The unloading of deep rock mass shows significant three-dimensional and nonlinear mechanical characteristics，and the design analysis method and engineering experience based on shallow tunnels cannot be directly transferred to deep engineering. The traditional calculation model does not reflect the mechanical characteristics of the deep rock mass. The generalized rock mass parameters are mainly determined by the displacement back analysis，and there is still a lack of design analysis theory and method for real-time and accurate acquisition of rock mass parameters and dynamic diagnosis of deep tunnels. The theoretical basis，in-situ parameter acquisition and applicability of GZZ rock mass strength based 3D continuous analysis method for deep rock mass are reviewed and studied. The non-associated plastic flow rule and elastoplastic constitutive model，considering 3D strength and dilatancy effect of deep rock mass，have been verified by true triaxial rock experiments，physical model tunnels，and in-situ tunnel monitoring data. The digital in-situ testing technology is promising to the obtain mechanical parameters of rock mass and thus to realize the 3D positive analysis and real-time design of deep tunneling，which overcomes the limitations of traditional displacement-based back analysis approach. It reveals the 3D extrusion law of the deep tunnel face and the mechanical mechanism of the principal stress rotation，and expounds the mechanical influence mechanism of the intermediate principal stress and the 3D stress state on the stability of the deep tunnel. The significance of the stress control-based design method and analysis idea in the deep tunneling precise analysis and control is discussed and investigated，which provides theoretical basis and technical support for intelligent construction and real-time design of deep and ultra-deep tunnels.

Red beds refer to the general name of red rock series deposited in various geological historical periods，which are a group of typical “slippery stratum”. Slopes in red bed areas are prone to form group-occurring geohazards in the process of heavy rainfall. By summarizing the development，distribution and main characteristics of red beds，it is found that the properties of red beds are controlled by sedimentary construction. The physical and mechanical properties and hydrologic properties of red beds are different due to the sedimentary environment. Red bed rocks have strong rheology and hydrophilicity，and are easy to swell，disintegrate，muddy and soften when exposed to water. The main types of geohazards in red bed areas are analyzed and summarized，and the formation mechanism of sub-horizontal translantional landslides and gentle shallow soil landslides in sand-mudstone interbedded areas is mainly introduced. The results show that the rainwater quickly infiltrates into the vertical cracks in sandstone and forms hydrostatic pressure，which often becomes the main driving force of sub-horizontal translational landslides. The long-term argillization and saturation softening of mudstone in sand-mudstone contact interface or weak interlayer by groundwater greatly reduces the strength of the bottom slip surface，which is also the main cause of large sub-horizontal translational landslides. Of course，the special slope hydrogeological conditions formed by the special lithologic combination of sand-mudstone interbedded are also the important reasons for landslides. The atmospheric influence depth in the slope residual layer and the interface effect in the base layer are the main reasons for the occurrence of shallow soil landslides during heavy rainfall，which determines the thickness of the landslide body. Finally，the risk prevention and control measures of red beds geohazards are discussed，and it is considered that establishing physical warning models through mathematical and mechanical analysis combined with field observation and laboratory tests is an effective way of red beds landslides warning. The landslides in red bed area should be controlled mainly by drainage and auxiliary by anti-slide measures.

Based on the analysis of failure characteristics of multiple underground engineerings under high in-situ stress condition，the mechanical behavior of rock before and after peak is incorporated into the support design of underground engineering. This paper establishes the stress-strength design method for huge underground engineering considering initial strength index and post-peak brittle-ductile transformation factor. The established support design method pays full attention to the rock crack propagation mechanism and post-peak mechanical behaviour. The initial strength index(ISI) is defined by the ratio of the crack initial strength of rock under uniaxial compression to the maximum principal stress of geostress，which reflects the crack propagation mechanism of rock and the initial stress level of surrounding rock. The post-peak brittle-ductile transformation factor (BDF) is defined by the decreasing rate of rock brittleness within engineering confining pressure，which demonstrates the magnitude and rate of post-peak stress reduction and rock brittleness. The support strength is obtained according to the basic principle of convergence-confinement method. A seven-zone evaluation system of surrounding rock failure mode and support design is established based on the statistical results of ISI，BDF and support strength of surrounding rock of several typical huge underground engineerings. The failure modes and support parameters of huge underground engineerings in Baihetan，Jinping II，Jiangbian，Yangfanggou hydropower station and Yixing pumped storage power station show the rationality and applicability of the established support design method.

