The effects of in situ stress，cumulative displacement and stress-strain relationship on the concentration，release，storage and dissipation of energy in surrounding rock masses are investigated in this paper. The energy released from the surrounding rock was found to decrease with the growth of damaged range. A theoretical model considering the combined effects of surrounding rock pressure and disturbing loads was proposed. The transformation of potential and kinetic energy in surrounding rock masses under the disturbing loads was clarified. The characteristic values of the energy threshold corresponding to different damaged zones were given. The key mechanism of zonal disintegration was revealed to be the magnitude of the combined characteristic energy. The accuracy of the theoretical and calculation method was validated by comparing with the in situ and in-door experimental data.

A novel multi-functional shear-flow coupled test system for rock joints was developed to reveal the (long term) mechanical properties，flow laws and failure mechanism for rock joints. The test system is mainly consisted of four parts：a host frame and servo loading system，a shear-flow box and sealing system，a seepage servo control system，a data measurement and acquisition system. The dynamic sealing problem of coupled shear-flow tests with 10% shear displacement(20 mm) of joint length under the water pressure of 3 MPa was settled. The loading methods of constant normal stress，constant normal displacement and constant normal stiffness can be implemented. The electronic balance and flow meter were used together to solve the problem of wide-range flow measurement. The pressure stabilizing system and differential pressure control system were developed to keep the water pressure constant in the water inlet and outlet. A hydraulic servo loading system under quasi static condition and a motor servo loading system under creep condition were designed to accomplish the corresponding experiments. The hydro-mechanical coupling experiments under three stress boundary conditions were performed to certify the accuracy and reliability of the test system.

A two-dimensional normal information diffusion method was proposed to infer the probability distributions of geotechnical parameters. A general expression of two-dimensional normal information diffusion model was established considering the correlation of two-dimensional geotechnical parameters and the principle of normal information diffusion. The two-dimensional distribution K-S test method was improved and the corresponding specific test procedure was proposed. The two-dimensional shear strength parameters of the geo-materials at a hydropower station project were used as an example，and their distributions were fitted with the two-dimensional normal distribution and two-dimensional normal information diffusion distribution. The proposed two-dimensional distribution K-S test method was used to test the fitting accuracy of the two distributions obtained above. The results show that the testing value of the two-dimensional normal distribution is lower than that of two-dimension normal distribution and that the distribution function of two-dimensional normal information diffusion is closer to the actual distribution of data，showing that the two-dimensional normal information diffusion distribution is better in describing the probability distribution of two-dimensional geotechnical parameters. The edge distribution of the two-dimensional normal information diffusion function obeys the one dimensional normal information diffusion distribution，and the accuracy is also higher than that of the one dimensional normal distribution.

The mechanism and characteristics of coal cracking under temperature shock was studied with micro-CT. The high-low temperature test system SLX–80 was used to exert the cold and heat shock on the raw coal with four temperature differences designated. The industrial micro-CT system was used to scan the coal samples before and after the temperature shock so that three-dimensional fracture structure of coal samples may be reconstructed with the image analysis system VG Studio MAX. The characteristics of cracking process were described quantitatively. The transient thermal stress theory was used to analyze the coal breaking mechanism under temperature shock. The results show that the temperature shock promotes the expansion and widening of cracks within the coal body and initiates new cracks. The crack volume，wall thickness and surface area are all positively correlated with the temperature difference，indicating that the thermal stress generated by different temperature differences determines the degree of damage of coal samples. A large number of acoustic emission signals were generated in the process of temperature shock and mainly concentrated in the first 600 s. The greater the temperature difference is，the higher the acoustic emission energy. The maximum thermal stress generated is tangential to the coal surface. The thermal stress exceeding the tensile strength of coal samples cause directly the initiation，propagation，and interpenetration of fissures.

