| Hunan Province’s regional geological environment conditions are complex.It is located in the Yunnan-Guizhou Plateau to the hills of the south and the Nanling Mountains to the transition zone of the Jianghan Plain.The terrain is surrounded by mountains on three sides,the horseshoe-shaped landscape towards the north opening to the mountains,hilly terrain,relatively abundant precipitation,short periods of heavy rainfall,and the intensification of human engineering activities in recent years,landslide geological hazards more and more frequent.Taking Zhuzhou 1#landslide disaster site as the research object,geophysical exploration techniques and geological engineering drilling were used to identify the engineering geological characteristics of the disaster site in detail,and numerical analysis models were established to comprehensively analyze the stability of the disaster site under natural working conditions and storm conditions.At the same time,by deploying monitoring equipment on the landslide body,monitoring data of the disaster site was collected and organized,and combined with machine learning,a displacement prediction model of the disaster site was established,and the displacement of the disaster site under heavy rainfall conditions was predicted.The stability of the disaster site was analyzed by combining the spatial and temporal evolution of landslide deformation.On this basis,the anti-slip pile reinforcement measures are proposed considering the analysis results of the stability of the disaster site,the macroscopic deformation characteristics,and the hazard objects.An optimization model was established for the problems in the design of anti-slip piles.The optimal design parameters of anti-slip piles were solved by combining FLAC3D and the improved simulated annealing（SA）optimization algorithm.The anti-slip piles at the disaster site were reinforced according to the optimization results.The results showed that the deformation of the landslide disaster site was effectively controlled after reinforcement.The research idea of this paper covers the whole process of landslide hazard point detection,Stability analysis,and Optimization design of support structure,which provides new ideas and methods for landslide geological hazard prevention and management.The main research contents and the research results obtained are as follows:（1）The relationship between resistivity and physical and mechanical properties of clayey soil:for the water content of the soil is the main factor affecting the physical parameters of clayey soil,the appropriate resistivity testing equipment was used to carry out indoor tests on the correlation between the physical parameters of clayey soil and its physical and mechanical parameters,the results show that as the water content of clayey soil increases,the resistivity value of clayey soil gradually decreases and finally tends to converge to the law of change,on this basis,the empirical relationship between the physical parameters of clayey soil and the indicators of physical and mechanical parameters was established,thus laying the foundation for the high-density electrical method in the hilly mountainous area mounded soil landslide detection.（2）High-density electrical method field landslide detection test:Based on the research results of the indoor test and on the basis of fully grasping the on-site investigation of the landslide potential disaster site,the high-density electrical method was used to conduct the field detection test of the disaster site,and the corresponding resistivity values of different strata at the disaster site were obtained,and the spatial distribution state of the strata and the distribution position of the sliding surface at the disaster site were identified.（3）Numerical analysis of landslide stability integrating geophysical exploration results:Based on the study of geophysical characteristics of the geotechnical body,the correlation between mechanical engineering parameters of the geotechnical body and geophysical parameters is established by combining the evolution mechanism of slope instability.On this basis,a three-dimensional slope geological model was based on combining the drilling and physical exploration results of the slope.The three-dimensional slope stability analysis was carried out by using FLAC3Dnumerical analysis software to obtain the deformation information of the disaster site.On this basis,the most prominent deformation profile of the disaster site was selected to establish further a two-dimensional model to explore the space-time evolution of the stress field,seepage field,and displacement field of the disaster site under the rainfall-induced factors,compare the space-time differences in the development of different rainfall intensities and slopes,and summarize the space-time evolution of the disaster site under the coupling effect of multiple fields.The risk of destabilization of the disaster site under rainfall conditions is evaluated based on the magnitude of the safety factor.（4）Research on landslide displacement based on machine learning:The data collected by each monitoring sensor on the hidden disaster site were collected and sorted out,and the main causal factors of the disaster site deformation were brushed out by combining with the landslide deformation mechanism,and the wavelet principal component analysis algorithm（WPCA）was used to denoise the monitoring data.Based on the time-series characteristics of each monitoring data and the characteristics of the LSTM algorithm,the WPCA-LSTM displacement prediction model was established.Considering the influence of previous rainfall on the disaster site deformation,the model was trained with different days of monitoring data.The results showed that the model trained with five days of monitoring data had a minor error value and the highest accuracy.The established model is used to predict the displacement value of the disaster site under heavy rainfall conditions and combined with the process of landslide deformation evolution.The disaster site is evaluated to have a significant risk of destabilization and damage under severe rainfall conditions.（5）Optimization study of the treatment plan of the hidden disaster site:Based on the numerical stability analysis and WPCA-LSTM displacement prediction results,the risk of destabilization and damage of the disaster site under heavy rainfall working conditions is comprehensively evaluated.Based on this,the support scheme of an anti-slip pile is proposed considering the deformation characteristics of the disaster site and its threat objects.For the problem of subjective arbitrariness in the traditional anti-slip pile design process,the optimization model of the anti-slip pile is established by a combination algorithm,and the optimal design parameters of the anti-slip pile are obtained.At the same time,the design of the support structure was carried out according to the principle of limit equilibrium,comparing the economy and safety of the support structure designed by the two concepts.The construction drawing design was carried out with the optimal solution as the guide.The effectiveness of the"SA+FLAC3D"combined algorithm in the optimization design of anti-slip piles was powerfully demonstrated by observing the deformation monitoring data after the treatment of the disaster site. |
