There are many studies on evaluating rock brittleness by different methods，but few reports on the quantitative evaluation of rock brittleness by constructing brittleness index from the perspective of constitutive parameters. Therefore，in this paper，the feasibility of constructing a brittleness index   with the damage statistical constitutive parameters m and   is discussed through theoretical research，case analysis and comparative verification. The results show that the thermal-mechanical-damage constitutive model of rock based on the Weibull distribution function can well express various constitutive behaviors such as brittleness，plasticity，strain softening，which lays a theoretical foundation for evaluating rock brittleness from the constitutive perspective. The brittleness index   is established by parameters m and  ( )，where for parameter m，which can effectively reflect the overall shape characteristics of the stress-strain curve of the rock，and for parameter  ，which reflects the post-peak characteristics. Under different confining pressures，loading methods and lithology conditions，the calculated results of brittleness index   are consistent with the experimental results，which effectively verifies the rationality of brittleness index  . Compared with the existing representative brittleness indexes，it is found that the brittleness index   can effectively reflect the characteristics of rock brittleness decreasing with increasing confining pressure and rock brittleness enhancement under unloading stress path，and it also has good applicability to evaluate the brittleness of different hard rocks. The method in this paper provides a new idea for evaluating rock brittleness from the perspective of constitutive parameters，and is helpful to enrich the analysis and evaluation of rock brittleness.

In order to explore the evolution law of mechanical properties of limestone and progressive damage characteristics of internal meso structure under the effects of freeze-thaw and uniaxial load，the uniaxial compression and acoustic emission monitoring tests of limestone with different freeze-thaw cycles were carried out. Firstly，the physical and mechanical parameters evolution law was obtained. Then，the correlation between acoustic emission signal and microcrack activity in freeze-thaw limestone was analyzed. Meanwhile，the damage cumulative change rule was quantitatively studied. Finally，the deterioration mechanism of freeze-thaw damage of limestone was clarified. The results show that the uniaxial compressive strength，elastic modulus and Poisson?s ratio of limestone decrease gradually with the increase of freeze-thaw cycles，and accompanied with the final discount rate to 61.43%，48.42% and 17.33%，respectively. However，there is gradual increase in the peak strain. The acoustic emission ringing count change process can be divided into calm stage，growth stage，and steep increase stage. As the increase of freeze-thaw cycles，the uniaxial acoustic emission signal of limestone is active with the local high-density released characteristics，and the acoustic emission b value presents a “V”-shaped change. Meanwhile，the local sudden drop of the dynamic b value indicates that large-scale cracks have been generated. Moreover，the type of internal microcracks in the limestone converts from shear to tensile，which contributes to the failure change from shear to split. The damage variables of limestone，which presents with a trend of abrupt or gradual mutation after 0，10 and 20 freeze-thaw cycles，increase gradually after 40，60 and 80 freeze-thaw cycles. In addition，the main causes of freeze-thaw damage of limestone are the volume expansion of pore and fissure structures caused by water-ice transformation，meso hydraulic fracture caused by local ice accretion and the free hydrodynamic response.

To explore the strain rate effect of high-temperature granite after cooling with water，MXQ1700 box-type atmosphere furnace was used to prepare granite samples with five different temperatures at 200 ℃，400 ℃，600 ℃，800 ℃，and 1 000 ℃，and then the high-temperature granite was cooled with water. The Separate Hopkinson Pressure Bar(SHPB) system was used to conduct dynamic compression tests with four different impact rates(pressures were 0.30，0.40，0.50 and 0.60 MPa). In addition，the pore size distribution，porosity，and micromorphology of the granite samples were obtained through mercury intrusion and scanning electron microscope(SEM) tests. The microscopic results show that the damage degree of the granite sample can be divided into two stages at 400 ℃. Before 400 ℃，the porosity was less than 2.20%，and the damage degree was lower. When the temperature exceeds 400℃，the proportion of medium-sized pores increases，and the degree of damage increases with the temperature rise. In terms of strain rate effect：under the same temperature conditions，the dynamic elastic modulus，dynamic peak stress，and peak strain of specimens all increased with the increase of the strain rate，but the strain rate effect of elastic modulus is not apparent. The rock fragmentation increased with the rising of damage，and the fractal dimension of the granite sample increases with the increase of the strain rate. The effect of internal damage and impact velocity on the strain rate were analyzed，and the fitting formula of strain rate with internal damage and impact velocity was obtained.

As the main channel for fluid seepage，the fractured pores of coal have an important influence on the permeability of coal by their expansion and connection evolution laws under the action of mining stress. In order to reveal the evolution of the micro-fracture and pore structure characteristics of coal with axial stress loading，this paper used the NanoVoxel-3502E X-ray three-dimensional scanning imaging system and Deben loading test device to carry out the micro-CT in-situ scanning test of coal uniaxial compression. Acquired real-time CT data under different strain conditions，extracted the pores and fractures distribution characteristics of the coal sample，defined voidage and evenly divided the coal sample into three parts(region 1，region 2 and region 3) according to the stage of the voidage curve，and analyzed the evolution law of coal sample meso-structure with axial stress. The results show that that the uniaxial compression of coal presents staged fractures，and the distribution characteristics of fractures and pores are closely related to the internal stress propagation. The force transmission under uniaxial compression caused region 2 to appear first as the fractures and pores developed，followed by the development of region 1 near the axial stress loading indenter and connected fractures and pores，and at the peak (1 100 N) and after failure(492 N)，the maximum volume and surface area of pores and fractures and voidage were the largest，while the fractures and pores of region 3 near the axial fixed indenter developed relatively slowly，and the maximum volume and surface area of pores and fractures and voidage were the smallest. The coal matrix(skeleton) was locally deformed and destroyed under the action of axial force，formed isolated pores and tiny fractures，and the spatial distribution of the fractures and pores aggravated the unevenness of the internal stress distribution in the coal，which in turn had a great impact on the process of coal rock fracture.

Clayey rock，in virtue of its properties，such as low permeability，self-sealing，etc.，is considered to be a candidate host rock for underground nuclear disposal facilities. In this paper，we take clayey rock as our study object. Analyzing the related test results of clayey rock both in the laboratory and in situ，a nonlinear creep damage constitutive model has been established. This theoretical model considers the creep damage，as well as the creep strain accumulation effect. Based on the software of ABAQUS，we compiled the corresponding UMAT subroutine. A numerical analysis of the long-term hydro-mechanical response of a nuclear underground repository was carried out. The results indicate that：(1) the nonlinear creep constitutive model can well show the creep process of clayey rock. (2) The deformation，as well as the pore water pressure of the host rock，are significant influenced by the tunnel excavation. The pore water pressure soon decreases with deformation increases after the tunnel excavation. (3) The deformation of the host rock is closely related to the excavation distance and creep time. The excavation effect for both deformation and pore water pressure weakens as the increase of distance between the target location and the excavated tunnel. Meanwhile，due to the time effect，the pore water pressure，as well as the deformation around the tunnel gradually increased. (4) The consensus of the numerical simulation and in situ measurements on the deformation of tunnel linings show the reliability of the theoretical model. The research here will give significant guidance on the long-term stability analysis of the clayey rock nuclear waste repository in China.