The twin shear unified bounding surface plasticity model(BUST model) can capture the plastic strain accumulation of geomaterials in small deviatoric stress state，which can?t be well predicted by classical twin shear unified elastoplastic constitutive model(UST model). To apply the BUST model in analyzing static and dynamic response of slopes or landslides，its finite difference format is formulated and compiled，and its reliability is further analyzed using numerical simulation. Firstly，the finite difference format of the BUST model is formulated both in general stress space and principal stress space. Specifically，the mathematical form of bounding surface，loading surface，mapping rule variable，plastic multiplier，plastic correction，and corner singularity treatment are derived. Secondly，the finite difference format of BUST model is compiled using C++ language，and its dynamic link library file is generated. Finally，a list of numerical simulation is conducted using FLAC3D software including a unit test for comparing BUST model and UST model，a triaxial compression test for predicting the stress-strain relationship of soft rock，a loess-mudstone slope simulation for predicting acceleration response. The results indicate that the finite difference method of the BUST model is reliable，and it can be applied in analyzing static and dynamic response of slopes or landslides.

In order to identify the key joints that have significant influence on the mechanical properties of rock mass from the complex random joint network，and to realize reasonable simplification of the joint network. The characteristics of the random joint network were studied by taking the dip angle as a variable. A random joint network was constructed based on Monte Carlo theory，and 7 kinds of joint specimens were prepared by 3D printing technology after eliminating a certain inclination angle. The uniaxial compression test was carried out. The variation coefficient method was used to improve the mechanical equivalent analysis method，and the dip angle characteristics of key joints were analyzed comprehensively based on the strength，deformation and energy characteristics of specimens. The results show that：(1) with increasing the joint removal angle，the peak stress and elastic modulus increase and then decrease，but the peak strain decreases and then increases. (2) The prepeak elastic energy density of the specimens increases with the joint removal angle，and the dissipation energy density shows a decrease-stabilize-increase trend. (3) The weights of 7 mechanical parameters are determined based on coefficient of variation method. It is considered that every dip angles has significant influence on specimens by comprehensive analysis of mechanical properties. The 90° and 45° dip angles have the most significant effect，while the 15° dip angle has minimal effect.

In order to explore the difference of dynamic response for different composite rock masses under impact loading in engineering，dynamic impact tests under different strain rates are carried out with split Hopkinson pressure bar(SHPB) test system and digital image correlation(DIC) on coal-rock and rock-coal composite rock masses. The properties of two composite rock masses after impact loading are compared including stress wave propagation，dynamic strength，energy dissipation，fracture fractal characteristics and failure modes. Furthermore，a constitutive model is constructed considering the characteristics of composite coal and rock. The results of experiment and theoretical analysis show that incident energies of two kinds of composite rock masses are almost the same under the same strain rate，dynamic strength properties and dissipation of rock-coal can be significantly greater than that of coal-rock especially when the strain rate is small，this is because wave impedance matching of rock-coal is better in the process of stress wave propagation，and the difference decreases with the increase of strain rate. The larger the dissipative energy of composite rock mass is，the larger the fractal dimension of fragmentation is，and there is an obvious linear correlation between dissipative energy and fractal dimension. Splitting failure is dominant on the coal side of different composite rock masses under impact loading，and then ultimate macroscopical failure will occur due to influence of interface effect with the development of the cracks on the coal side. A constitutive equation considering strain rate effect and damage evolution is established based on the element combination，it is in good agreement with the experimental results，which verifies the correctness of the constructed model.

Strainburst is a rockburst phenomenon induced by the redistribution of surrounding rock stress in the process of roadway excavation，which poses a major threat to the safety of personnel and equipment in mining and tunnel construction. In this paper，the self-developed strainburst experiment system is used to carry out strainburst experiments of intact sandstone samples and samples with different numbers of boreholes. High-speed photography is used to collect images of the whole process of rockburst，and monitoring methods such as acoustic emission and digital speckle are used to collect relevant data. Based on the data，the failure characteristics of the sandstone strainburst under the conditions of different drilling numbers are studied，and the influence of the number of drilling holes on the strainburst is discussed under. The results show that with increasing the number of boreholes，the intensity of rockburst decreases，the accumulative count and accumulative energy value of acoustic emission decrease，and the activity degree of acoustic emission decreases，indicating that the micro-fracture in the rock body decreases. When the kinetic energy decreases，the rockburst risk level decreases from a strong rockburst，and the fragmentation degree of the ejected debris decreases. When the rockburst occurs，the vertical and horizontal maximum displacements of the rock sample increase，and the deformation capacity increases. The experiments show that the measures of borehole pressure relief have a good prevention and control effect on rockburst disasters.

