In order to study the acoustic emission characteristics during the stick-slip process of fault，friction-slip experiments were carried out on the prefabricated coarse-grained syenite fault model with different dip angles，and the evolution process of the acoustic emission characteristics of fault stick-slip instability under different loading rates and lateral pressures was analyzed. The results show that：(1) In a stick-slip process，the number of acoustic emission events in the stress accumulation stage increases steadily，and the acoustic emission energy levels are relatively small；when the fault is unstable，the acoustic emission events increase suddenly，and accompanied by high-level acoustic emission event，the energy is much greater than other events，indicating that more and more intense acoustic emission activities are produced when the fault is unstable，and the sudden increase in acoustic emission energy occurs in the sub-instability stage of the fault，and the occurrence of high-energy acoustic emission events can be used to the fault stick-slip predicted. (2) The stick-slip characteristics of faults are different under different loading conditions. When the lateral pressure and dip angle change，the stress state of the fault plane affects the contact state of the concave and convex body，and the macroscopic expression is the change of friction strength of the fault plane. With the increase of lateral pressure，the shear stress of slip-induced fault increases and the stress drop increases. When the inclination angle changes from 34°to 45°，sticky-slip is less likely to occur，and the strain energy released during instability is greater. The peak stress of sticky-slip event increases continuously at the inclination angle of 45°，while the shear stress of fault plane increases continuously at the inclination angle of 56°，but sticky-slip does not occur. When the loading rate decreases，the stick-slip period becomes longer，and there is a relatively long time for adjustment and physical healing between the interrupted planes during the loading process. The stress state of the fault plane changes，and the stress drop of fault instability increases. (3) When loading conditions change，acoustic emission characteristics change. When the lateral pressure increases，the growth rate of AE count increases，and the AE activity becomes more frequent. At the same time，the amplitude of high frequency increases，and the generation and expansion of fault micro-rupture become more complicated. When the dip angle changes from 34° to 45°，the amplitude of low-frequency signal increases significantly，which indicates that the change of dip angle makes the fault appear larger micro-rupture.

In order to study the dynamic propagation law of cracks with different lengths in sandstone materials under blasting loads，the crack initiation law and crack propagation speed of sandstone are studied with the help of crack propagation gauge. The dynamic parameters of sandstone and crack propagation time measured in the test are imported into ABAQUS code，a finite element software，to establish a numerical calculation model. The crack propagation speed and dynamic stress intensity factor of mode I cracks and fracture toughness of sandstone with different crack lengths are obtained by experiment-numerical method. The test results show that：(1) In the blasting tests，the initiation time and delayed fracture time of pre-cracks with different lengths are different. When the lengths of pre-cracks vary between 60 and 80 mm，the change of initiation time and delayed fracture time is not obvious，and it basically increases linearly when the lengths vary between 80 and 120 mm，and then the change tends to be flat；(2) The dynamic propagation speed of crack in sandstone is not a not a constant value and fluctuates during crack propagation，when the lengths of pre-cracks vary between 60 and 140 mm，the pre-crack length has little effect on the average propagation speed of dynamic crack，and the variation range is within 10%；(3) With the increase of pre-crack length，the initiation toughness of sandstone increases gradually，that is，the difficulty of sandstone initiation increases，and its propagation toughness is inversely proportional to the propagation speed，which is still valid when the crack length changes.

In the process of slope rock mass from stable state to unstable state，the dynamic index will change significantly. Therefore，the introduction of appropriate dynamic index is the key to realize the accurate identification of unstable rock mass. In this study，six dynamic indexes such as time domain，frequency domain and energy index are introduced，and the PSO-SVM algorithm is used to carry out the experimental study on the identification method of unstable rock mass. The results show that compared with the identification model of unstable rock mass with single dynamic index in time domain and double dynamic index in time domain and frequency domain，the prediction effect of the identification model of unstable rock mass based on multi-level dynamic index is the best，with MSE of 0.004 772 and of 0.984 865. Therefore，the accurate quantitative analysis of unstable rock mass can be realized by using the identification method of unstable rock mass with a variety of sensitive dynamic indicators. According to the statistical analysis，the importance order of identification sensitivity of six dynamic indicators is the mean square frequency＞the margin index＞the relative energy of the first frequency band＞the center frequency＞the impact energy＞the pulse index. The study provides a relatively rich and sensitive dynamic monitoring index for the identification of unstable rock mass in the engineering site，which is helpful to establish a set of unstable rock identification method based on big data analysis，so as to meet the practical needs of disaster prevention and mitigation of collapse disasters in alpine valley areas.

Based on the analysis of fracture mechanics，the fracturing collapse mechanism and stability analysis method of columnar dangerous rock mass(CDRM) in southwest mountainous area are proposed in this study. Combined with the development characteristics of columnar dangerous rock mass，we generalize the geomechanical model composed of the upper intact rock mass and the bottom multi-crack structure in reason. And we also introduce rock fracture mechanics to analyze the cracks initiation and propagation characteristics in multi-crack structure，and then establish an analysis method of base controlling the stability of CDRM. This study shows that：under the self weight load，the form of crack propagation in base is wing crack. Accordingly，we can divide the failure of rock bridge into tensile failure between parallel cracks and shear failure between collinear cracks. Through the analysis，the cracks initiation and propagation degree are the main basises for evaluating the stability of CDRM. Taking Jianchuandong and Zengziyan dangerous rock mass both in Chongqing as examples，this study reasonably evaluates the risk of such geological disasters.

To study the effect of seepage on shear mechanical properties of rock joints，direct shear tests on 10 irregular sandstone joints in non-seepage natural state and 25 irregular sandstone joints in seepage state were carried out using the rock joint shear-seepage test device which is modified by RDS–200 rock direct shear apparatus. The influences of seepage pressure，normal stress and joint roughness on the shear mechanical properties of rock joints are analyzed，and the empirical formula for the peak shear strength of irregular sandstone joints under seepage is proposed. The results show that compared with the natural state without seepage，the peak shear strength of rock joints decreases，the peak shear displacement increases，and the pre-peak shear stiffness decreases in seepage state. Based on the empirical formula of peak shear strength of rock joints proposed by Barton，considering the effective stress principle and the influence of seepage water pressure on the basic friction angle together，the empirical formula of peak shear strength of irregular sandstone joints under seepage condition is proposed. Only two test parameters which can be easily to obtain(rock joint wall strength JCS and seepage water pressure) are added in the formula，and it has a good fitting effect on the test data of rock joint shear strength under seepage condition，which is helpful to accurately estimate the peak shear strength of rock joint under seepage condition.