The size effect on the probability distribution of rock compressive strength is very important for accurately determining its probability distribution and obtaining the correct reliability calculation results. The compressive tests were carried out on cubic red sandstone blocks with side lengths of 80，120 and 160 mm respectively. The results show that：(1) the probability distribution type of peak compressive strength of rock obeys normal distribution and will not change with the change of size. (2) The point estimation，interval estimation and disproof method show that both average and standard deviation of rock compressive strength have obvious size effect. When the size increases from 80 to 160 mm，the average decreases from 56.64 to 51.14 MPa，the standard deviation decreases from 14.33 to 3.96 MPa. The standard deviation decreases more significantly. (3) Taking the plinth with size of 160 mm as an example，the reliability calculation results are greatly influenced by the probability distribution under different sizes. The failure probability obtained by the probability distribution under 80 mm size is 11.00%，and the failure probability obtained by the probability distribution under 160 mm size is 0.12%. The former result is 91.67 times that of the latter. In addition，if only the size effect of the average value is considered and the size effect of the standard deviation is ignored，it will be realistic Therefore，The size effect of probability distribution of rock compressive strength in practical engineering should be paid attention.

In order to study the influence of the meso-heterogeneity on the initiation，propagation and coalescence of microcracks in Beishan granite，three types of meso-structure characterization models(cluster uniform model，Voronoi grain model and cluster tessellation model) were constructed based the combined finite-discrete element method(FDEM)、digital image processing(DIP) and grain based model(GBM). Based on the three types of models，uniaxial compression tests were carried out to study the influence of heterogeneity on the mechanics characteristics，acoustic emission(AE) characteristics and grain-scale crack propagation law. The research results show that three types of models can capture the evolution from micro-damage to macro-failure，intergranular tension cracks were firstly generated，then intragranular cracks were generated，and the cracks were mainly tensile cracks. The meso-structure characterizations effect significantly crack initiation stress(CI) and crack damage stress (CD). The AE characteristics of Voronoi grain model and cluster tessellation model were more consistent with the laboratory tests than cluster uniform model. With the increase of hard mineral content，the uniaxial compressive strength(UCS)，elastic modulus，CI and CD increased，but Poisson?s ratio decreased. With the increase of grain size，UCS and CD decreased. With the increase of stiffness heterogeneity factor，UCS，CI and CD decreased.

The urban shallowly buried box tunnels are usually built using jacking method. The face stability of the boring machine is a major problem in box tunnelling due to the fluctuation of the support pressure in the working chamber. The excessive support pressure triggers the passive failure of working face. In this paper，the problem is analysed by constructing planar admissible kinematically mechanism. In the proposed mechanism，the rotational and translational mechanism are included and the compatibility is satisfied by constructing an interface between the mechanisms using spatial discretization technique. The upper bound analysis is applied for stability analysis and the upper bound solution of support pressure is derived as a result. By optimizing the mechanism，the critical support pressure as well as associate failure mode is obtained. The parametric analysis indicates that the frictional angle significantly impacts the failure. The increase of frictional angle contributes to the failure zone expanding and the proportion of the block involved in translational movement to that in the rotational movement is reduced. As a result，the higher support pressure is required. With the increase of cohesion，the support pressure increases correspondingly but the failure mode is virtually unchanged. With the surcharge increasing，the failure region is enlarged and the support pressure is increased. The numerical simulation is employed for verification. The comparison indicates that the current solution matches to the numerical solution better than the previous solutions especially at high L/D. The applicability of the current solution in practice is discussed.

The traditional surrounding rock classification method is mainly based on the drill and blast method，which is not suitable for evaluating and predicting the TBM tunneling performance. Therefore，the purpose of this paper is to establish a rock mass boreability classification method suitable for TBM construction and achieve accurate perception and recognition of the boreability grade. For this purpose，based on the rock mass and TBM machine data sourced from four tunnels，the correlations between the rock mass parameters and TBM boreability evaluation indices were analyzed，and the multi-criteria decision-making(MCDM) method TOPSIS was used to establish the rock boreability classification system for TBM tunnel. Since it was difficult to obtain the parameters of rock mass in actual TBM excavation directly，the proposed rock mass boreability grade was recognized by using TBM driving parameters，including the thrust force per cutter(Th)，penetration rate(PR)，cutterhead revolutions per minutes(RPM) and penetration rate per revolution(PRev). In the perception and recognition，Bayesian optimization was used to determine the optimal combination of input parameters of various machine learning classification algorithms to achieve the maximum accuracy of grade recognition. The applicability of different optimized algorithms for the boreability grade recognition was compared，and then the optimal perception and recognition method was determined. Finally，based on the real-time TBM data of Jilin Yinsong tunnel project，the accuracy and effectiveness of the proposed classification，as well as the perception and recognition method of rock mass boreability were verified. The research results can be used to predict the boreability of rock mass in the stable operation sections of TBM projects，and provide reference for the optimization of TBM driving parameters and the establishment of intelligent control system.

Water absorption is one of the most important cause of deep soft rock deformation and failure. In this paper，X-ray diffraction，scanning electron microscopy and test of water-absorbing without pressure were carried out on deep soft rock samples in the Yixin coal mine. Water absorption characteristics and its influencing factors of rock samples in the study area are comprehensively analyzed. Results have shown that the patterns of the water absorption trends of the three types of rocks were all characterized by a similar kind of dynamic process with decreasing water absorption rate which could be well fitted by a negative exponential function. However，significant differences were found in water absorption ability among the three types of rock samples which could be sorted in an order of mudstone(0.74%)＞siltstone(0.56%)＞medium sandstone(0.27%). The content of clay minerals and fractal dimension have a significant effect on water absorption capacity，which all show a good positive correlation. There are obvious differences in the effect of different mineral compositions on the water absorption capacity. Illite can obviously promote the water absorption capacity，while quartz and kaolinite can inhibit the water absorption capacity.

Tunnels crossing active faults have appeared in water conservancy，highway，railway and subway projects in recent years. However，the failure mechanism of tunnels under fault dislocation and prevention method don?t keep up with the development of engineering construction. Therefore，a comprehensive collection of domestic and foreign cases of tunnels that have been built across active faults and cases of earthquake fault damage were analyzed，and the failure process and characteristics of the tunnel under creeping and stick-slip dislocations were analyzed. A review of the engineering geological model，mechanical model，and dislocation displacement model adopted by the dislocation failure mechanism research of tunnels，as well as the analytical methods，numerical simulations and physical simulation methods used. The existing prevention and control measures and their principles are summarized. On the basis of the above research，the engineering problems，failure mechanism and prevention measures were discussed and prospected. It is proposed that railway tunnels across active faults should not only consider the safety of the tunnel structure，but also the safety of surrounding rocks，rails，lines and trains influenced by creeping，stick-slip，and ground motions of active faults. It is believed that the study of dynamic evolution of properties of fault surrounding rock，the dynamic simulation of stick-slip fault and the influence of the radius of the fault curve of the line should be strengthened. Some design principles of railway tunnels during design and operation phase are respectively given. The results have certain reference value for the design，construction，operation and maintenance of the tunnel project across active faults.

In-situ stress is very important for tunnel design and construction. In order to solve the problem that the boundary stress of each sub-region can not be continuously transferred and the whole stress field of the project can not be analyzed in the whole line in the subregional inversion of large-scale stress field，the following researches are performed. Firstly，combined with the engineering experience of inversion of the in-situ stress field，the application of boundary loads when simulating the tectonic stress field in the axial direction of the tunnel is studied based on the principle of multiple linear regression. Secondly，a“overlapping partition- combination”inversion method of in-situ stress field is proposed. Finally，the method was applied to the Shengli extra-long road tunnel in Tianshan，Xinjiang，and verifying the validity of the method. the stress field distribution characteristics of the entire engineering area was analyzed. The results show that when simulating the tectonic stress field，the boundary load should be applied from the side with the higher terrain in the calculation domain. The“overlapping partition-combination”inversion method of in-situ stress field effectively avoids abnormal stress field results caused by improper boundary loads in the direction of the tunnel axis during the overall regression inversion of the engineering area，and ensures the continuous transition of the sub-regional boundary stress. The distribution law of the overall in-situ stress field obtained by the partition combination is reasonable. The inversion method can provide reference for inversion of in-situ stress field of extra-long tunnel engineering.