In view of the size effect of internal crack propagation and penetration when rocks of different sizes containing multiple flaws are loaded，sandstone disk samples with diameters of 50，100，150 and 200 mm containing parallel double flaws were made. Their thickness，flaw length and flaw spacing change in proportion to the diameter. Static loading tests were carried out on disc samples on GCTS RTX–3000 testing machine，and the crack propagation process of the samples was monitored by comprehensively using acoustic emission(AE) and digital image correlation(DIC) technology. The results show that with the increase of the sample size，the failure load increases linearly，the weakening effect of prefabricated flaws on the bearing capacity of the samples is enhanced，and the energy consumed per unit volume of sample failure decreases gradually. There is a competitive relationship between multiple crack propagation. The crack initiation sequence determines the propagation direction. The crack propagation of samples with diameters of 50，100 and 200 mm has the same direction. Cracks expand in opposite directions of samples with diameter of 150 mm. AE can accurately reflect the damage evolution process inside the sample. Using DIC technology to digitize the crack propagation process，the crack opening displacement and propagation speed can be accurately analyzed. The variation law of crack propagation speed of samples with different sizes is similar，that is，the crack starts to grow at a low initial speed and reaches a peak speed within a short distance，then the propagation speed drops sharply，and the final crack propagation speed decreases slowly and tends to be stable.

In this study，on the basis of the analysis of the source characteristics of the cylindrical charge，the generation mechanism and development process of the R-wave in rock drilling and blasting was unmasked by interpreting the reflection of curvellient body waves at the free surface. Then，to further investigate the growing process and evolution characteristics of the R-wave，the wave-type and wave-components induced by the cylindrical charge in the infinite and semi-infinite spaces were compared using the 3D dynamic finite simulation. Moreover，the onsite blasting experiment was also conducted to verify the growing characteristics of the R-wave. The results indicate that the R-wave is generated with the reflection of the explosive-source-emitted curvellient body waves at the ground surface，and it grows gradually with the epicenter distance r. The R-wave starts forming at about 0.6h (where is h the buried depth of the explosive source) far from the epicenter，and some gentle R-wave motions could be recognized when r exceed 2.0h. Whereas，the R-wave is still not fully developed and cannot be easily recognize until the incident angle of the curvellient body waves exceed a critical value，where r is larger than 5.0h. In the rock drilling and blasting，the source emitted P-wave?s motion approaches to the horizontal direction with r increasing，and its contribution mainly awards to the horizontal vibration. The source emitted S-wave component could be ignored and is easily covered by the fully developed R-wave，as the S-wave gradually deviates from its dominant radiation orientation. What?s more，the R-wave will dominate the vertical ground vibration when r exceeds 25.0h～35.0h.

In order to improve the stability of mined-out areas after filling in a metal mine，the damage law of cemented tailings backfill with different contents(0%，0.5% and 0.7%) of glass fibers was studied. The stress-strain curves of cemented tailings backfill with different contents of glass fiber were obtained by uniaxial compression. The energy dissipation law of backfill was analyzed according to the damage constitutive equation，and its match with rock mass was obtained. The results show that：(1) in a certain range，the strength of backfill is improved with the addition of glass fiber but the elastic modulus is reduced. (2) The damage value of backfill when reaching the peak stress is reduced with adding glass fiber，but with the increase of its contents，the process of damage value reaching the maximum of backfill is slower. (3) The calculation model of peak specific energy of backfill with different contents of glass fiber is obtained，and the peak specific energy of backfill increases with the increase of contents of glass fiber. (4) Based on the filling mining of the mine，the design model of the depth of the goaf and the strength of the backfill is obtained，which can provide a scientific basis for saving the backfill cost of the mine.