For the single continuous jointed rock mass with different dip angles，the concepts of initial joint damage，load damage and total damage were proposed. Based on the damage mechanics，a damage evolution model and a damage constitutive model were established considering the coupling of structural effect and loading. The method of total differential was employed to establish the strength criterion of jointed rock mass considering the effect of the dip angle of joints. The models were verified and analyzed. The theoretical curves of the models constructed in this paper agree with the experimental result，which verifies the correctness of the models. The initial joint damage shows an inverted U-type distribution that increases first and then decreases with the increase of the dip angle of joints. The ratio of the maximum to minimum values of the initial joint damage reaches 2.13，indicating that the dip angle of joints has a great influence on the initial damage of the jointed rock mass. The total damage curve shows an S-type distribution that rises slowly first，then rises rapidly and finally rises slowly to 1 with the increase of the dip angle of joints. The total damage rate curve shows a normal distribution that increases first and then decreases with the increase of the strain and a U-type distribution that decreases first and then increases with the increase of the dip angle of joints. The total damage and total damage rate decrease with the increase of confining pressure，indicating that the confining pressure has an inhibitory effect on the damage evolution.

The dynamic response，variation of stability coefficient，deformation process and failure modes of two kinds of bedding rock slopes with different dip angles under frequent microseisms were investigated with the shaking table test and discrete element method UDEC. Under frequent microseisms，the natural frequencies show a downward trend，whereas the damping ratios present an upward trend. The structural damage of the slope keeps developing and accumulating and is manifested as the initiation，propagation and penetration of the bedding planes and secondary joints. The permanent deformation of the slope increases gradually and the stability coefficient decreases. The acceleration responses of two kinds of slopes show the “elevation effect”and“aspect effect”，and the dynamic response decreases with the increase of the loading time of microseisms. For the gently dipped bedding slope being loaded with strong earthquakes at the later stage of the shaking table test，the failure of the slope presents the layer-by-layer cracking and shedding from the top to the bottom，but the damages under the frequent microseisms simulated by UDEC are mainly concentrated on the slope surface. For the steeply dipped bedding slope with non-exposed layer parallel to the slope surface，the stability of the slope under microseisms is generally good. The failure surface under strong earthquake is mainly composed of three parts：a steep edge section of the trailing edge，an undulating section of the middle step and a gentle shear section of the leading edge.

In order to study the bond softening effect on rock failure using the discrete element，the softening relationship of the bond crack was implanted into the Flat-Joint constitutive model. The softening effect is considered when the bonds of particles are breaking. PFC2D can call the subroutine(DLL) of constitutive model written in C++. According to the stress-strain curve，the variation of the numbers of softened bonds and broken bonds，the uniaxial compression process of rock is divided into 5 stages：a linear ascent stage，a nonlinear ascent stage，a nonlinear descent stage，a brittle stage and a residual stress stage. Bond softening occurs mainly at the nonlinear stages. The effects of strength and strain softening on the macroscopic parameters such as the peak stress，the length of nonlinear zone，the number of softening bond and Poisson ratio were analyzed in detail according to the horizontal softening curve. Finally，the fracturing energy of bond was analyzed preliminarily with the composite failure criterion. The method of simulating the rock failure with discrete element was improved.

With the large testing system for tunnel seepage，experiments were carried out to analyze the influence of the rock permeability coefficient，the initial height of water head and the rates of drainage on the distribution of seepage field in surrounding rock and the water pressure on lining. The calculation formulas for the influencing range of seepage field and water pressure of lining were presented. The obtained results were verified by numerical simulation. The influencing range of seepage field is affected by the permeability coefficient of surrounding rock，the initial height of water head and the rate of tunnel drainage. The analysis of the correlation between the above factors and the influencing range of seepage field by the single variable method proves that the greater the permeability coefficient and the rate of tunnel drainage，the larger the influencing range of seepage field. The effect of permeability coefficient on the influencing range of seepage field is greater than that of the rate of tunnel drainage. There is a linear positive correlation between the influencing range of seepage field and the initial height of water head. The influencing range of seepage field coefficient is related to the rate of tunnel drainage and permeability coefficient，but has nothing to do with the initial height of water head. The influence range of seepage field coefficient，the rate of tunnel drainage and permeability coefficient satisfy the quadratic surface equation. The average water pressure on lining，the influencing range of seepage field and the rate of tunnel drainage also satisfy the quadratic surface equation. The calculation formula for the average water pressure was obtained. The above formulas are in good agreement with the numerical results.