To investigate the dynamic mechanical properties and stress wave propagation law of coal rock mass under the action of in-situ stress and joints，the dynamic compression tests of coal rock under different static loads (0，2，4，6 and 8 MPa) and different joint coincidence coefficients(JMC：0.8，0.9，1.0 and no joint) were carried out on modified split Hopkinson pressure bar(SHPB) coupled static and dynamic loads by using artificial joint specimen. The results show that，with the decrease of joint match coefficients，the amplitude of the reflected wave increases，and the transmitted wave decreases. With the static load increased from 0 MPa to 8 MPa，the transmission coefficients of samples with different joint match coefficients increase，and the smaller the joint match coefficients，the greater the increase of transmission coefficient. The influence of static load on the transmission coefficient of samples with low joint match coefficients was more obvious. The peak stress increases with the increase of static load，while the peak strain decreases with the increasing static load. The specific stiffness and dynamic modulus of the joint specimen increase with the increasing static load. It shows that the existence of static load inhibits the deterioration of jointed coal and rock mass. Based on the stress-strain relationship and the standard linear solid model，the experimental results are verified. With the increase of static load and joint coincidence coefficients，the seismic wave quality factor increases，and the sample dissipated energy decreases. It is considered that more attention should be paid to the influence of joint and static load on the transmission characteristics of stress waves in underground engineering.

In order to consider the effects of nonuniform mesoscopic(local) stress field and strain field on the macroscopic mechanical response of porous rock under loading, a multi-scale elastoplastic constitutive model based on incremental variational theory was established. Firstly，the porous rock was assumed to be a composite material composed of a pressure-sensitive solid matrix and spherical pores. In this context，the incremental variational principle for porous rock was proposed. Secondly，the local mechanical behaviors of solid matrix were assumed to obey a threshold and isotropic hardening parabolic Mises-Schleicher plastic model, the local increment potential of solid matrix as well as the effective incremental potential of porous rock were deduced. The macroscopic stress-strain relationship of porous rock was obtained by optimizing the effective incremental potential and combining with the Mori-Tanaka homogenization method. Finally，a numerical algorithm for the multi-scale elastoplastic constitutive model was developed based on the traditional return mapping algorithm，and a UMAT subroutine was compiled to embed the multi-scale model into Abaqus software. The accuracy and effectiveness of the multi-scale model were verified by comparing the simulation results of the multi-scale model with the finite element unit cell model(reference solutions) as well as the test data of typical Vosges sandstone. The results show that the numerical solution of the multi-scale model based on the incremental variational theory was in a good agreement with the finite element reference solution，and can accurately reproduce the macroscopic mechanical behavior of the Vosges sandstone.

Rock avalanche dynamics is always a hot and challenging issue in the field of landslide. Focusing on this issue，the Luanshibao rock avalanche is chosen as an example with its whole propagation process being simulated by MatDEM，to further illustrate its fluidized propagation and frictional heating effect from a mesoscopic point of view. It is reached that：(1) The velocity of the sliding mass presents an obvious vertical zoning feature with the bottom forming an intensive shearing zone characterized by lower velocity and the upper part characterized by higher velocity. (2) The maximum velocity of the mass center is 33.3 m/s. Its horizontal runout and vertical drop are 1469.9 and 374.1 m，respectively，with the equivalent friction coefficient being 0.255. (3) Frictional energy occupies 44.3% of the total loss of the potential energy，which mainly distributes along the basal facies in the translation zone. Correspondingly，the entrainment effect in the translation zone is obviously higher than that in the accumulation zone. (4) The frictional energy generated in the propagation can cause the basal facies to rise by 64 ℃ with frictional heating effects generated. (5) In the translation zone and the rear part of the accumulation zone，the equivalent friction coefficient displays in low values，which is attributed to the coupling effect of momentum transfer，entrainment effect and frictional heating effect. The simulated results have a good consistency with the field investigated data，which is conducive to the revealing on its propagation mode and frictional heating effect. The simulated results can provide significant scientific value for rock avalanche kinematics and dynamics.

Given the lack of the understanding for the hydration leads to softening mechanism of deep shale and the corresponding constitutive model，based on the laboratory experiment and theoretical model research on the deep shale of longmaxi formation in Sichuan Basin，establishing the softening constitutive model of hydration. Firstly，carrying out the experiment of free immersion in clean water. Then CT scanning is performed on rock samples with different soaking time，and CT gray histogram is drawn to analyze the meso-damage mechanism of hydration，and carrying out the triaxial compression test on soaking samples. Finally，establishing the softening constitutive model of shale under high closure pressure by using continuous damage mechanics and Weibull statistical theory. It is found that the main cause of shale hydration damage is the initiation and expansion of internal microcracks，the mechanical parameters，cohesion，peak strength and elastic modulus，decrease with the increase of hydration time，Poisson?s ratio increase with the increase of hydration time. In the stress-strain curve，the compaction stage and yield failure stage are obviously prolonged，and the model parameters m and decrease with the increase of hydration time，which macroscopically explains the softening of shale mechanical properties caused by hydration.

The surrounding rock of salt caverns，being ideal locations for the storage of energy and disposal of nuclear waste，presents with significant creep deformation over a long period of time due to being continually affected by a combination of stress and temperature. To determine the effects of temperature and stress on the creep of salt rock，triaxial creep tests at varying conditions were conducted，the mechanisms of the effects of temperature and deviatoric stress at steady-state creep rate were determined，and fitting was conducted to obtain the parameters of steady state creep rate constitutive equation. The linear dashpot of the Maxwell model was replaced by a fractional Abel dashpot to establish a fractional Maxwell model that can be used to consider the effects of temperature. A new fractional viscoelastic-plastic creep damage model considering temperature was achieved by connecting the viscoplastic damage model to the fractional Maxwell model. The parameters of the creep damage model were extracted from the results of the experimental triaxial creep test using 1stOpt software. The theoretical model results and tests data were compared to determine the validity of the creep damage model. The new model can well simulate the creep deformation in whole evolution.

The existing rock damage calculation methods have the problem of inconsistent energy calculation results for the uniaxial loading and cyclic loading. To solve this problem，both the existing calculation methods and the energy conversion mechanism are analyzed，a rock damage calculation method based on damage release energy is further proposed，and the indoor tests were conducted to verified its reliability. The research results indicate that the dissipated energy and the elastic energy released when rock sample fails are related to its damage state. These two kinds of energies can be defined as damage release energy. It is the absorbed energy from the loading to the failure in the case of uniaxial loading. For the case of cyclic loading，it is the sum of that of the initial cycle，the intermediate ones，and the final loading stage. The increment of damage release energy in the initial cycle can be obtained by integrating the loading stage and unloading stage，that in the intermediate cycle is the difference between the dissipated energy of the current cycle and the damping energy of the previous one，and that in the final loading stage is difference between the input energy of the cycle and the damping energy of the previous one. The indoor tests show that the total damage release energies for the same type of rock samples are about the same under uniaxial loading and cyclic loading. The damage evolution of rock sample during cyclic loading can be analyzed on the premise that only the uniaxial loading results are obtained. Besides，this method can better reflect the damage evolution of rock sample with the loading process.

