Research On Stability Of Water-sealed Cavern Shaft Based On Intelligent Identificarion Of Rock And Soil Mass

With the continuous advancement of industrialization,energy security determines national security.It is urgent for our country(China)to build more energy reserve bases to cope with various possible energy crises.In this paper,an Underground Water-sealed caverns as the research object.Firstly,a large number of field measured data and image data are obtained,and the shortcomings of Q value method are analyzed.Then the Q value method is optimized with image processing technology.Finally,the optimized method combined with the limit analysis theory was used to analyze the shaft stability,and the conclusions are as follows:(1)Through analysis and research,it is found that the grading results of Q value method fluctuate greatly due to subjective factors.Therefore,multiple classifiers are built by the convolutional neural network(Efficent Net)to optimize the surrounding rock classification method.Finally,the accuracy of surrounding rock classification model training is 88.6%,joint development degree model training is 86%,joint surface strength model training is 78.8%(The feature recognition is relatively complex,with less image data and low accuracy),occurrence condition model training accuracy is92.1%.(2)Aiming at the problem of large operation space of classification parameters in Q value method,this paper uses image processing technology to successfully change the traditional manual acquisition method to mechanical acquisition of borehole core and surrounding rock joints,which improves the acquisition rate and accuracy of joints.The process of borehole core image processing is divided into pre-processing,pre-processing and post-processing,and finally the rock mass integrity evaluation index Jn is obtained.The image processing of surrounding rock consists of three modules:Firstly,Holistically-Nested Edge Detection,image threshold segmentation and morphological processing to initially extract joint information.Secondly,the outline and skeleton were extracted from the preliminary information by mathematical morphology,and then the joint trace was detected by Hough transform.Finally,the detected joint trace was further optimized by clustering,and joint dip and inclination were extracted.And the evaluation indexes Jn and RQD of rock mass integrity were obtained.(3)Through the analysis of the stress environment of the shaft,it is shown that the shaft bears different tensile and compressive stress combinations.The strength failure criterion of concrete under multi-axial stress is introduced,and the ultimate bearing capacity of the shaft considering the bidirectional stress state is proposed.In addition,taking the rock-like material concrete as the object,the ideal elastic-plastic limit analysis method is adopted,and the reference value of elastic-plastic limit strain of shaft wall is derived by introducing the shear strength of concrete and the compressive strength of shaft,and the theory of limit strain of shaft wall is derived.Finally,the theory of limit strain analysis of shaft wall is improved by numerical simulation.(4)The optimal surrounding rock classification method is applied to the stability analysis of shaft surrounding rock,and the engineering model of the shaft is established based on the classification results.Then the limit strain theory is introduced to determine the shaft stability.Finally,it is found that safety protection measures should be taken for the location of the well head to avoid large deformation.In addition,the junction between the shaft and the cavern is a weak area,which needs to be further strengthened.