In order to obtain the permeability variations of coal samples with different heights under cyclic loading，triaxial loading seepage experiment device was used to carry out the cyclic axial loading and unloading compression tests on coal samples with the same diameter and different heights. The results show that the permeability decreases with the increasing of the axial compression in the loading stage. The permeability increases with the decreasing of the axial compression in the unloading stage. The relationship between the permeability and the axial compression is a negative exponential function and the stress sensitivity of permeability decreases with the increasing of cycle numbers. The permeability has a significant difference during the loading and unloading stage and the permeability difference decreases with the increasing of cycle numbers. The permeability reduction mainly occurs in the initial stage of loading and unloading. There are significant differences in stress sensitivity for coal samples with different heights. The stress sensitivity decreases with the increasing of the height to diameter ratios of coal samples. During the loading and unloading stage，the reduction of dimensionless permeability increases with the increasing of the height to diameter ratio. After cyclic loading and unloading，the reduction rate of permeability increases with the increasing of height to diameter ratio. The recovery rate of permeability decreases with the increasing of height to diameter ratio.

For transversely isotropic material，there is a significant difference in fracturing energy along and perpendicular to the bedding plane of layered rocks. In this paper，a fracture path criterion taking the fracture energy anisotropy into consideration was proposed. Under the linear elastic condition，the criterion is transformed into a three-parameter model of fracture toughness，elastic parameters and the ratio of stress intensity factors. The influence of the fracturing energy anisotropy on the hydraulic fracturing path was studied through finite element simulations with the criterion considered in the cohesive elements. The study indicates that the fracturing energy anisotropy is a dominant factor limiting the hydraulic fracture propagation. There is a limit in the ratio of the fracturing energy perpendicular to/along the bedding in different rocks. When the ratio for the fracturing energy perpendicular to/along the bedding is less than the limit value，the fracture will propagate across the layer. As the ratio is around the limit value，the fracture will extend across the layer with deflection. For the ratio is greater than the limit value，the fracture will be deflected along the layer interface.

In recent years，hydraulic fracturing is an important technique widely used in mining fields for methane exploitation in coal bed，gas drainage and rock burst preventions. In order to improve the speed and accuracy of arrival picks of microseismic P-phase in hydraulic fracturing of coal seam，a new method which combines a multi-threshold wavelet packet transform(MWPT) and an improved Hilbert-Huang transform(IAIC) was developed. The MWPT module has the multi-rules for different frequency bands based on wavelet packet transform method. The IAIC module is applied for speediness and precise determination of P-phase arrival time with two steps：rapid searching and accurate picking. Taking a typical microseismic event with 12 channels as an example，the methods of IAIC，ER，MER，WFM and PAI-S/K were used to pick up the first arrival time and compared to the manual picking results. MWPT method was proved to be applicable for microseismic signals from hydraulic fracturing of coal seam. The SNR has been advanced about 12.67 dB with reserved signal energy eigenvalue of 97.15%. Four methods except PAI-S/K method had the high accuracy in picking microseismic P-phase arrival，especially the IAIC method reduced the error from 34.17 ms to 7.08 ms. Although the methods of ER，MER，WFM are in the same time-window system，and sensitive to the waveform fluctuation，they still lag behind the manual result attributing to selective time-window. Compared to the manual method，IAIC method is more sensitive to signal denoising by MWPT method，with computation cost(0.124 s) and precision(7.08 ms)，the picking error between －10 ms and 10 ms is even up to 62.5%，between －20 ms and 20 ms is up to 100%.

To address the transient characteristics of landslide processes and to overcome the deficiency of static forecasting models，a model for predicting the transient landslide displacement was proposed based on time series theory and long and short term memory neural network(LSTM). In the model，the moving average method was applied to decompose the cumulative displacement into the trend term and periodic term. Subsequently，the trend displacement was predicted by a polynomial model. A LSTM model，based on the response of inducing factors，was established to predict the periodic displacement. Finally，the trend displacement and periodic displacement were superposed to achieve the cumulative displacement. Baishuihe landslide，a typical stepped landslide in Three Gorges Reservoir area，was taken as an example to test the prediction performance of the proposed model and the support vector machine(SVM) was used for comparison. The results demonstrate that the transient model(LSTM) achieves higher prediction accuracy than the static model(SVM)，especially during the period of stepped deformation. Furthermore，the prediction accuracy of the LSTM model is not limited by the timeliness analysis of training sets.