The rapid declination of the internal pressure in salt rock storage caverns leads to the dilatation and damage of the surround rocks. The effects of the initial confining pressure and unloading rate on the dilatation damage characteristics of salt rocks are studied in the present work，through the triaxial test with unloading confining pressure at constant axial stresses. The experimental results show that: at the same unloading rate，the higher the initial confining pressure，the larger the dilatation in the unloading stage. The mean volume strain at the confining stresses of 12 and 18 MPa increases by 7.1% and 21.4%，respectively，compared with that at the initial confining pressure of 6 MPa. Additionally，slow unloading rate of confining pressure is conductive to dilatation. As the unloading rate decreases from 0.2 MPa/s to 0.02，0.002 and 0.000 2 MPa/s，the mean volume strain increases by 60.2%，103.5% and 282.1%，respectively. In the unloading process，the larger the dilatation，the more serious the internal damage. The deformation modulus in the unloading stage decreases with the unloading of confining pressure. The larger the initial confining pressure and the unloading rate，the larger the deformation modulus reduces. The damage variable and the energy absorbed by salt rock in the unloading stage increase linearly with the increase of initial confining pressure，and on the other hand，decrease exponentially with the increase of unloading rate. The more energy absorbed，the more serious the internal damage.

Through the MTS815 rock mechanics test system，conventional triaxial loading test and unloading tests with different initial confining pressures and different unloading rates were carried out on the deep-buried marble in Jinping. Based on the energy evolution characteristics of the marble deformation and failure process under different stress paths，the progressive failure stages are divided. The energy change law corresponding to each characteristic stress in the process of rock progressive failure is further explored. It can provide reference for underground geotechnical engineering design and dynamic disaster prevention. The research shows that energy evolution curve of deep-buried marble under different stress paths have obvious stages. It can be divided into four progressive failure stages：elasticity，plasticity，yield expansion and post-peak. During the progressive failure process of marble，the characteristic stress and characteristic energy have obvious confining pressure effects. With the increase of confining pressure，the characteristic stress increases linearly. At the same time，the total characteristic energy，elastic strain energy and dissipation energy are increased. Compared with 25 and 50 MPa，the peak elastic energy at 80 MPa is the largest. The energy released is more at failure. It is manifested as a tendency to explode in the middle rock. Under unloading conditions，the peak stress is smaller. The marble's carrying capacity is reduced. The peak total energy，elastic strain energy，and dissipation energy also decrease. Under the same initial confining pressure，the crack initiation stress，peak stress and unloading rate of deep-buried marble show nonlinear relationship. The damage stress increases linearly with the increase of unloading rate. Contrary to loading，peak total energy，dissipated energy decreases with increasing unloading rate. Marble exhibit no rock explosion tendency at different unloading rates. When the unloading rate is 1 MPa/s，the peak dissipated energy is the smallest. In contrast，the peak elastic energy is the largest. The degree of rock damage is small，and more energy is released during failure.

In order to determine the ultimate mining height for ensuring the stability of coal wall，the characteristics of coal rib spalling under different coal seam conditions were studied. Based on the phenomenological method，the paper established the three-dimensional model and the equilibrium differential equation of the coal rib spalling，and obtained the section trajectory of the failure surface for coal rib spalling. On the basis of this，the stress equation of coal rib spalling was established by using three-dimensional simplified wedge model. By using unified formula to characterize two different rib spalling of soft coal shear failure and hard coal tensile failure，the paper analyzed the coal rib spalling mechanism of shear failure and tensile failure and put forward the formula of the ultimate mining height for ensuring the stability of coal wall. According to actual mining parameters of coal seam condition，the influence on coal wall stability limit height of coal seam parameters which include Poisson?s ratio，cohesion and internal friction，joint fissure，cohesion and internal friction of the coal and rock interface was mainly studied. The results showed that the coal wall stability limit height is increased by the increasing of the Poisson's ratio，cohesion and internal friction of coal rock interface cohesion and coal seam. Furthermore，on account of the research on the coal wall stability control effect of face guard supporting intensity，the paper put forward the estimate formula of the coal wall ultimate stability height considering the face guard supporting intensity and the improvement scheme of hydraulic powered support face guard device by analyzing the structural characteristics of different face guard devices. Finally taking the soft coal seam of Yuzhou mine in Henan province and the hard coal seam of Yulin mine in Shanxi province as examples，the validity of the research results was verified. The theoretical research results were basically consistent with the production practice data，which better explained the coal wall stability characteristics of coal seam under different conditions. The research could provide theoretical support for the coal wall stability control and the design optimization of hydraulic powered support in the super-large mining height mining face.

In order to accurately predict rockburst disasters，the four integrated learning of Boosting，Bagging，Stacking，and Voting are applied to rockburst disaster prediction. The prediction performance of common machine learning algorithms，Boosting and Bagging is compared. The base model selection method of Stacking and Voting is proposed. First，275 groups of rockburst cases at home and abroad were collected to construct the original rockburst data set，statistical parameters of different rockburst grades in the original rockburst dataset were analyzed. TSNE(t-distributed Stochastic Neighbor Embedding) algorithm is used to visualization analysis the original rockburst data. The analysis shows that there are a large number of outliers in the original rockburst data set and the data is unbalanced. Secondly，Yeo-Johnson transformation and K-means SMOTE oversampling are used successively to normalize and balance the data to reduce the effects of outliers and data imbalance，respectively. The normalized and balanced rockburst data has enhanced separability. Then，Training prediction is made for 15 machine learning algorithms including general machine learning，Boosting and Bagging. The accuracy of macro average and Friedman statistical hypothesis test were used to compare the prediction performance of various models. Finally，a prediction method of rockburst intensity grading based on Stacking and Voting is proposed. The models with low accuracy or high similarity of prediction results were eliminated by using the exhaustive method. General machine learning，Boosting and Bagging algorithms retain the models with high precision and as different as possible as the base models for Stacking or Voting. The results show that the prediction performance of Boosting，Bagging，Stacking and Voting is mostly better than that of ordinary machine learning. The Stacking and Voting rockburst intensity classification prediction methods combining diversity and accuracy weights can effectively improve the rockburst prediction performance.

In order to explore the change law and meso-mechanism of the mechanical properties of concrete in chemical environments such as saline soil. Limestone，which is commonly used as a coarse aggregate for concrete in Northwestern China，is used as the research object. The chemical corrosion tests of limestone immersed in CaCl2 solution and Na2SO4 solution with pH = 2，5，7，9，12 were carried out. Nanoindentation mechanics test and scanning electron microscope observation technology are used to carry out the micromechanical test research of limestone under chemical solution erosion. The results of the study indicate：After chemical corrosion，the elastic modulus and hardness of the limestone samples are reduced to varying degrees. The limestone sample shows a certain timeliness. The corrosiveness of limestone samples is CaCl2 solution＞Na2SO4 solution. At the same time，the degree of influence on the limestone sample is acidic environment＞alkaline environment. The changes of pH and ion concentration in the solution are closely related to the damage of the sample. It shows that the higher the pH and the higher the ion concentration，the lower the mechanical parameters of the sample. Limestone is most sensitive to the pH of the solution. The main effects on the chemical corrosion of limestone are H+ and OH－ in the solution. Finally，a new damage variable defined by elastic modulus is introduced to describe the degradation degree of the mechanical properties of limestone after chemical damage At the same time，the relationship between damage variables of limestone and other mechanical parameters is analyzed. The study found that the damage variable is closely related to the hardness. The smaller the hardness，the greater the decrease in mechanical parameters.