In this study，drying experiments at different temperatures were conducted to study the variations in the quality，fracture，pore，microstructure and expansion of upper-Shaximiao Formation red mudstone in eastern Sichuan Province. Field emission scanning electron microscope(SEM)，N2 adsorption measurement and digital radiography(DR) were used to evolution micro-structure，meso-pore and macro-fracture of rock samples during heat treatment，aiming to obtain the mass loss and expansion rules of rock samples reveal the influence of heat treatment on the expansion mechanism of red mudstone. The results show that：(1) the mass loss of rock samples presents two patterns as rapid water loss-slow water loss and slow water loss with the increase of heat treatment temperature. The complete water loss of the red mudstone requires a heat treatment temperature greater than 130 ℃. (2) Clay minerals exhibit interlayer opening and cracks along layers after high-temperature heat treatment，while they mainly show intralayer cracks after low-temperature heat treatment. The contribution of heat treatment temperature to macro cracks after heat treatment is low less than primary fissure. (3) The fine particles (0.5～2nm) in rock pores increase significantly with the decrease of heat treatment temperature，and the number of these particles reaches the peak at 110 ℃. Heat treatment temperature has an obvious effect on the micro-pore sizes of rocks. Compared with the single peak distribution of pores in natural state，the pore distribution after the high temperature heat treatment presents a bimodal type. (4) The expansion pattern under the low temperature treatment is mainly controlled by macro-fractures，while the expansion pattern under high temperature treatment is dominantly controlled by the macro-fractures and micro-pores. In addition，increasing the heat treatment temperature can significantly increase the curve slope and final expansion in the rapid expansion stage. The results reveal the characteristics of micro-structures and micro-pores and the development of macro-fractures under different heat treatment conditions and discuss the influence of heat treatment temperature on the expansion processes of upper-Shaximiao Formation red mudstone，which may provide a new insight into predicting the drying degree of mudstone and provide a reference to the correction of final expansion rate under specific heat treatment conditions.

Microwave-assisted rock breaking is a new method to achieve efficient and low-damage excavation of hard rocks. There are some differences in the wave absorption capacity，water loss，warming and damage laws of different hard rocks，and the comparative study has important theoretical and engineering significance for accurate and efficient rock breaking. By comparing the mass changes of the specimens before and after pretreatment and microwave radiation，the water loss law of the three types hard rocks were obtained. The whole process of surface temperature rise and fracture of hard rock in microwave field is monitored by infrared thermal imager，and the temperature rise and macro crack propagation characteristics of three types of hard rock under different water content were obtained. The mechanism of hard rock damage effect with increasing radiation energy was elaborated by the changes of longitudinal wave velocity，acoustic emission and SEM microstructure of three types hard rocks under different time of radiation in dry condition. The results show that the effect of water in the lower power microwave radiation within 300 s is as follows：the evaporation of water is the dominant factor in the mass loss of hard rock，water has a significant role in promoting the warming of hard rock，and the degree of influence of water content on the average temperature of the surface area is divided into three stages according to the radiation time “small(0‐50 s)，large(50‐200 s)，stable(200‐300 s)”，in which the saturated water absorption has the most significant effect on the temperature rise and the sandstone and limestone with high saturated water absorption burst in the cavity at 55 s and 240 s respectively. The damage law of different hard rocks with temperature rise in dry state is as follows：the crack propagation of limestone is more obvious than that of sandstone and granite due to large surface temperature difference. COMSOL numerical simulation of continuous radiation for 420，600 and 720 s has a good correlation with the change trend of the damage factor obtained from the experiment. After the three types of hard rock damage，the acoustic emission of the specimens presents three stages of stability，growth and failure，and the uniaxial compressive strength loss is from fast to stable，and the ring count and energy are different in order of magnitude；At the micro scale，the fracture surface of granite and sandstone after 720 s of microwave radiation appears the bonding of molten minerals，which leads to the slow reduction of the longitudinal wave velocity，which explains the internal cause of the damage stabilization；With the increase of the radiation time，the mass loss，the damage factor characterized by the wave velocity，the heating rate and the strength loss obtained by the numerical simulation all show a consistent change law from fast to slow and a “inflection point” from fast to stable. The research results can provide some reference for the differentiation，low energy consumption and high efficiency damage of different hard rocks.