Backfill mining is often considered as one of the effective methods to control the movement of overlying strata and to protect the ground surface in underground metal mining. In order to analyse and evaluate deformation characteristics and current situation of overlying surface after 8 years of multi-level completely backfill mining in the west of exploration line 10 of Taohuazui gold mine，an AHP-Fuzzy evaluation model of stope safety was established and the safety analysis mode of“stopes to overlying strata and surface”was put forward. The whole backfill mining process according to the historical sequence of past 8 years was simulated，and the numerical results were compared with the monitored data of surface subsidence in the past 2 years. The results show that 22 of the totally 25 backfill stopes are in good safety condition and belong to level II. The existing condition of orebody，the mining method and the distance from the shafts of 11 evaluated factors are the most significant in influencing the stopes and current regional safety. The strata can be divided into“relative significantly disturbed area”and“micro disturbed area”. The relative significantly disturbed area tends to form a certain height of“Natural rock movement arch”with the morphology of approximately a 3D ellipsoid. The effects of ellipsoid on overlying surface are weak when the backfill mining reaches a certain depth. The micro disturbed area tends to form a“Micro subsidence basin”. The surface of mining area gradually changed from undisturbed status to 0.5–1 mm micro-disturbed status in 2014. The time effect of backfill on surface safety is weak. All the observation points of three closed loops arranged on the surface are in stable fluctuation，which fluctuate along the horizontal lines of 0，－3 and －6 mm respectively. The surface settlement did not exceed the allowed deformation value during the observation period. The calculated results agreed largely with the monitored data. The overall safety of surface is good.

Rock sheds are often used to protect the mountain roads from dry granular flow impact from landslides or avalanches. In order to further clarify the protective effect of cushion layer on rock shed，the impact mechanism of dry granular flow against rock shed and the cushioning effect of the cushion layer made of gravel were studied through the laboratory model experiment. The high speed camera recordings in the experiment showed that the gravel cushion layer deformed notably in the process of a dry granular flow impact. The force sensor measured the distribution of the impact force along the rock shed in the normal and tangential directions. The experimental results reveal that for the cases without a cushion layer，the maximum normal and tangential components of forces are close to the footing of the rock shed directly facing granular flow impact. With the addition of a cushion layer，the maximum impact force decreases and shifts approximately 30° to 45° away from the footing，indicating that the cushion layer can not only reduce the magnitude，but also change the mode of distribution of the impact force. It was found from the internal forces calculated that cushion layers can reduce the internal bending moment and internal shear force and increase the internal axial force，which enhances the safety of the rock shed. It was also verified that the finer granular material has higher capacity of energy dissipation. The experimental results fully indicated that a cushion layer played a significant role in buffering impact and dissipating energy in the process of a dry granular flow impacting on a rock shed.

A new kind of step tapered hollow pile with huge diameter was successfully applied in the stratum with bead-like karst caves. The load transfer mechanism of the pile is complicate. According to the analysis of the load transfer characteristics of the pile，the effect of enhancement of the resistance at variable cross-section to the shaft resistance was revealed. A new load-transfer function model for step tapered hollow pile with huge diameter was proposed on the basis of disturbed state theory. The bearing capacity at variable cross-section was analyzed，and the enhancement effect of shaft resistance caused by the end resistance was presented. An approach for calculating the enhancement coefficient   was proposed on the basis of Mindlin?s solution in semi-infinite space. The proposed analytical solution was verified by comparing with the numerical results and two existing analytical solutions. In the new load transfer model，the hyperbolic model and parabola model were selected to describe the relative intact state of pile-soil interface located on the pile side and the pile tip respectively，and Mohr-Coulomb model was used to assume the fully adjusted state. Thus，the load-transfer function model of pile-soil interface was proposed systematically. The effect of the model parameters was analyzed numerically. The proposed calculation model was found to be an excellent to describe the behavior of the step tapped hollow pile with large diameter. The resistance at variable cross-section has a significant enhancement effect on the shaft resistance. The load transfer function model based on the Disturbed State Concept theory can describe the nonlinear，hardening and softening characteristics of pile-soil interface. The enhancement effect of the shaft resistance caused by the resistance at variable cross-section can be revealed obviously.