As we all known，how to accurately locate the source position is of great significance in acoustic emission testing. The traditional AE localization method has some problems，such as wave velocity model and inaccurate on arrival-time picking，which brings a low localization accuracy. Therefore，a new localization method that combining spectrum analysis and convolutional neural network was developed，which don?t need the wave velocity model and the arrival-time picking for acoustic emission localization of rock. To be more specific，based on the acoustic emission full waveform signals collected by using active source acoustic emission test，we got spectral distribution of acoustic emission signals on each surface of cuboid granite sample using wavelet analysis method firstly. Then surface source positioning of acoustic emission signals was carried out，according to the spectral difference of acoustic emission signals，which cleverly transformed and the complex three-dimensional spatial positioning into two-dimensional plane positioning. At the same time，a convolutional neural network designed to transform the acoustic emission signals of the broken lead points on each surface using the short time Fourier transform. Finally，we taken the contains both time domain and frequency domain information of the signal spectrum characteristics as the convolution of the neural network input，and the two-dimensional space coordinates of the break point as the output features to train，test and verified the two-dimensional localization results of acoustic emission sources of lead points on each surface progressively. The results show that the average location errors of surface XOZ－(which attaches four transducers)，YOZ+，YOZ－，XOZ+ are 0.5 cm，0.7 cm，0.7 and 1.0 cm，respectively. To sum up，this study significantly improves the AE localization accuracy and avoids the shortcomings of traditional localization methods，which provides a new idea for rock AE localization.

The layered structure is a common type of rock mass structure in nature，which is widely distributed in China. However，under the action of tectonic stress in the west，it is easy to form a special rock mass combination structure，namely，a layered slab-rent structure. In this paper，the mechanical properties of slate with layered slab-rent structures in the western region are studied. The results show that the internal structure of slate rock has a great influence on the failure mode and mechanical properties of rock，and the anisotropy is obvious. In order to eliminate the interference of the rock structural plane，the mechanical properties of the layered slab-rent structure are studied by using the prefabricated interlayer thin interlayer of granite with high integrity and good homogeneity. The research results show that the peak strength and elastic modulus of rock mass decrease linearly with the increase of interlayer plate-cracking structural plane density. The larger the interlayer plate-cracking structural plane density is，the smaller the strain energy absorbed by the rock mass in the compression deformation process is，and the degree of damage is also weakened. When the failure occurs，cracks first appear at the plate-cracking structural plane，and then the cracks expand and penetrate into the failure. On this basis，the density parameter of the structural plane is introduced to optimize the deformation constitutive equations of the existing elastic homogeneous discontinuous structure and fractured structure rock mass. The optimized constitutive equation can better describe the constitutive relationship of the layered slab-rent composite structure rock mass.

The permeability of coal seam in most mining areas in China is low. Fracturing is currently one of the effective methods for enhancing the permeability of coal seams，which can strengthen the effect of coal seam gas desorption and seepage. In order to further study coal seam gas adsorption，desorption and seepage characteristics before and after fracturing，a true triaxial test system for the whole process of gas adsorption-desorption，fracturing，and seepage in coal was independently developed. Gas adsorption，desorption，seepage，and hydraulic fracturing test were conducted under true axial stress. The test system was mainly composed of high-pressure chamber，stress loading-control system，gas injection-adsorption-desorption system，fracturing system，test system，and data acquisition-control system，et al. The sample size was 300 mm×300 mm×300 mm，the maximum loading stress was 30 MPa，the maximum gas injection pressure was 20 MPa，and the pressure endurance of the hydraulic fracturing system was lower than 80 MPa. The main material of the sealed high-pressure chamber was 42CrMo forged steel，which has high safety and sealing performance. The results showed that the maximum adsorption amount of gas in triaxially stressed coal( = 10 MPa， = 8 MPa and = 6 MPa) was approximately 6.50 cm3/g，the maximum strain of coal in the maximum principal stress( ) direction was 4.48×10－3，that in the intermediate stress( ) direction was 4.90×10－3，the adsorption and deformation characteristics both were in accordance with the Langmuir-type change law. The cumulative desorption amount of gas in coal increased exponentially with the time during the desorption process. The desorption velocity of gas in coal showed an exponential decay trend with time，and the residual deformation in the maximum principal stress direction of coal was smaller than that in the intermediate stress direction. Coal was fractured along the maximum principal stress direction under hydraulic fracturing，and the maximum initiation pressure was 15.31 MPa. The permeability of coal before fracturing was from 0.005×10－15 to 0.023×10－15 m2，and it was from 0.028×10－15 to 0.163×10－15 m2 after fracturing，an increase of 4.34–6.06 times. The permeability of coal had a negative exponential relationship with effective stress. This system can provide a certain experimental basis for coal seam gas drainage，fracturing and permeability enhancement technology.

When the thickness of the coal seam is unstable and the strength of the roof is much lower than the coal seam，The surrounding rock of the roadway generally has problems such as large deformation of surrounding rock，large roof rupture depth，and obvious regional differences in roof stability. For such problems，taking the extremely soft roof of the mining roadway in 16–3 coal seam of Mindong No.1 Coal Mine as the engineering background，the comprehensive research methods such as theoretical analysis，field measurement and numerical simulation are adopted to analyze the characteristics of mine pressure behavior of the mining roadway under different top coal thicknesses，and reveal the failure law of the extremely soft roof of the mining roadway under different top coal thicknesses in the process of mining influence. The results show that the plastic failure of roof will pass through the medium hard coal seam to form the interlayer expansion failure zone under the mining influence. The larger the thickness of top coal is，the smaller the interlayer failure range will be. When the thickness of top coal exceeds 5.0 m，the interlayer failure phenomenon tends to disappear. The large-scale interlayer failure zone generates severe deformation pressure，which has a huge extrusion effect on the lower hard coal body，leading to the overall large deformation and even complete instability of the roof. It is the internal reason for the severe deformation and failure of the roadway roof in the thin top coal area. Accordingly，taking the area of the hidden interlayer failure as the benchmark and the thickness of the top coal as the grading index，the roof of mining roadway in 16–3 coal seam is divided into three levels. The control strategies of different roof levels dominated by grouting cable bolt and cable bolt synergistic anchoring are given，and good engineering application results have been achieved.

Predicting the mechanical response of underwater shield tunnel is essential to maintain the long-term stability. Therefore，a multi-factor coupling model driven by SHM(structural health monitoring) data is developed on the basic of deep learning algorithm，where Autoencoder and RNN(recurrent neural network) are adopted to learn the spatial and temporal dependencies respectively，and considering the effect of external load applied on tunnel structure. As a study case，the presented model is formalized on the monitoring data obtained from the SHM system installed in Wuhan Yangtze River tunnel. A series of data experiments are conducted to discuss the reasonability and predicted capability of the presented model. Experimental results indicated structural historical behaviors especially in last 15 days，spatial correlation，temperature，and water level are the main factors affecting the future mechanical behavior. The predicted capability of model dropped with the increase of prediction time scale. Compared with some typical models，the model presented in this study expressed the best performance，whose learning rate reaches 96%，and predictive accuracy exceeds 93% in next two weeks. As a promising application，the proposed model can be used to predict structural mechanical response under extreme conditions and provide scientific guidance for structural stability evaluation.