Xiong?an New Area is rich in geothermal resources. In the future，large-scale exploitation and utilization of the deep geothermal storage will have important strategic significance for making green Xiong?an，reducing the environmental pollution in Capital Circle region and improve the energy structure of Beijing，Tianjin，and Hebei. There are several buried Quaternary faults in the Xiong?an New Area，such as Rongcheng，Niudong and Xushui—Dacheng faults. For the upcoming exploitation of the Jixian-Changcheng geothermal system，whether water injection will affect the stability of the main buried faults near the reservoir is of critical importance. At present，there has been no report correlations. In this paper，taking the Rongcheng geothermal reservoir as a case study，we firstly investigated the regional in-situ stress field，and then evaluated the initial stability of the main buried faults using the Mohr-Coulomb fracture criteria. Secondly，we calculated the excess pore pressure caused by water rejection at representative deep geothermal wellbores by the Hsieh and Bredehoeft hydrological model. Subsequently，with the modeled excess pore pressure，we estimated the fault slip potential on the main buried faults associated with water injection，based on a probabilistic approach. Finally，we discussed the influence of injection rate，injection time and fluid withdrawal on the mapped fault slip potential. Our results suggest that：(1) The excess pore pressure generated by water injection at the rate of 171 m3/h for 40 years is less than 12 MPa at representative deep geothermal wellbores，within a distance of 3 km near the Rongcheng boundary faults. The excess pore pressure decay shows power law-like behavior to distances of less than 9 km. (2) The stability of the middle segment of the western Rongcheng boundary fault (namely F9–2) is significantly affected by water injection，while this influence on other buried faults is not significant. (3) The fault slip potential on the mapped fault F9–2 increases exponentially with water injection rate as well as with injection time. Projected in 2070，the fault slip potential on the mapped fault F9–2 is 33.16%，indicating a high risk. (4) The continuous pumping in deep geothermal wellbores near the boundary fault of Rongcheng uplift can lead to the decrease of the minimum principal stress and the increase of the maximum differential stress；as a result，the stress accumulation on the mapped fault F9–2 gradually approaches the critical state. Therefore，the effect of upcoming geothermal extraction on the stability of faults nearby warrants great attention.

Carrying out the evaluation of the boreability of tunnel boring machines(TBMs) has an important guiding significance for scientific planning of key construction factors. Based on 219 sets of data of Karaj water conveyance tunnel，Zagros water conveyance tunnel，and West-Qinling railway tunnel，this paper proposed a deep belief network model for the TBM boreability evaluation，improved by Bayes optimization algorithm and early-stopping strategy. This model used rock uniaxial compressive strength(UCS)，rock quality designation(RQD)，the angle between the dominant structural plane and the tunnel axis( )，and rock mass cuttability index(RMCI) as input variables，and used field penetration index(FPI) as the output variable. In the data preprocessing stage，Kriging interpolation and improved CRITIC algorithm were separately used to complement the missing values in the database and implement data weighting. Taking Yinsong water conveyance tunnel and LXB water conveyance tunnel as examples to test the practicality of the model：For 37 sets of test data of Yinsong water conveyance tunnel，the root mean square error (RMSE)，mean absolute percentage error(MAPE)，and coefficient of determination( ) were 2.18，8.25%，and 0.926 2，respectively；for the 49 sets of test data of LXB water conveyance tunnel，the RMSE，MAPE，and   were 2.83，8.14%，and 0.981 7，respectively. Furthermore，by quantitatively comparing the RMSE，MAPE and   of the model before and after data weighting，data weighting was an effective way to improve the prediction performance of the model. Finally，a comparative analysis of the model in this paper with BP neural network，support vector regression，K-nearest neighbor and random forest was conducted in terms of prediction accuracy and running speed，verifying the superiority of the model in this paper.

The aim of this paper is to study the asymmetric collapse mechanism of the overlying rock mass of shallow tunnel under seismic load. Based on the Hoek-Brown failure criterion，combined with the upper bound theorem of the limit analysis and the variational method of the variable endpoints problem，the analytical solution of the asymmetric collapse curve is proposed. The effects of seismic body wave and rock parameters on the asymmetric collapse curve are analyzed. It can be found that，without considering the damage of rock mass，the S-wave will tilt the geometric shape of collapsing block，which is beneficial to restrain the collapse. And the restrain effect is more significant in the rock mass with better mechanical properties. The backward propagation of P-wave will promote the collapse of the rock block and make the geometrical area of the collapsing block smaller. This implies that under seismic load，a smaller range of allowable velocity field can lead to the ultimate failure state of rock mass. The conclusion of this paper will help to provide a safer design and construction guidance for engineering practice.

The red mudstone with the characteristics of low strength, easy weathering, and easy expansion of water absorption，which is widely distributed in the Sichuan Basin and resulting a high threat to the high-speed rail operations due to strict control of rail deformation in Sichuan. In reference to the continuous implementation project of subgrade deformation for certain transportation line，the swelling characteristics of the weakly weathered mudstone were correlated with the influence of the mudstone by the pressure head，which were described in this essay. It has been found that the swelling characteristic of weakly weathered mudstone shows a significant time lapse and there is no observation of converging about one month for in-situ swelling test. However，there is still a great potential of swelling which indicates a long and slow process of the swelling of weakly weathered mudstone in the natural environment. The indoor test infers that the absorption capacity of the red mudstone increases with a growing water pressure and the time lapse curve shows three stages: rapid growth，decelerating growth, and gradual stabilization. The specimen reaches 90% of its capacity within 14.4 to 49.1 hours and achieves 98% in 145.2 hours. Based on the research results, the differences in the expansion deformation characteristics of indoor and outdoor mudstones and their internal mechanisms were compared and analyzed，which inspires the exploration of continuous deformation of mudstone subgrade.

Prestressed anchor cable is the most common support method in slope reinforcement，and revealing its prestressing time-history change law is of great significance to the long-term stability study of reinforced slopes. Firstly，based on the shear lag model，the force characteristics and load transfer mechanism of the prestressed anchor cable anchoring structure's three elements(anchor bars，mortar and surrounding rock) are studied ，and the calculation formulas for the shear stress at the interface of the anchoring structure and the axial stress distribution of the anchor bars are obtained. Furthermore，according to whether the shear stress at the anchor bar-mortar interface exceeds the bond strength of the two contact interfaces，a prestress coupling model and a prestress decoupling model are proposed；among them，the prestress decoupling model is established based on the shear lag model，which can calculate the final stable value of the anchor cable prestress. Finally，taking Maya High Slope Anchorage Project of the Shuibuya Hydropower as an example，the time-history variation law of the prestress is explored and the prediction formula is put forward. The research results show that the shear stress distribution and the axial stress distribution are characterized by negative exponential attenuation；The fitting result of actual monitoring data proves the rationality of the prediction formula；The maximum error between the calculated stationary value and the actual pre-stress stationary value is only 2.17%，which verifies the accuracy of the prestress decoupling model.