Based on the rigid-body-spring method，a new three-dimensional discrete numerical model is proposed to simulate the progressive failure process of brittle rocks. The model describes the interaction between rigid blocks by means of spring-set groups distributed on the interface of blocks. Thus，the progressive crack propagation is naturally considered. To obtain a basic mesh with high homogeneous and isotropic crack direction，a new random Voronoi mesh technology with a growth “nucleus” place strategy is proposed based on the “point saturation” strategy. Then，the basic functions of three-dimensional modified rigid-body-spring method are proposed. Considering mesoscopic shear and tensile failure modes，a failure criterion based on Mohr-Coulomb criterion with maximum tensile strength is established，and the calibration strategy of mesoscopic parameters of the new model is provided. Finally，the proposed model is validated by comparing with the results of laboratory triaxial tests. Effects of the pattern of the spring-sets configuration and the block size on the simulation results are also discussed. The results show that the proposed model can correctly describe non-linear stress-strain responses，evolution of micro-cracks，and macroscopic failure mode of the rocks under different confining pressures.

To study stress distribution and failure characteristics of a longwall panel floor with a negative gate pillar，coupled finite and discrete element numerical modelling method was adopted to construct a FLAC-PFC coupling numerical model. Spatiotemporal evolution of floor stress distribution，failure characteristics and effect of caved rock accumulation on floor stress and failure were analyzed throughout the process from setup room excavation to compaction of caved rock in gob to roof strata settlement. Based on elastic mechanics，a mechanical floor model was established to derive failure depth. The results show that：(1) Asymmetric accumulation of caved rocks in gob and asymmetric collapse of roof result in asymmetry of stress and failure of floor. (2) A “bottleneck” structure is formed by asymmetric failure of the floor and rotary cave-in of key blocks of roof，which hinders the further asymmetric slip of gob rocks. (3) The floor stress distribution under the influence of asymmetric accumulation of gob rocks presents a spacial distribution of “large top and small bottom”. And the stress concentration on both sides of the gob increases first and then decreases over time，and the floor stress experiences a transition of “compressive-tensile-compressive”. (4) Numerical simulation，theoretical calculation and field measurement results verify the asymmetric characteristics of stress and failure of floor caused by asymmetric accumulation of gob rocks. The study serves to provide theoretical support and scientific basis for determining location of gob side entry and its surrounding rock control of future adjacent panel with a negative gate pillar.

Liquefaction-induced pile failure during earthquake is the main reason for building and public infrastructure damage. A centrifuge shaking table test was carried out to explore the seismic response of pile group foundations in the liquefaction sites. A finite element numerical model considering nonlinear liquefaction large deformation under dynamic static coupling，for carrying out plastic response analysis of pile group foundation under earthquake，was established. The research indicates that the liquefaction of the foundation under the action of an earthquake first appears on the surface of foundation around the pile. As the PBA increases，the liquefaction range of the foundation develops toward the depth of the foundation and both sides of the pile. The position of the maximum bending moment of pile group appears at the bottom of the pile，soil surface and caps. The larger deformation of the straight pile occurs at the bottom of the pile and the surface of the foundation，and the larger deformation of the oblique group piles appears in the middle of the piles. The excess pore pressure ratio of the foundation decreases with the foundation depth，but increases with the PBA. The foundation will produce foundation uplift within 20 times the pile diameter of the pile，and foundation seismic subsidence will occur in the far pile area. The group-pile suffers less compression damage and serious tensile damage. The pile bottom produces tensile failure during 0.3 g strong earthquake condition.

In order to provide support for the safety control of shield tunnel under-crossing the existing tunnel，genetic algorithm(GA) is used to optimize the structural parameters of Bi-directional Long Short-Term Memory(Bi-LSTM)，including the time series，units in hidden layer，hidden layers，Bi-LSTM layers，and dropout. And then，the existing tunnel settlement prediction model，named GA-Bi-LSTM，is constructed by comprehensively considering engineering geological parameters，spatial parameters，and shield construction parameters. Based on the settlement monitoring values and the corresponding construction parameters of the section project of Changsha Rail Transit Line 3 crossing Changsha Rail Transit Line 1 in parallel，the model is trained and tested. The results show that the mean absolute error(MAE)，root mean square error(RMSE) and sample regression value(R2) of the GA-Bi-LSTM model are 0.42，0.45 and 0.90 respectively，and the average relative error is only 10.78%. Compared with BP，SVM，LSTM and Bi-LSTM models，GA-Bi-LSTM model has better prediction accuracy，indicating that the model has better reliability and practicability，which can provide a new idea and method for the settlement prediction of the new tunnel under-crossing the existing tunnel.