To study the seismic dynamic response characteristics of the tunnel crossing slope，a shaking table test was designed and carried out. The acceleration and dynamic strain responses of the cross tunnel and the slope are analyzed based on the test results. The spectral characteristics of the slope are analyzed quantitatively by introducing the Hilbert-Huang transform. The conclusions are as follows：(1) The acceleration responses of cross tunnel structure exhibit significant regional differences. Furthermore，the acceleration response is particularly intense at the crown and invert of the upper-span tunnel，the intersection，and the portal section of the under-crossing tunnel. (2) The tunnel lining is subjected to tensile-compression cyclic loading under earthquake. Additionally，the crown and invert of the upper-span tunnel bear relatively large pressure. The seismic additional moment is mainly concentrated in the left arch waist and invert，which should be the main seismic fortification area. (3) There is a synergistic deformation effect between the tunnel structure and the slope when the tunnel crosses the slope. The presence of the upper-span tunnel will accelerate or worsen the seismic damage in the middle and upper parts of the slope. Simultaneously，the seismic inertia force and landslide thrust are counteracted on the tunnel structure in the process of slope deformation and instability，resulting in lining damage or even failure. (4) The marginal spectral amplitude of each measurement point increases with the increase of elevation. The filtering effect of seismic waves is different for rock stratum interface such as bedrock，weak interlayer，slope，and lining structure. Furthermore，the spectral composition of seismic wave has transformed，resulting in a sharp change of marginal spectral amplitude. (5) The seismic energy response of each measuring point migrates to a relatively low frequency with the increase of the input seismic amplitude. The marginal spectral amplitudes of the crown of the upper-span tunnel and top of the slope show multiple peaks whose abrupt changes are the most pronounced. It shows that the upper part of the slope，especially the crown of the upper-span tunnel and top of the slope，is a sensitive area under seismic loading.

In order to promote the green development of geotechnical engineering and improve the resource utilization efficiency of solid waste，a green and high performance red mud-based grouting material was prepared by red mud，blast furnace slag，steel slag and desulfurized gypsum. The effects of red mud content and water to cement ratio on rheological property，setting time，mechanical strengthof red mud-based grouts were explored. The hydration characteristics of red mud-based grout are revealed based on the rheological kinetic characteristics of slurry. The leaching characteristics and stabilization mechanism of heavy metals in red mud-based grout were determined. The reinforcement mechanism of red mud-based grouts to the soft soil stratum，and the interface of the grout-soil were analyzed. The results show that the engineering performance of red mud-based grout is comparable to that of Portland cement slurry. Red mud has morphological effect and hydration characteristics，which can improve the rheological properties of slurry，but its gelling activity is low，with the increase of its content，the mechanical strength of stone tends to decline. When the content of red mud is 40%–60% and the water cement ratio is 0.8–0.9，the red mud-based slurry has the better working performance. The results of heavy metal analysis show that the leaching amount of heavy metals in red mud-based stone body is less than 0.2 ppm，which has the characteristics of green environmental protection for geological environment. The grouting simulation test shows that the red mud-based grouting material is mainly compaction-splitting diffusion in soft soil stratum，and can effectively improve the compressive strength of soft soil stratum by 47.6%. The test shows that the red mud-based grouting material has obvious performance advantages and better engineering applicability. The research results can provide green high-performance and low-cost grouting materials for underground engineering construction，and will also provide ideas for the resource utilization of solid wastes such as red mud.

The study on the leakage risk of oil-gas pipeline in coal mining subsidence basin is essential to protect the pipeline's safety and maintain the stability of transportation projects. In this paper，the oil-gas pipelines in the loose layer of Ordos Basin are taken as the research object. First，the different coupling states between pipeline and soil are analyzed. And the mechanical models of elastic beam and elastic foundation beam are used to calculate the pipe deflection. Next，the deformation limit of the pipeline is calculated according to the industry code. And the reliability theory is used to link the pipeline's deformation and the leakage damage. Second，the pipeline deflection and its deformation limit are compared and segmented. The axial damage risk level of the pipeline was determined. On this basis，combined with the influence range of pipeline leakage，the quantitative partition method of pipeline leakage risk in subsidence basins is established. Meanwhile，the corresponding correction method and protection scheme are proposed. Finally，the engineering example is compared to the theoretical solution，and a good agreement is obtained. The results show that the risk level of pipeline damage increases with the gradual increase of pipeline deflection. At the same time，the impact area of leakage gradually increases. The method realizes the quantification of oil-gas pipeline leakage risk in subsidence basins，which is essential to increasing maintenance efficiency and protecting pipeline safety.

Visualization test of whole shear process is of great significance to study progressive failure characteristics of rock joints. The study combines three-dimensional laser scanning，printing technology with transparent photosensitive resin material(veroclear) to produce high-precision transparent models of rock joints. Carry out uniaxial compression experiments at different temperatures，the experimental results show that the performance of this material is similar to that of rock material at －30 ℃–－45 ℃. Local modification of shearing equipment based on semiconductor refrigeration technology，The refrigeration performance tests under different working conditions show that after 50 minutes of operation，the cooling performance of semiconductor refrigeration equipment is gradually reduced by forced convection，and the temperature inside the cavity is maintained at about －17 ℃. Combined with the dark channel prior defogging algorithm and the visual monitoring method，the surface morphology of the whole shear process of the rock joints surface can be obtained more clearly，which effectively solves the problem of specimen fogging low temperature test conditions. The shear test shows that the temperature of the sample is always lower than －30 ℃ in the whole process of the test at low temperature，which meets the temperature requirements of photosensitive resin material as rock-like characteristics. The feasibility of the test equipment and test technology is further verified，which creates conditions for a visual test method for the whole shear process of rock joints.

It is difficult to estimate the accurate value of the shrinkage limit by conventional methods(i.e.，calculation method and graphic method) when the residual shrinkage stage of SSCC cannot be negligible. The soil-water characteristic curve(SWCC) and the soil shrinkage characteristic curve(SSCC) are applied to explore the method of determining the shrinkage limit of remolded clay in the present study. The results show that a simple method for correctly estimating the shrinkage limit of remolded clay is proposed based on the idea of combining numbers and shapes. This method does not require a complete shrinkage curve，but only requires two main parameters，namely the water content corresponding to the air entry value and the volume change when the sample is completely dried. The accuracy and applicability of the proposed method are verified by using experimental data from both laboratory and literature available. When the residual shrinkage phase of SSCC can be neglected，the shrinkage limit formula deduced in this study is essentially consistent with the specification，but the former is simpler and has a clear physical meaning. Combined with conventional SWCC experimental data，the formula can also be employed to predict the SSCC of remolded clay and its SWCC based on degree of saturation. The predicted curve is in good agreement with the experimental data. These findings are very important for the subgrade deformation calculation of cohesive soil，as well as the identification of the expansion and shrinkage grade of expansive soil.