The clay gravel layer of the Anqing Formation is widely distributed in the upper reaches along the Yangtze River in Anhui Province. It is a special engineering geological body composed of gravel and clay. After the excavation of the project，it is very easy to cause slope collapse. In this paper，the clay gravel bed of the Anqing Formation was studied. Firstly，the type and composition of cement in the clay gravel bed were analyzed by X-ray diffraction method，and the cementation characteristics of the clay gravel layer were clarified，including matrix cementation and soil-stone interface cementation. Then，the experiment of artificial rain erosion model in the slope of clay gravel bed was carried out，and the variation law of slope destruction instability pattern of clay gravel layer with different gravel content was studied. The results show that when the gravel content is small (30%)，the slope failure is mainly dominated by gully，and its failure form is mainly controlled by the clay layer. When the gravel content increases to 50%，the gravel skeleton effect is gradually reflected，and the slope failure form is greatly different from the soil slope. The slope failure is manifested as the evolution of erosion hole - erosion pit - shallow layer collapse. When the gravel content is large (70%)，the gravel skeleton structure is relatively stable. Even if the clay particles are washed away in large quantities，the overall stability of the slope is better. Finally，the variation of substrate moisture content and wet peak migration depth of slopes with different gravel content were analyzed，and the internal mechanism of slope failure of clay gravel layer was further explained.

The tensile failure and instability induced by medium and low strain rate disturbance load is one of the main causes of dynamic disasters in deep coal resources and residual coal resources mining. And it is of great significance to analyze the tensile mechanical response characteristics of coal and rock under dynamic disturbance for disaster prediction and prevention. In this regard，Brazilian splitting test of sandstone under medium low strain rate disturbance load was carried out .To analyze the influence of different pre-static load levels and disturbance load amplitude on dynamic tensile mechanical response characteristics of sandstone. The results show that：The degree of damage due to disturbed loading was positively correlated with the pre-static load level, and the AE energy level increased by about 11 times when the pre-static load increased from 7 kN to 20 kN. Macroscopically, this is manifested as a reduction in tensile strength. The tensile strength of disturbed sandstone with pre-static load of 11 kN and 16 kN is 12.4 MPa and 11.5 MPa，respectively，which is 5.6% and 12.7% lower than that of static tensile strength. The existence of stress threshold value of disturbance load is verified. The sandstone in high stress state will not be destroyed in the process of small amplitude dynamic disturbance load，and the tensile strain increases logarithmically. The sandstones in high stress state are destroyed during the application of disturbing loads greater than the stress threshold. And the number of cycles of perturbation loading before damage decreases exponentially with increasing amplitude. The increase of pre-static load level and dynamic load amplitude both promotes the expansion of secondary tension cracks in the vicinity of the central tensile crack in the rock sample. The precursor of strain surge before dynamic tensile failure of rock specimen can provide reference for dynamic disaster warning.

In order to explore the evolution law of two-dimensional fractal parameters of limestone thermal shock fracture，the high-resolution X-ray three-dimensional CT detection system was used to observe the standard limestone specimens treated by thermal shock tests at different temperatures，and the two-dimensional fractal parameters of limestone were counted and calculated with image processing software. To study the evolution law of two-dimensional fractal parameters after thermal shock fracture of limestone. The results show that when the high-temperature limestone is cooled by thermal shock in 20 ℃ water，as the temperature increases from 200 ℃ to 600 ℃，the two-dimensional fractal dimension D increases monotonically，and the increase range of D is less than 400 ℃–600 ℃ at 200 ℃–400 ℃ stage. For limestone with strong random distribution of fissures，the fissure development index K is introduced，and the fitting relationship between K and limestone temperature T under thermal shock in 20 ℃ water is obtained. The thermal shock fracture of limestone at 400 ℃ in water at different temperatures，as the water temperature increases from 20 ℃ to 100 ℃，the two-dimensional fractal dimension in any direction gradually decreases，and there are directional differences. The thermal shock fracture of 400 ℃ limestone in water at different temperatures shows that the three-dimensional fractal volume dimension of the fracture surface is consistent with the two-dimensional fractal dimension，and they have a highly fitted linear relationship. The study results can enrich and develop the theory of high temperature rock mechanics.

Characterization of the spatial variability of rock mass parameters differs significantly from that of soil parameters due to complex structures inherently existing in the rock masses. At present，the spatial variability modeling of the mechanical parameters of highly fractured/weathered rock masses and those dominated by a single structural plane and associated slope reliability analysis have been extensively studied. However，few attempts have been made to depict the spatial variability of the mechanical parameters of wedge and anti-dip rock masses and conduct associated slope reliability analysis. In this paper，a method for spatial variability modeling of the mechanical parameters of structural planes and slope reliability analysis is proposed for the wedge and anti-dip rock masses. The varying ranges of the autocorrelation distances and scales of fluctuation of the mechanical parameters for different types of rock masses are systematically summarized. An interface program between the spatial variability modeling of the mechanical parameters and probabilistic slope stability analysis with FLAC3D software is developed. An explicit function relationship between the factor of safety of the slope and the random field vectors of structural plane parameters is constructed using the Back-Propagation(BP) neural network. Then the non-intrusive stochastic finite difference method is adopted to explore the influence of the spatial variability of the mechanical parameters of structural planes on the probability of slope failure. The results indicate that the autocorrelation distances and scales of fluctuation of the mechanical parameters are different among different types of rock masses. The single exponential autocorrelation function is the most widely used theoretical autocorrelation function. The direction simulation and direct Monte-Carlo simulation methods are adopted to validate the effectiveness of the proposed method. The proposed method can provide an effective means for the spatial variability characterization of the mechanical parameters of structural plans for the wedge and anti-dip rock masses in the slope reliability analysis. In addition，the probability of slope failure will be overestimated if the spatial variability of mechanical parameters is ignored，which can further result in conservative slope reinforcement design schemes.

In order to investigate the properties and influencing factors of unsaturated flow through a single rock fracture under different confining stress，unsaturated flow experiments are performed on four single limestone fractures under various saturations and confining pressures by use of the self-developed rock fracture unsaturated flow test system. The experimental results show that：(1) With the increase of confining pressure，the air entry pressure increases，and the decline slope and amplitude of unsaturated permeability coefficient and capillary pressure relationship curve decrease. Under the same capillary pressure, the fracture unsaturated permeability coefficient corresponding to high confining pressure is higher than that under low confining pressure. (2) By comparing the effects of capillary pressure and confining pressure on unsaturated permeability coefficient，it is found that the effect of confining pressure on unsaturated permeability coefficient decreases with the increase of capillary pressure. Similarly，with the increase of confining pressure，the effect of capillary pressure on unsaturated permeability decreases. (3) With the increase of roughness，the fracture saturated permeability coefficient decreases and the residual saturation increases slightly. (4) Based on the fracture unsaturated flow test data under different confining pressures and capillary pressures，the relationship between fracture unsaturated permeability coefficient and capillary pressure and confining pressure is fitted. From the comparison between the fitting surface and the test data, the fitting relationship can accurately describe the change of fracture unsaturated permeability coefficient with capillary pressure and confining pressure. The research results on the process and law of unsaturated flow in fractures under variable stress can provide a basis for the stability analysis of engineering rock mass under the condition of rainfall infiltration.