To investigate the behavior of lateral facing deformation of a geosynthetic-reinforced soil(GRS) wall in a tiered configuration and to quantify the correlation between the maximum lateral facing deformation and the factor of stability(FS) of the corresponding wall，a validated finite difference numerical method was used to calculate the lateral facing deformation and the maximum tensile force mobilized in each layer reinforcement，which were inputted in the shear strength reduction(SSR) method to determine the factor of safety of the tiered GRS walls. Further，a parametric study was performed to analyze the effect of properties of backfill and foundation soil，reinforcement properties，and configuration mode of the tiered GRS walls on the lateral facing deformation and the factor of stability. The results show that：(1) for the two-tiered GRS walls with other parameters remaining constant，an increase in friction angle or cohesion of the backfill causes the enhancement of the wall stability，and thus decreases the maximum lateral facing deformation and the maximum tensile forces mobilized in the reinforcement layers. Increasing reinforcement length in upper or lower wall results in a reduction of lateral facing deformation and an increase of FS. When the reinforcement length in upper or lower wall reach the critical value，e.g. 0.7H (H is the total wall height) or 0.6H，the maximum lateral facing deformation and FS value tend to be stable. Besides，reducing the vertical spacing of reinforcement or increasing the reinforcement stiffness causes a decrease in lateral facing deformation and an increase in FS. (2) Increasing the offset distance for a multitiered GRS wall with equally individual wall height，the maximum lateral facing deformation tends to decrease and then to be stable. The critical offset distance，beyond which each tier wall functions independently，is determined to be the 1.2 times the individual wall height for the backfill with a recommended friction angle ? = 34° by guidelines. (3) The maximum lateral facing deformation and the corresponding FS for a multitiered GRS wall with the same total wall height and normalized offset distance decrease with the number of tiers，and then tend to increase once the number of tiers reaches a certain value. When the wall height ratio of the upper and lower walls is no great than 1，the change of the wall height ratio does not affect remarkably the lateral facing deformation. In addition，as the total height of the two-tiered GRS walls increases，the maximum lateral facing deformation increases and the FS values decrease. (4) An empirical function based on the numerical results was proposed to establish the correlation among FS，the normalized maximum lateral facing deformation and the normalized offset distance. It is helpful for engineers in practice to evaluate quantitively the stability of a two-tiered GRS wall based on the specified maximum lateral facing deformation.

The fortification level is an important factor to determine the effect of liquefaction disaster prevention. However，there is still less knowledge on the difference of risk of earthquake liquefaction damages under different fortification levels，and as a result，the seismic codes of building and structure engineering，highway engineering，railway engineering，water transportation engineering in China still adopt basic seismic ground motion as the fortification index to deal with site liquefaction hazards. A simplified model is set up，and based on mature and engineering acceptable site liquefaction analysis methods，the formulas for calculating the liquefaction-induced damage risk under basic seismic ground motion(moderate earthquake)，rare seismic ground motion(rare earthquake) and extremely rare seismic ground motion(huge earthquake) are derived. The difference of liquefaction-induced damage under three fortification levels is discussed，and based on the comparative analysis of liquefaction damages caused by the 2021 Maduo M7.4 earthquake，the necessity of fortification against site liquefaction under rare earthquake is demonstrated. The results show that the liquefaction-induced damage risk varies significantly under the three different fortification levels and the liquefaction-induced damage risk increases obviously when the level of seismic ground motion increases. In terms of the engineering sites in seventh and eighth grade regions widely distributed in China，for the near liquefaction site and slight，medium liquefaction sties，under the rare and extremely rare seismic ground motions，site liquefaction index and its possibility both have a sharp rise. On the principle of the most disadvantageous，liquefaction index at least will be increased by at least one grade，most by two grades，partly by three grades，and the occurrence probability is increased by 20%–30%，reaching high and very high levels. There is a good correspondence between the seismic ground motion level of liquefaction fortification，actual seismic ground motion intensity，actual site liquefaction severity and bridge damage grade in the 2021 Maduo M7.4 earthquake，which shows the correctness of the theoretical derivation of the paper and the necessity of liquefaction fortification for the rare earthquake ground motion. In view of the consistency of seismic resilience and the goal of "not falling in a big earthquake"，the seismic ground motions of relevant engineering design in China should not only adopt the basic peak ground acceleration，but also take the rare peak ground acceleration as the fortification value of site liquefaction deformation control.