The existing methods for solving the load problem of layered foundation are not perfect. The transfer matrix method often has ill conditioned matrix and data overflow，while the analytical layer element method or stiffness matrix method has some defects，such as difficult inversion of high-order matrix. Firstly，the matrix solution of soil state variables with undetermined coefficients is derived based on the basic equation of three-dimensional elastic statics and Fourier positive transformation. Then，combined with the boundary conditions，the transformation relationship of undetermined coefficients of soil state variables and the inverse Fourier transform，the known coefficient solution of layered foundation state variables is obtained. Finally，the method verification and parameter analysis are carried out for homogeneous semi infinite foundation，three-layer unequal thickness foundation and generalized Gibson foundation. The results show that compared with transfer matrix method and analytical layer element method，this method is not only reasonable and accurate，but also has the advantages of stable numerical calculation，faster solution speed，self-check and unlimited number of applicable soil layers. The relative displacement of interlayer contact surface is directly proportional to the interface slip coefficient and inversely proportional to the distance between loading surface and interlayer contact surface. The change value of interface relative displacement is twice the change value of load distribution width. The number of layers of generalized Gibson foundation is 40，which can meet the requirements of calculation accuracy. Under different Poisson's ratio and load depth，the volume strain and displacement decrease with the increase of heterogeneous parameter m. The influence of parameter m on the vertical state variable of soil is more significant，and the value of m is more noteworthy when it is 0.0–0.5.

Clay-gravel mixtures(CGMs) are composed of clay and gravel. With the accelerating construction of engineering projects in cold regions，it is of great importance to figure out the mechanical properties of frozen CGMs. In this study，CT scanning tests were first carried out on compacted CGMs at room temperature，the skeleton structure and pore structure distribution characteristics of samples with different gravel contents were obtained. Besides，a series of unconfined compression tests were carried out on frozen CGMs to study the effects of temperature，gravel content，water content，and loading rate on their stress-strain characteristics and failure strain energy densities. The test results show that frozen CGMs exhibit typical brittle failure characteristics，which are greatly affected by temperature，gravel content，water content and loading rate. The failure strain energy density decreases with increasing temperature and gravel content，and increases nonlinearly with the increase of the loading rate. Furthermore，the failure strain energy density of the frozen CGM first increases and then decreases with water content，indicating there is a“critical water content”at which the frozen CGM bear the strongest resistance to failure.

A total of 150 single compression tests and 18 uniformly-graded sample one dimensional compression tests for three different size groups carbonate sand were conducted to investigate the crushing characteristics of single particle and uniformly-graded sample and their association relationship. According to the load-displacement curve of single particle compression，crushing strength and ultimate strength were defined to distinguish two crushing states，i.e.，the local crushing state and global crushing state，and then the strength characteristic and the fractal characteristic of single particle were discussed by considering two different crushing states. Through detail analysis of uniformly-graded sample grading evolution subjected to one-dimensional compression，the crushing fractal evolution and fractal limit of uniformly-graded sample was explored. Based on fractal theory，a fractal evolution function for uniformly-graded sample was established by introducing the calculation of breakage ratio，which reveals the evolution mechanism of fractal behavior of uniformly-graded sample and its relationship with fractal characteristic of single particle. The feasibility of the proposed fractal evolution function for uniformly-graded sample was finally verified using the experimental data.

Rubber sand mixture(RSM) can be used as a cheap energy dissipation material. Reinforcing RSM with soilbags can improve its strength and stability as a base isolator beneath the foundation of structures cushion. A large cyclic shear test device was used to study the hysteresis characteristics of soilbag reinforced rubber sand mixture(SBRSM) under cyclic shear load，and the effects of the main factors such as the mass content of the rubber particles，the vertical pressure，the strain amplitude，the filling ratio of soilbag and the arrangement type of the soilbag on the two dynamic parameters，i.e.，the dynamic shear modulus and the damping ratio，of SBRSM were investigated. The results indicate that：(1) the dynamic shear modulus( ) of SBRSM increases exponentially with the vertical pressure，and decreases with the rubber content(RC) with approximately linear lg -RC relationship，and the slope of the lg -RC relationship decreases slightly with the strain amplitude increasing. (2) In the strain range greater than 1%，the damping ratio of SBRSM decreases slightly with both of rubber content and strain amplitude，and decreases in power function with the vertical pressure. (3) of SBRSM decays with the decrease of filling ratio. Though the damping ratio is less affected by filling ratio of soilbag，a lower filling ratio would reduce the influence of vertical pressure on the damping ratio. (4) Both lower and higher damping ratio of SBRSM would be gained when laying soilbags vertically comparing with laying them horizontally，and the horizontal shearing direction affects the dynamic properties of SBRSM little when soilbags are paved horizontally. The mechanisms of dynamic deformation characteristics of SBRSM are revealed by comparison of the dynamic characteristics between SBRSM and RSM without soil bag reinforcement.

Miniature pore water pressure transducer is one of crucial testing techniques in geotechnical engineering monitoring and physical model test. With centrifuge model test featured by factors of centrifugal acceleration，viscous liquid and high frequency vibration，influences of these factors on miniature pore water pressure measurement are worth to explore and understand. A set of centrifuge tests consisting of dry sand，saturated sand and pure liquid models are carried out to calibrate the measuring reliability of pore water pressure transducers by varying a single state parameter in the tests，the response characteristics and measurement error of three typical miniature transducers are revealed and analyzed. The results show that as increasing centrifuge acceleration，amplitudes of all transducers in the dry sand model present distinct increasing or decreasing trends independent of their buried depths. However，as centrifuge accelerations less than 50 g，all variations of the amplitudes are less than 1.0 kPa，indicating that the influence of the centrifuge accelerations is negligible. With viscosity coefficient ranging from 1cst to100cst，response time and dynamic resolution of the transducers are both impaired. For the transducers of 2Mie，DSP–II and HC–25，the response time grows from 4.92，4.98 and 10.75 ms to 13.42，13.15 and 20.01 ms，while their dynamic resolutions weaken from 54，59 and 121 Pa to 109，113 and 213 Pa. Such influence are worthy of taking into account for the study of liquefaction threshold and pore water pressure increment relation. The influence of transducer placement directions on pore pressure measurement in the saturated model test is noticeable. The pore water pressure ratios of the transducers placed parallel to the shaking direction exhibit unexpectedly remarkable oscillation with the magnitudes varying from 0.78 to 1.02. Simultaneously，the pore water pressure ratios of the transducers placed vertically arise considerable values between 1.04 and 1.20. The influences arisen in parallel and vertical placement are likely attributed to the disturbances of dynamic cyclic shear stress and excess pressure-induced pore water seepage respectively. The intensive electromagnetic interference on the transducer responses appears inconsiderable. The calibration method and results provide important guidance and dataset for the development pore water pressure measurement techniques and the revision of specifications for dynamic centrifuge model tests.

The tram pile plate structure，a new embankment structure，is proposed in recent years. In order to achieve the purpose of bearing the upper load by the piles and soil between the piles，friction piles with sparse arrangement are used in the embankment. The embankment settlement and soil displacement field of the tram pile plate structure are studied by centrifugal model test. The results show that the tram embankment settlement meets the requirements of the code. The upper load is shared by piles and soil between piles，and there is no void under the plate. Pile length and pile spacing have a significant effect on embankment settlement. The embankment settlement increases with the increase of pile spacing，while it decreases with the increase of pile length. The embankment settlement increases gradually at the initial stage of operation. In the later stage of operation，the increasing degree of embankment settlement is small，and the embankment settlement tends to be stable. The settlement of the soil around pile decreases gradually along the depth，and there is an inflection point near the pile end. The relative displacement of pile-soil does not have a negative value and gradually increases with the depth. The settlements of the soil on pile tip and under plate are the same. There is differential settlement between the soil around the pile and in the middle of the span. The greater the depth，the more obvious the difference.