Alpine and high altitude areas are affected by freeze-thaw cycles throughout the year，and frost heaves in fractured rock masses are widespread. In order to explore the evolution law of frost heaving force in fracture frost heaving deformation and reveal the frost heave damage mechanism of fractured rock，using the test system of distributed pressure film sensor，temperature sensor and strain gauge to fully monitor the freeze-thaw process of saturated cracked red sandstone samples with different crack widths in real time，then the curves of crack frost heave force，temperature and crack strain with time were obtained. The experimental results show that during the freezing and thawing process，a constant temperature platform appears at 20 mm in the water saturated red sandstone fracture， indicating that the fracture water is freezing，and the larger the fracture width is，the earlier the constant temperature platform appears；The strain of rock surface is synchronous with the inflection point of temperature，the inflection point of fracture strain lags behind the inflection point of temperature，but the change of fracture strain is synchronous with the change of maximum frost heaving force；When the constant temperature platform appears，the strain of rock surface rises rapidly，mainly caused by the joint action of the shrinkage of the rock skeleton and the frost heaving of the rock pore water；The change of temperature on the rock surface is faster than that in the fracture，which leads to the lag of frost heaving of fracture water；During the thawing process，the frost heaving force and strain both climb again，and appear the secondary peak；For cracks with a width of 2–4 mm，the maximum frost heaving force increases with the increase of crack width；In the whole process of freezing and thawing，the distribution of frost heaving force is irregular，the frost heaving force increases from the four sides to the inside，and the frost heaving force around basically remains unchanged；The analysis of freeze-thaw mechanism shows that：a frozen layer is first formed around the rock mass，and then the fracture water freezes and gradually forms an ice plug. When the ice plug no longer slips and reaches a completely closed state，the fracture strain rises rapidly，the frost heaving force rises to the maximum，and the stress concentration at the fracture tip leads to microcracks. During melting，the surface of the rock melts and secondary frost heaving occurs in the fracture. and secondary frost heaving occurs in the fracture. The test preliminarily analyzes the freezing law of pore water and crack water in the fractured rock mass，which provides the basis for reasonably analyzing the frost heaving damage mechanism of fractured rock.

To improve the rock crushing efficiency and reducing energy consumption of ore，the dynamic mechanical properties and breakage behaviour of magnetite under one-dimensional coupled dynamic and static loads were studied using the improved Hopkinson pressure bar(SHPB) apparatus. The effects of axial compression ratio and impact velocity on the crushing effect and energy utilization rate of magnetite ore were analysed. Results show that impact velocity mainly affects the impact resistance and crushing degree of magnetite ore. The smaller the sample fragmentation the more significant the influence of axial compression and the larger the cumulative strain，when the axial compression ratio is constant，with impact velocity increasing. The larger the axial compression ratio，the larger the cumulative strain increases. The axial compression ratio mainly affects the damage degree of the inner structure before the dynamic impact and the energy utilization rate of impact load of the magnetite ore. With axial compression ratio increasing，the dynamic compressive strength of magnetite samples has a maximum value. Under one-dimensional combined dynamic and static loading，the crushing effect and energy utilization rate of magnetite ore are controlled by axial compression ratio and impact velocity. The impact resistance of magnetite ore is enhanced by“low axial compression + high velocity”. The compression and shear effect of“high axial compression + low velocity”is not significant and the crushing effect is poor. “High axial compression + high velocity”makes the sample broken into a large number of debris powder，which has good crushing effect with low energy utilization rate. When the“extrusion + impact”rock breaking method is used to improve the crushing effect and energy utilization rate of magnetite ore，the optimal combination of axial compression ratio and impact velocity should be put forward on the premise of clear crushing fragmentation requirements and considering the combined effect of axial compression ratio and impact velocity comprehensively.

In order to study the damage evolution law of red sandstone under different cyclic loading and unloading paths，the MTS815 rock mechanics test system was used to perform constant lower limit cyclic loading and unloading test and variable lower limit cyclic loading and unloading test on red sandstone. Meanwhile，acoustic emission(AE) and digital image correlation(DIC) monitoring technique were carried out to capture microfracture and strain evolution information during test. The results showed that compared with the uniaxial compression test，the average compressive strength of the rock under constant lower limit cyclic loading and unloading increased by 6.5%，while the increase under variable lower limit cyclic loading and unloading was not significant. The average ratio of Felicity decreased continuously with the increase of stress under constant lower limit cyclic loading and unloading，while the ratio of Felicity basically remained unchanged in the later period of variable lower limit cyclic loading and unloading. The evolution trend of the apparent strain field was consistent with the variation characteristics of acoustic emission. Under constant lower limit cyclic loading and unloading，the damage deformation of rock increased gradually before failure，showing the characteristics of continuous increase. However，under cyclic loading and unloading of the variable lower limit，the damage strain was relatively dispersed in the early stage of loading and unloading. It was not until the stress exceeds 66% of the peak strength that the large apparent strain points of the rock were dispersed and rapidly concentrated，and the local regional deformation increased sharply，showing the characteristics of abrupt change，and the samples showed severe brittle failure，which should be paid attention to in practical engineering.

Two adjacent tunnels which connected by a transverse passage are widely used in urban subways. The dynamic response of the special structure section is more prominent under the train vibration load. Two parallel cross shield tunnel models with and without transverse passages were carried out in this paper. The dynamic response characteristics of parallel cross shield tunnel under train load were studied. The influence mechanism of cross passage on the vibration response of two main tunnels was explored. Results show that the existence of the cross passage will reduce the dynamic response of the tunnel where the train load is located，and its influence range is about twice the width of the cross passage. While，For another adjacent main tunnel，when the center frequency is lower than 125 Hz，the one-third octave frequency of the lining structure increases due to the existence of the cross passage，and the lower the frequency is，the greater the impact is. The influence mechanism of the cross passage on the vibration response of the two main tunnels is as follows：Since the integrity and compactness of the cross passage is much higher than that of the soil，the vibration wave generated by the train load in the cross passage has a lower loss than the soil. Moreover，the lower the frequency of the vibration wave is，the faster the loss during the wave propagation process is.

Controlling by rock mass structure，the stability of rock mass in meizoseismal areas is particularly prominent. In this paper，the dangerous rock mass along the Ludian National Highway in Yunnan was taken as a typical example. A method to visually identify structural planes and quickly extract characteristic information by using point cloud directly was proposed. Firstly，based on the point cloud，the random Hough algorithm and the Fuzzy C-means algorithm were used for clustering analysis. Then the Fast-Marching algorithm was applied to reconstruct the dominant structural planes in three dimensions，to realize the characteristic information extraction of geometric information and spatial distribution state of structural planes. Meanwhile，based on the extracted characteristic information of structural planes，combined with HSV technology，the in-situ dangerous rock mass was identified and counted. The Monte Carlo simulation was also used to compute the in-situ rock volume distribution. Finally，the kinematic analysis of the dangerous rock area was carried out to judge the instability failure mode of the dangerous rock mass. The research results have a certain reference value for the investigation and evaluation of road dangerous rock mass in meizoseismal areas and the energy level design of protective structures.