This study proposed an elastic bearing pile-soil coupled model and obtained an analytical solution of soil around pile in frequency domain considering three-dimensional symmetric vibration of soil around piles. The semi-analytical solution of surrounding soil in time domain can be solved with inverse Fourier transform，and the validity can be proved by compared with the result solved by FEM software. On this basis，parameters of pile-soil coupled system can be analyzed during variable conditions，and laws of dynamic response of surrounding soil can be concluded as well，which can be used to provide practical guidance for engineering application.

Recent international damage surveys indicate that the performance of ground improved by cement-deep-mixing(CDM) method is generally better than expectations. Enhancing the understanding of the seismic response of CDM improved ground is important in developing suitable design methods of seismic resistance. Two series of dynamic centrifuge model tests to the composite soft clay grounds improved by CDM grids were presented. The unimproved ground and improved ground sites with embedded and floating CDM grids installed following a TRD-like method were tested. The responses and features of peak acceleration amplification，spectral ratios，ground settlement and shear stress-strain were analyzed. The results show that the acceleration responses of two embedded grid-improved grounds are linearly amplified at large and the peak accelerations are amplified by a factor of 1.4 on average，which is in contrast to the response where a floating grid was used. The unimproved soft clay below the floating grid-improved ground significantly affects the local amplification effect and limits the surface PGA especially when the clay is softening. In this case，a reduction in high frequency content were observed. The settlements of grid walls in all cases typically occur mostly due to the post shaking reconsolidation. Differential settlements were observed between the grid walls and the enclosed soils in both  embedded grid-improved ground sites，indicating that the negative skin friction was generated. Conversely，the settlements of the grid walls and the enclosed soils in the floating grid-improved ground are in agreement. The results of back analysis indicate that the grid walls by SSI(Soil-Structure Interaction) restrict or mobilize the shear strain response of their enclosed soils dependent on the frequency content and intensity of seismic motions. The effect of mobilizing gave a reasonable explanation for the settlements of the enclosed soils in the two embedded grid-improved grounds during the high frequency earthquakes.

The large-scale triaxial apparatus GDS was upgraded with an unsaturated testing module to study the long-term dynamic characteristics of unsaturated road base and subbase materials. The testing materials were taken from a common quarry in Zhejiang Province and mixed with kaolin at a mass ratio 3% to simulate the actual road base and subbase materials. During the tests，the axis-translation method was used to control the matric suction. A series of large-scale cyclic triaxial tests were conducted on the crushed tuff aggregate mixtures under four initial matric suctions and two cyclic stress amplitudes. Special attention was paid to the effects of the initial matric suction on the accumulated axial strain and resilient modulus. The testing results indicate that the soil-water characteristic curve(SWCC) of the unsaturated materials presents a hysteresis phenomenon in the water retention tests. For the same matric suction，the moisture content in the drying path is higher than that in the wetting path. Under cyclic loadings，the resilient modulus increases with the increasing of matric suction. The influence of initial matric suction on resilient modulus is more obvious at smaller matric suctions(0–30 kPa) than at larger matric suctions. The accumulated axial strain decreases with the increasing of matric suction，and the decreasing rate also declines with the increasing of matric suction. A nonlinear relationship exists between the accumulated axial strain and matric suction，and a linear relationship exists between the accumulated axial strain and moisture content.

The boundary of the model box in shaking table test directly affects the reliability of the test data. Lining the flexible material in the side wall of model box has become an important method to mitigate the boundary effect. A model in terms of concentrated mass to represent the soil-box-flexible material was established based on the principle of mode superposition. A theoretical formula for determining the rational parameters of flexible material parameters for model box was deduced. The influence of elastic modulus，thickness，Poisson?s ratio，density and damping ratio of flexible material was analyzed. When the input wave and the parameters of soil are determined，the parameters of the flexible material are governed by an indefinite equation. With the designated input wave and soil parameters，the error of the optimal elastic modulus of flexible materials is only 3.3% by using the theoretical formula derived in this paper and the numerical calculation. The maximum error of the soil displacement in the free field and in the model box is only 3.7%，showing that the flexible material can significantly weaken the boundary effect of the rigid model box.