Under freeze-thaw cycling，the soil structure was failed，soil porosity was increased，and the pore distribution characteristics was varied，resulting in the degradation of the soil shear performance. A series of tests on lime-soil and fiber-lime-soil，including freeze-thaw test，the triaxial compression test，and NMR test，were completed to research the shear performances and microstructure variation of lime-soil and fiber-lime-soil under freeze-thaw cycling. The results showed that the cohesion and internal friction angle of lime-soil and fiber-lime-soil gradually decreased with the increase of freeze-thaw cycles. The freeze-thaw deterioration was divided into three stages，namely，large strength decline stage，small strength decline stage and strength stability stage. The stress-strain characteristics of lime-soil transformed from strain softening to strain hardening because of adding fibers，and the soil failure form changed from brittle failure to plastic failure. With the increase of freeze-thaw cycles，the porosity and the pore radius of soil increased，the micro pore volume decreased，and the small pore volume，medium pore volume and large pore volume increased. After 10 freeze-thaw cycling，the large pore volume varied little. Because of the spatial restraint effect of fiber on soil and the friction between fiber and soil，the increase of pore radius and porosity of soil were limited to a certain extent，which resulted in the improvement of the shear performance and freeze-thaw resistance of fiber-lime-soil.

For the problem of one-dimensional large deformation consolidation of soft soils with time-dependent loading，based on the existing governing equation of consolidation，the general analytical solutions for one-dimensional large deformation consolidation are obtained by using the method of variable substitution and separation variable. The method used in solving the general analytical solutions is simple，and the solutions have a clear form，which is convenient for calculations. Referring to the general analytical solutions，the expressions for analytical solutions of consolidation under several common loading patterns are given. By analyzing the specific analytical solution presented in the paper and comparing the proposed analytical solution with finite difference solution，the correctness of the proposed analytical solutions is verified. Based on the solutions proposed，the consolidation behaviors are analyzed. The results show that under two-stage instantaneous loading pattern，the consolidation rate slows down and the maximum value of excess pore water pressure decreases with the extension of first-stage loading time. Under two-stage linear loading pattern，the effect of increasing the loading rate on the consolidation rate is small when the loading rate increases to a certain extent. Under cyclic loading pattern，with the increase of cyclic loading time，the variation range of the consolidation degree curves becomes larger after it tends to be stable. Under the time-dependent loading，the consolidation rate defined by settlement increases with the increase in the final loading，while the consolidation rate defined by pore water pressure decreases. The analysis results can provide theoretical guidance for the actual loading construction process of soft soils.

The sinking speed prediction of open caisson foundations has important practical significance for ensuring the sinking safety and sinking steady，and to prevent potential construction risks. Based on the structural stress monitoring data at the bottom of open caisson foundations，a two-dimensional convolutional neural network (CNN) and a three-dimensional CNN are applied for proposing a sinking speed category prediction model and a sinking speed value prediction model. The spatial and spatial-temporal characteristics of the structural stress monitoring data are extracted to predict the sinking speed. The accuracy and practicability of the two prediction models were verified by applying them to an open caisson foundation of the main tower in the Changtai Yangtze River Bridge Project. Then，the real-time prediction of the sinking speed category in the sinking process was simulated，and the influence of the prediction step and the spatial-temporal characteristics of the structural stress on prediction accuracy were analyzed. The results show that the proposed models can successfully predict the category and value of the sinking speed. The reliability and practicability of the models were verified in practical engineering that the proposed sinking speed prediction models have good prediction performance. Moreover，the spatial-temporal characteristics of structural stress have an important influence on prediction accuracy. The real-time sinking speed prediction of open caisson foundations was achieved，which has important reference value for intelligent decision-making in the monitoring during sinking process.

考虑到宽幅路基的“聚热效应”和复杂的多年冻土环境，拟建青藏高速公路建设所面临的关键问题是如何保证路基的长期热稳定性。基于现场监测数据和传热传质理论，建立分离式通风管路基三维数值模型，分析与预测未来50年通风管在青藏高速公路分离式路基中的工程效果。结果表明：分离式通风管路基具有较好的降温效果，能够保证路基及其下部多年冻土的长期热稳定性。但是，当隔离带宽度小于10 m时，后幅路基管道内部风流变化特征受到隔离带宽度的显著影响，导致其对下部多年冻土的降温效果弱于前幅路基。此外，在隔离带较窄情况下，两幅路基之间隔离带区域存在局地增温效应，将对路基及其下部多年冻土产生严重的热扰动，不仅引起多年冻土上限下降、温度升高，而且增加了路基两侧下部土体温度场的不对称性。尝试将两幅路基通风管连通来弱化局地增温效应，但连通后路基及隔离带区域下部土体的热状况反而更差，说明这种方法未能有效解决局地增温效应的热影响。

In order to research the applicability of glutinous rice slurry for ancient ruins restoration under drying-wetting cycles，the ancient ruin samples with different contents of glutinous rice slurry were prepared by static pressing，the dynamic triaxial test was carried out after different times of drying and wetting cycles to obtain the dynamic characteristics and structural evolution law of the artificially prepared ruins. The results show that the dynamic strength of soil can be improved significantly and the dynamic deformation can be controlled effectively，When the mass ratio of soil to glutinous rice slurry is 90∶10，the backbone curve and dynamic shear modulus of the sample are the largest，and the damping ratio is the smallest. The drying-wetting cycles reduce the backbone curve and dynamic shear modulus of the artificially prepared site soil，and increases the damping ratio. With the increase of the content of glutinous rice slurry，the influence of drying-wetting cycles on the dynamic characteristics of soil gradually weakened. Dynamic shear modulus ratio and damping ratio tend to be stable after 5 cycles of wetting and drying，nonlinear regression analysis is performed on them respectively，and the corresponding normalized fitting model was established. The dynamic structural parameters defined by the comprehensive structural potential theory can well quantify the deterioration degree of soil structure in the drying-wetting cycles and shear process，The structural parameters decrease with the increase of drying-wetting cycles and confining pressure. When the content of glutinous rice slurry increases，the structural parameters first increase and then decrease.

A series of isotropic consolidation tests，three-dimensional net principal stresses loading tests and wetting tests were carried out on intact expansive soils under measuring suction by using unsaturated soil collapsible true triaxial apparatus to simulate the swelling deformation characteristics of expansive soils in practical engineering. The effects of net average stress on suction and compression deformation were studied. The variation of suction under three-dimensional net principal stresses loading，and the effects of intermediate principal stress，stress ratio and net confining pressure on suction and swelling deformation are discussed. The results show that the suction decreases with an increase in net average stress，and the smaller net average stress(p≤100 kPa) has little effect on the variation of suction and pore ratio，while the larger net average stress(p＞100 kPa) has a great influence on the change of suction and pore ratio. From the e-lnp relationship curve，it is determined that the yield stress of expansive soil is = 100 kPa，and the yield suction is = 86.4 kPa. When shearing，the suction decreases slightly or remains unchanged at first，q/p＞1，the suction decreases again and then tends to be stable，q/p≤1，the suction increases again and then tends to be stable. When humidifying，the suction decreases with an increase in moisture content， intermediate principal stress parameter b and net confining pressure ，but increases with an increase in k value. There is a three-dimensional deformation when the intact expansive soil is humidified. In this paper，the deformation forms of expansive soils can be summarized into four forms，which is closely related to the intermediate principal stress parameter b，stress ratio k and net confining pressure .