Rockburst is a huge disaster faced by underground engineering，and rockburst prediction can reduce the harm caused by rockburst. Machine learning is the research hotspot and development direction of rockburst prediction methods，However，at present，each machine learning algorithm performs differently，and works independently of each other. Without fusion，they cannot complement each other?s advantages，resulting in low accuracy，generalization and stability of all machine learning algorithms. In this paper，eight machine learning algorithms(four ensemble algorithms and four basic algorithms) which are widely used at present are integrated by the Stacking integration algorithm，so as to give full play to the advantages of each algorithm and realize the complementary advantages. In order to ensure the diversity of new feature information，combined with the principles of various machine learning algorithms and the characteristics of rockburst sample library，three sets of Stacking integration algorithms with multiple rockburst prediction indices were proposed，each of which had different base models and multiple meta-models. The difficulty of accepting feature information and selecting meta-models in traditional Stacking integration algorithm is solved. The accuracy，accuracy，recall，and F1 values of each Stacking algorithm and independent algorithm were compared and analyzed，the results show that the Stacking algorithm can effectively integrate all the machine learning algorithms，and the prediction performance is significantly improved. Among the three sets of Stacking algorithms，the base-model of Stacking algorithm 2 is composed of Random Forest Classifier，Extra Trees Classifier，Gradient Boosting Classifier，and LGBMClassifier integration algorithms with high prediction accuracy，and the precision of new feature information provided by the base-model is the highest，and the prediction improvement of each meta-model is the most significant. Among the three sets of Stacking algorithms，the accuracy of the SVC algorithm and the LGBMClassifier algorithm in the meta-model of Stacking algorithm 2 is the highest in the three sets of meta-models. The Stacking ensemble algorithm can learn more feature information by using different groups of meta-models with good performance to vote for rockburst prediction. The accuracy of the test set is finally stable at 94.12 %，which is better than the prediction performance of independent machine learning and common theoretical criteria(the highest accuracy is 91.18 % and 53.9 %，respectively). Finally，the rock burst prediction of Zhongnanshan tunnel shaft is carried out by the established Stacking integration algorithm，and the prediction results are consistent with the actual situation.

The three different damage parameters for fiber concrete are determined for theoretical researches based on the fiber concrete tests with the digital image correlation(DIC) technology，and the mechanics model for surrounding rock-tunnel structure is also obtained. On the basis of the finite element displacement method expressed by stress intensity and stability factor with analytical solution，the influences of different crack location，depth，width，angle，surrounding rock grades on the structural stability of lining cracks are analyzed under the cases of conventional concrete and different types of fiber concrete for secondary lining. The results show that lining cracks are subjected the compression and shear stress under the train excitation loads. Compared with the conventional concrete，the fiber concrete can effectively enhance the structural stability of the cracked lining，and reduce the stress concentration at crack tip during the dynamic loads of train. Among them，the glass fiber concrete has the strongest reduction effect，the mixed fiber is the second，and the PVA fiber is the weakest. Furthermore，the influence of cracks from arch foot to arch roof on the stability of lining structure tends to climb status，and compared with the crack width，the crack depth and inclination angle are more obvious for the effect of stability of lining structure under train loads. For the case of the inclination angle  = 45°for crack in arch roof of lining tunnel，the stability of lining structure is weakest. Taking  = 45°as the dividing line，the area of   is more dangerous than the area of  ，and the danger increases with the increase of train speed. Moreover，when the rock around the tunnel is grade V with low strength，the reinforced effect of fiber concrete on the structural stability is greatly reduced. At this moment，even if the depth of crack in arch roof of tunnel lining is very small，the tunnel lining structure also will enter a dangerous status，so special attention should be paid to the repair of such cracks in the engineering practice.

In order to explore the microscopic damage and deterioration mechanism of dolomite under static loading，combining the NMR relaxation mechanism and   spectrum relaxation signal characteristics，a multi-peak normal mathematical model is proposed，at the same time，the NMR before and after the triaxial compression test Resonance detection signals verify the model. Combining the characteristics of NMR relaxation signals before and after the test and NMR imaging technology，the damage and deterioration mechanism of the microscopic pore structure in the rock is analyzed from the NMR   relaxation spectrum signal and the pore size distribution characteristics. The results show that：(1) The proposed mathematical model can more completely characterize the NMR relaxation signal information，and the inversion fit is = 0.99. Combined with the NMR relaxation information during rock macroscopic failure，the proposed mathematical model can be used to Quantitatively evaluate the damage state of the rock. (2) Through the comparative analysis of the NMR test results before and after the rock compression failure test，the NMR test of the rock sample showed three typical peaks before the rock failure，and the small pores and mesopores have good connectivity，The distribution of large pores is less，after the rock is macroscopically destroyed，the connectivity of the medium and large pores is significantly enhanced，and the mesopores and large pores have typical“close-merge”characteristics. After the failure，the T2 spectrum peak area，and total spectrum The peak area ratio exceeds 90%，and the macroscopic damage value D＞0.9. (3) When the rock is damaged under different confining pressures，the nuclear magnetic relaxation signal characteristics of the internal pore structure are the same，and the distribution of large pores is dominant. The relationship between the difference in pore size distribution and the confining pressure is not significant. The research results will provide a reference for the scientific and quantitative evaluation of the damage state of rock mass engineering.

Exploring the longitudinal deformation mechanism of the existing shield due to construction of new tunnel above is significant to assess the adverse impacts of above-crossing tunnelling on the existing shield tunnel. Current analytical methods commonly simplify the existing shield tunnel as an equivalent continuous beam resting on a linear elastic foundation，of which will overlook the weakening of circumferential joint and the nonlinear deformation characteristics of the ground. Firstly，the joint discontinuous shield tunnel model and the nonlinear Pasternak foundation model are adopted to consider the weakening of circumferential joint and the tunnel-soil nonlinear interaction. The differential equation for shield tunnel longitudinal deformation owing to above-crossing tunnelling is derived. Secondly，the numerical solution for shield tunnel longitudinal deformation is solved using the finite difference method and the Newton?s iteration method. Finally，the predictions from the proposed method are compared with two field cases and the results predicted by the current equivalent continuous beam model. It shows that the longitudinal deformation and the bending moment based on the equivalent continuous beam model express continuous characteristics. While the longitudinal deformation and the bending moment predicted by the proposed method can reflect the discontinuous characteristics，because the weakening of circumferential joint is considered by the proposed method. The results obtained by the proposed method is much closer to the real shield tunnel deformation characteristics. Through the parametric analyses，it is found that increasing the clearance distance between the new and old tunnels will effectively reduce the tunnel heave，bending moment and joint opening. By increasing the length of segment ring，the tunnel heave will decrease slightly，but the joint opening and bending moment will increase. To increasing the stiffness of the circumferential joint will effectively lead to the decrease of tunnel vertical displacement and opening of joint，but it will increase the bending moment of tunnel.

To improve the performance of particle swarm optimization(PSO) in model parameter inversion，an improved PARTICLE swarm optimization(FPSO) algorithm is proposed. By using linear piecewise function and exponential function，the learning factors   and   are adjusted by asynchronous strategy to realize dynamic adjustment of inertia weight and   and   with the number of iterations. The particle diversity is guaranteed by the immune mutation operator and convergence rate is guaranteed by the formation search，which is applied to the parameter inversion analysis of the incremental rheological model of the Arts face dam. The results show that：(1) the FPSO algorithm can effectively improve the accuracy and efficiency of the inversion analysis because it better fits the motion characteristics of particle from the early high-speed global search to the late local fine search，and the global search ability is greatly improved. (2) By using the measured settlement of the dam during the downtime，the rheological parameters of the rockfill are inverted based on FPSO，and the calculated results are in good agreement with the monitored values.