Gas hydrate reservoirs are characterized by loose，weakly cemented，and the framework particles are finer，which causes the sand production more complicated and serious. Indeed，sand production is one of the most serious challenges restricting the efficient exploitation of gas hydrates. According to the gas hydrate reservoirs in the Shen hu area of the South China Sea，we design and carry out experiments of sand production to reveal the mechanism during the exploitation of gas hydrate，achieving predicting the sand production for fundament of taking effective sand control measures during hydrate exploitation. The self-built multi-field and multi-scale coupling experiment platform was used to carry out the core displacement sand production experiment. The nuclear magnetic resonance(NMR) instrument was used to monitor the pore distribution and structural changes in the core during the sand production process in real time. The produced sand is timely collected，the influence of flow rate on sand production rate and sand production are analyzed as well. Then，a sand production model for hydrate reservoirs is established based on the theory of hydraulic erosion，and the model parameters are calibrated according to the experiment data. The results show that the sand production is mainly affected by the water erosion. The greater the water flow，the greater the sand production rate and the amount of sand. Besides，the sand volume is linear with water flow，thus the sand production can be efficiently predicted by the established model.

Swelling is one of the key factors affecting the stability of expansive soil slope. At present，the assumption of expansive force is insufficient and the mechanism is not clear when analyzing the stability of expansive soil slope. In response to these problems，the multidirectional swelling under specific constraints and its influence on shear strength are being considerd based on the Morgenstern-Price limit equilibrium method，and a project is taken as an example to illustrate the calculated process，which also is used to analysis the influence of swelling and residual shear strength on the stability of expansive soil slope. The results of the study presented in this paper show that constrained condition play a significant effect on the swelling pressure，which will be markedly reduced when there is a micro gap. The calculated swelling pressure of the slope presents a“trapezoidal”distribution，which transform the shear strength into residual strength when it causes the swelling strain exceeds the critical strain，and there is a critical swelling pressure，which makes the stability coefficient of the expansive soil slope suddenly change. The residual shear strength has dual effects on the stability of expansive soil slope，on the one hand，which affects the position with respect to the junction of residual and non residual area，and the other is the direct impact as a mechanical parameter for slope stability calculation. The research of the study are provided for reference for further improving the calculated theory of expansive soil slope stability and carrying out the research on corresponding engineering prevention measures.

In the in-situ state，the clay rock is saturated，but in the actual laboratory study，the clay rock has lost water and is unsaturated due to the influence of drilling core，transportation，storage，and processing. Therefore，restoring the clay rock to the in-situ saturation state as far as possible before the further experimental test is an important prerequisite for further experimental tests. This paper takes clay rock as the research object，through the clay rock saturation test without external force and the saturation test under in-situ stress，the focus is on the hydration expansion properties of clay rock and the deformation law in the process of saturation from a macro and mesoscopic perspective. The research shows that with different water saturation time，under the action of no external force，the claystone with central circular pores gradually fills the original pores due to the expansion，disintegration and mixed accumulation of clay minerals. Under the in-situ stress，the action of brine still causes the rapid hydration and expansion of the clay minerals in the clay rock，and the macroscopic appearance of the sample volume increases rapidly while the porosity increases significantly. The expansion deformation is negatively correlated with the initial saturation of the sample. Further，based on the Comsol Multiphysics multi-field coupled finite element，the fluid flow module in the software establishes the geometric model of the clay rock sample and carries out the numerical simulation of the saturation process of the clay rock. Finally，pore pressure coefficient B during saturation and the effect of loading steps on the clay rock saturation process are discussed. The research results will provide experimental guidance and theoretical basis for the related research on the physio-mechanical properties of clay rocks and other rocks.

Biochar has a porous structure，higher specific surface area and stronger adsorption characteristics. Adding biochar to the traditional clay cover of landfill，i.e.，biochar-amended clay cover，so as to achieve the purpose of biologically reducing the methane released from the landfill. Adding biochar could change the pore structure of the soil and improve the air mobility in the soil, but the mechanism of influence on the air flow characteristics of the landfill cover is not clear. Therefore，the air permeability characteristics of biochar-clay mixed soil were used as the main evaluation indicators to explore the influence law and mechanism of biochar content and particle size on the air permeability characteristics of biochar-clay mixed soil. The air permeability coefficient ka of biochar-clay mixed soil with different biochar contents and different biochar particle size groups was measured by using the self-developed flexible wall air permeability testing device，and the changes of pore size distribution are analyzed by nuclear magnetic resonance(NMR) technique. The research results show that for the biochar-clay mixed soil whose dry density is not less than 90% of the maximum dry density of clay，there was an optimal biochar content，and its air permeability coefficient is the lowest. When the content of biochar is less than the optimal value，the air permeability coefficient decreases with the increase of the content of biochar, and the trend is more obvious at lower dry density. When the biochar content was larger than the optimal value，the air permeability coefficient gradually increases with the increase of the content of biochar. Comparing the air permeability coefficient of biochar-clay mixed soil with different biochar particle size groups without overlap，the permeability coefficient decreases with the decreased of biochar particle size. The NMR microscopic test results were used to explain the change mechanism of the air permeability coefficient of biochar-clay mixed soil increasing with the increase of biochar content and particle size under specific conditions. This study can provide a reference for the selection of biochar dosage and particle size for the design and construction of the final cover layer of biochar-amended clay in landfills.

受现有土工试验仪器功能的限制，目前仅能实现绕单一应力主轴旋转的应力路径，对应力绕2个及以上应力主轴旋转的研究仍是土力学的空白。从控制试样沉积面旋转和控制应力加载方向2个角度，系统梳理应力方向依赖性的试验研究方法。设计研发倾斜沉积砂土空心圆柱试样的制样装置，在砂雨法制备空心圆柱试样技术的基础上，通过控制电机转速使成样模具整体定速转动，使得砂土颗粒在重力和离心力的作用下沿径向倾斜沉积，进而制得沉积面与水平面夹角为定值的倾斜沉积砂土空心圆柱试样，实现试样沉积面绕 的定向旋转。再借助扭剪仪控制应力绕 旋转，即可实现对应力的3个大小量和2个方向量的定量控制。采用生石灰–彩砂试制18个沉积面倾斜的试样，对沉积面倾角进行图像分析和统计学分析，结果表明，所制试样的沉积面倾角与理论计算值较为接近，倾角偏差的分布较符合正态分布，倾斜沉积砂土空心圆柱试样制样装置的制样可靠度较高，为进一步揭示岩土材料特殊的力学性质、丰富和完善土力学理论奠定了技术基础。