In this paper，the horizontal vibration of piles in unsaturated foundation considering the vertical load is studied. On the basis of the Timoshenko beam model and three dimensional poroelastic unsaturated continuum medium theory，the soil-pile interaction model considering the vertical load at the pile top is established. Through the direct solving method，variable separation method and the boundary conditions of the soil-pile system，the analytical solutions of horizontal displacement，rotation angle，bending moment，shear force and dynamic complex impedance of pile head are obtained. The solution in this paper is verified through the comparisons against existing theoretical solutions. Further，the differences between the Euler and Timoshenko Beam theories are illustrated under the condition of vertical load and pile cap mass. The major findings can be summarized as：(1) The vertical load only has influences on the stiffness factor of the horizontal dynamic impedance at the pile top，and will not cause any change in the resonance frequency except for the reduction of the minimum value. (2) The complex impedance calculated by the Euler beam model is always larger，but the Timoshenko beam model should be preferred for safety reasons. (3) With the increase of vertical load，the horizontal displacement at the pile top，the rotation angle at the middle and upper part of the pile，the bending moment and shear force at the middle and lower part of the pile all increase slightly，which may threaten the structural safety. (4) When the pile cap mass is considered，the dynamic response of the pile decreases to a certain extent.

The coarse-grained soils used in the subgrade filling of Beijing—Xiongan intercity railway was selected to carry out surface vibration compaction test，and the dry density threshold of the coarse-grained soils under different vibration frequency and excitation force combination was studied. Then，the nonlinear dynamic response between vibration equipment and filling material is studied by using VMD Hilbert transform. The energy transformation of compaction process is analyzed based on the law of energy conservation. And the compaction degree of the filling material is evaluated by the Hilbert marginal spectral energy of acceleration signal. The results show that：the dry density of the filling material in the surface vibration compaction test presents a trend of “rapid growth，slow growth，keep stable”，and the dry density threshold of filling material is different under different compaction parameters. By using the VMD to decompose the acceleration signal，the fundamental and harmonic components can be accurately identified and separated，and there is no mode aliasing. By analyzing the Hilbert spectrum and Hilbert marginal spectrum of acceleration signal，the propagation law of acceleration signal energy in time-frequency domain during compaction process is obtained，which shows the transfer from low-frequency fundamental component to high-frequency harmonic component. In addition，there is a logarithmic relationship between the Hilbert marginal spectral energy of the exciter acceleration signal and the dry density of the filling material， so it is feasible to use the Hilbert marginal spectral energy of the vibration signal decomposed by VMD to evaluate the compaction degree of the filling material.

In order to study the static liquefaction characteristics of fine-grained tailings，the inner mixing method was used to prepare tailings with different fine-grained contents. The static liquefaction characteristics of tailings under different confining pressures and fine contents were studied by undrained consolidated triaxial tests. The test results show that the lower the effective consolidation confining pressure is，the more obvious the static liquefaction of saturated tailings will be. Meanwhile，three liquefaction parameters including undrained brittleness index，liquefaction potential index and instability potential index can be used to reflect the static liquefaction characteristics of tailings. With the increase of fine content，the three liquefaction parameters of tailings show a trend of first increasing and then decreasing，and reach the maximum value when the fine particle content is 30%，while tailings are most prone to static liquefaction at this time. Finally，the static liquefaction characteristics of tailings are classified by using modified state parameters. The rationality of classifying static liquefaction characteristics of tailings by modified state parameters is demonstrated by the test results of other tailings and correlation test analysis.

Fiber-Reinforced Backfill(FRB) has obvious advantages in filling the of goafs. To investigate the microscopic damage mechanism of FRB，a series of tests were conducted on glass fiber reinforced ultra-fine tailings backfill with different cement tailings ratios by using compression machine，scanning electron microscope，nuclear magnetic resonance instrument and acoustic emission monitor to systematically analyze the fracture expansion and damage evolution process of FRB from microscopic level. Studies have shown that：the incorporation of glass fiber can improve the initial structure of the backfill material and reduce the initial defects，among which the improvement effect is most obvious for the specimen with 1∶12 cement tailings ratio. There are obvious differences between the damage patterns of CPB and FRB. The damage of CPB is dominated by primary fissures through the specimen，with spalling at local locations，while the damage of FRB is dominated by more small-scale secondary fissures，which is caused by the crack-resisting effect of glass fibers on the backfill material. Due to this reason，the acoustic emission(AE) characteristics of FRB and CPB also show a significant difference：the CPB specimen shows an“AE drop zone”in the ringing count after the yield point(about 85% of the peak stress)，while the FRB specimen shows an“AE rise zone”in the ringing count after the yield point. The results of the study can not only provide theoretical guidance for promoting the application of glass fiber reinforced backfills，but also provide acoustic emission criteria for predicting the peak strength of FRB materials.

The dismantling of the inner support in deep foundation pit caused the deformation of the retaining structures，which has a significant impact on the surrounding environment. The space curve differential equations for retaining structures within each independent coor-dinnates system were proposed by applying the Winkler elastic foundation beam theory. And the non-limit equilibrium state of soil pressure of retaining structures was considered，and the reduction coefficient μ was uesd to quantitatively describe load reduction in horizontal direction during the transformation of basement structural plate and inner support. Then the deformation of retaining structures by this numerical method were solved. And taking the deep excavation project of Xiaguangli building of Chaoyang District and Shenzhen Pingan building as a background，the horizontal deformation of the retaining structures in different stages of the dismantling inner support was calculated，and the influence of the process of the dismantling inner support on the horizontal displacement of the retaining structures was analyzed. The results show that the calculated values of this method are close to the site monitoring date. During the process of dismantling inner support，the horizontal displacement of the retaining structures is approximately a parabola like curve with the maximum point moving upward. The increment of horizontal displacement is mainly concentrated above 1/2H depth of retaining structures. The value range of reduction coefficient from 0.1 to 0.3 is reasonable. The influence of value on the deformation of retaining structures decreases with the increase of depth. The horizontal displacement of retaining structures below excavation depth is not affected by value.

The restoration technology of earthen sites is a hot issue in the protection engineering. Based on laboratory test，model test and analytic hierarchy process(AHP) theory，the engineering characteristics of traditional modified soil(MS) and its suitability for earthen sites restoration were studied. The results showed that：(1) the engineering properties of the MS were improved by adding a certain proportion of sticky rice，cementitious material or fiber material to the plain soil(PS). The shear strength，impermeability and durability of PS are increased by 1.3‐3.5 times，3‐1 650 times and 1.3‐12.5 times，respectively. (2) Six indexes for comprehensive suitability evaluation of various kinds of MSs during the restoration of earthen sites were put forward，which are the safety，durability，anti-seepage，hygroscopic and desiccant characteristics，economy，construction difficulty. It indicates that the suitability of MS with adding sticky rice and mortar is the better than that of the lime，plant fiber materials. The cement-modified soil should not be used in earthen sites restoration.

In order to prevent the degradation of permafrost under subgrade in permafrost area，and test the refrigeration effect of the developed solar driven compression refrigeration apparatus，the field test was carried out in Qinghai Permafrost Research and Observation Base of the Ministry of Transportation. The results show that the apparatus can continuously cool the soil in cold season and warm season. The refrigeration mode can affect the distribution of refrigeration temperature along the pipe wall，and the bottom-up refrigeration mode is more effective to cool the permafrost subgrade. The average refrigeration temperature of the apparatus is positively correlated with the ambient temperature. The average refrigeration temperature in cold season is lower，the lowest is －7.97 ℃，and the highest is －3.98 ℃ in warm season. When the average refrigeration temperature is lower than －6.5 ℃，the refrigeration capacity of the apparatus is greater than that of the soil，the soil cold storage capacity increases and the ground temperature decreases，on the contrary，the ground temperature increases. The refrigeration radius of the apparatus is affected by the ambient temperature，and the refrigeration radius is 1.95‐2.61 m. The apparatus can be arranged on both sides according to the distance of 4‐5 m in subgrade engineering. To sum up，the solar driven compression refrigeration apparatus can be effectively applied to the temperature control of permafrost subgrade and can guarantee the reliability and safety of permafrost subgrade in a long term.

The drilling with pre-stressed concrete pile cased(DPC) pile is a new type of pre-drilled large-diameter PHC pipe pile，and its pile toe can be embedded in the moderately weathered rock. In order to reveal the influence of pile toe type on the bearing capacity of rock-socketed DPC piles，an ABAQUS finite element numerical analysis model was established by combining concrete damage plasticity model，interface bonding model and Mohr Coulomb model of equivalent Hoek Brown criterion. The influence factors such as hole bottom shape，pile shoe size and plugging concrete at the pile toe were emphatically analyzed. Numerical results show that，under the condition of no plugging concrete at the pile toe，the toe resistance ratio can reach more than 50% for the cylindrical cone groove at the bottom of the pile hole，and the influence of groove depth(15–30 cm) is limited. However，for the trapezoidal cone groove at the bottom，the end bearing capacity is obviously weakened，which is only 61% of that of cylindrical cone groove. The open-ended steel pile shoe can lead to the progressive destruction of the bedrock; the toe resistance ratio increases from 54% to 60%–76% when the thickness of pile shoe increases from 24 mm to 40–130 mm. Considering that increasing the thickness of pile shoe will affect soil discharge，the reasonable thickness of pile shoe should be determined by field tests. For the rock-socketed DPC pile with pile shoe，plugging concrete at the pile toe is an effective measure to improve the bearing capacity. However，the end bearing capacity will be weaken when there is sediment in the lower part of the plugging concrete. Finally，the improvement measures for the construction technology of DPC pile were put forward.

Steel pipe-frozen soil synergistic structure is a new freezing model，controlling the development of the frozen wall boundary during the freezing process is the key to restraining the impact of frost heave and thaw settlement on the environment. Taking the freezing reinforcement project of Jiangpu Road Station of Shanghai Metro Line 18 as the background，model tests were designed and conducted based on similar theoretical，which on the development process of frozen wall boundary of steel pipe-frozen soil synergistic structure，analysing the influence law of steel pipe and circulating water on the boundary development，and conclusions were drawn as followed：The development of frozen soil can be inhibited by steel pipe set at the boundary of the frozen wall，which can also expand the temperature gradient at the boundary of the frozen wall，making the frozen wall more uniform. When freezing 32 days，the temperature difference between the inner and outer sides of single-row and double-row steel pipes can reach 11.2 ℃ and 7.6 ℃，while the temperature of the frozen soil outside the steel pipe is 6.7 ℃ and 10.7 ℃ higher than the corresponding position at the bottom of the freezing pipes. The outward expansion of frozen soil boundary can be effectively controlled by 4 ℃ of the circulating water，further improving the uniformity of the frozen wall. The temperature difference between the inner and outer sides of the frozen soil boundary steel pipes reaches 18.3 ℃ when freezing 32 days，and the temperature of the frozen soil outside the steel pipes is 16.9 ℃ higher than the corresponding position at the lower part of the freezing pipes. The research results show that the expansion of the frozen soil can be effectively controlled by the steel pipes or 4 ℃ circulating water placed at boundary of the frozen wall，improving the uniformity of the frozen wall，and significantly weakening the influence of frost heave on the surrounding environment during the freezing process，while the effect of 4 ℃ circulating water at the boundary position is better.

In order to study the wetting-induced deformation characteristics of unsaturated sandy loess under complex hydro-mechanical paths，the single-line triaxial wetting tests of the two complex hydro-mechanical paths of consolidation-loading-wetting-loading and consolidation-unloading-wetting-unloading under deviator stress are carried out by GDS unsaturated triaxial apparatus for unsaturated sandy loess. The effects of hydro-mechanical path and stress ratio on the wetting-induced deformation characteristics of unsaturated sandy loess are comprehensively analyzed. The research results indicate that：(1) Both the axial strain and the wetting shear strain decrease with the increase of stress ratio. (2) When the stress ratio is 0.25，the specimen fails to reach saturation，and then the dilatancy failure occurs. The wetting shear strain and void ratio of the specimens that experienced the loading path develops faster than that experienced the unloading path，and are the first to reach failure. (3) When the stress ratio is 0.5 and 0.75，the wetting deformation characteristics are basically the same. The wetting volumetric strain and moistening shear strain both increase first and then become stable with the increase of the wetting parameters. The void ratio first decreases and then becomes stable with the increase of the wetting parameters. The maximum wetting volumetric strain，wetting shear strain and the initial void ratio before the moistening of the samples that experienced the loading path are all greater than that experienced the unloading path.

To study the effect of compaction degree on the pore structure of unsaturated remolded weak expansive soil，and to investigate the influence of the change of pore structure of expansive soil on the soil-water characteristic curve(SWCC). Taking the remolded expansive soil in Hami of Xinjiang as the research object，the mercury intrusion test was used to determine the impact of compaction on the microscopic pore structure of the soil sample，and use the thermodynamic relationship model to study the fractal dimension of expansive soil. The pressure plate instrument and filter paper method were used to test the soil of the remodeled soil sample under different dry densities-water characteristic curve. The expansive soil SWCC two-parameter fitting model that can consider the effects of compaction is proposed. Finally，based on the pore characteristics of the mercury intrusion test，the soil-water characteristic curve is corrected with the help of the capillary principle to calculate and combine the test data of the filter paper method. The results show that as the degree of compaction increases，the pore distribution curve of expansive soil presents an initial Trimodal distribution and gradually tends to be bimodal. The compaction effect is manifested in the microscopically as the conversion of large pores to small pores caused by the compression of large pores. The fractal characteristics of pores show that compaction has increased the roughness of the inner wall of the pore and the complexity of the pore structure. The suction of the soil matrix decreases with the increase of water content. Under the same volumetric water content，the greater the initial dry density of the soil，the greater the suction of the matrix. The two-parameter model can better reflect the influence of the initial dry density of the soil on the development law of the soil-water characteristic curve of the expansive soil. The calculated value of the soil-water characteristic curve based on the capillary principle has the same function model development law as the actual measured curve of the filter paper method，and the two tend to gradually“far away”as the initial dry density of the sample increases. The modified soil-water characteristic curve model can uniformly describe the development law of the matrix suction of the unsaturated weakly expansive soil with the pore structure，water content and density state.

The particle morphology of graded gravels of high-speed railway subgrade has a significant impact on the macro-mechanical properties. Reasonably quantifying the three-dimensional morphological characteristics of the particles is the basic work of the research on the micro-mechanism of fillers. Based on the laser scanner，the point cloud data of 700 gravel particles in 7 particle groups in the range of 2.36–31.5 mm are acquired. And these data are reconstructed to form three-dimensional grid models. By introducing the root mean square curvature (RMS) which describes the surface unevenness of the particles and considering the particle mesh model smoothing process and the RMS curvature distribution characteristics，the angularity parameters are defined by the proportion of the point cloud quantity in the edge area，and the texture parameters are defined by the difference of the model before and after smoothing. Combined with the shape parameters described by the relationship between the long，middle and short axes，the indexes system to describe the multi-scale morphological characteristics of gravel particles is formed. By selecting the particles with typical morphological characteristics，it is verified that the established index system has a good morphology discrimination ability. According to the statistical analysis of the scanned gravel particles，the shape，angularity and texture of the gravel particles in different groups basically present the characteristics of normal distribution. As the particle size increases，the proportion of block particles increases，and the angularity and the texture characteristics are more prominent. The research results can provide a reference for the quantitative evaluation of particle morphological characteristics from macro，medium and micro scales.