Analysis Of Identification Methods Of Ground Fissures Based On Marker Layer Difference

Xi'an ground fissure is one of the most famous geological disasters in the world,which are widely known for their wide distribution,intense activity and serious disasters.When the ground fissures are identified according to the drilling data,the loess beam-depression transition zone has a certain inclined background,which affects the identification of fault of the marker layer.Tilt to rule out the formation cause and the influence of interference,discriminant mark layer fault fracture were more objective and reasonable,enhance the accuracy of determination,reduce the error,metro line 1 ~ 5 were analyzed and summarized in this paper to crack investigation data of ancient soil layer in the survey site profile,sort out mark layers as bottom elevation difference ?h as well as the corresponding hole spacing l statistical results,further calculation analysis,based on the strata breaks with the wrong breaks the existence of differences,analysis of mark layer fault identification method of the ground fissures.The main research results are as follows:(1)According to the elevation data of the marker bottom layers in each borehole,the left and right elevation difference functions of the marker bottom layers are established,and their characteristics at the fault location of the marker layer are as follows: the fault location of the marker layer is between the maximum value of the left and right elevation difference function.(2)Statistical analysis was made on the bottom elevation difference and inter-hole dip of the adjacent marker strata in different geomorphic units,the empirical probability distribution was calculated,and the dip background value of the ancient soil layer in different geomorphic units was obtained: the dip background value of the gentle geomorphic strata was 0.099,and the dip background value of the loess ridge-depression transition zone was 0.301.(3)Calculated and analyzed the difference of the dip between the holes in the marker layer,and obtained the boundary experience curve of the difference of the dip between the holes in a single section of the ancient soil layer fault,which was applied to the analysis of the survey results of a site in xi 'an,and the discriminant results were consistent with the actual situation.(4)Four factors,including hole spacing,elevation difference function value,inter-hole plane slope value and inter-hole plane slope value,were selected as the indicators for the identification of fault faults.In view of the characteristics of ancient soil layers of different geomorphic types,four machine learning classifier models were established by using logical regression method and Kernel-SVM method.(5)In the four machine learning classification models established,the accuracy and AUC of the two logistic regression algorithms are higher than the corresponding Kernel-SVM classifier model,respectively.In general,the accuracy of the logistic regression classifier model is higher than that of the Kernel-SVM classifier model.(6)Longitudinal comparison of two classifier models based on logistic regression algorithm and two Kernel-SVM classifier models shows that the accuracy and AUC values of the two classifier models of gentle strata are higher than those of the inclined strata classifier model of loess ridge and depression transition zone,and it is easier to distinguish the fault of the marker strata of gentle strata.(7)The trained two types of machine learning classifier models are applied to engineering examples,and the results show that the predictive results of the logistic regression classifier are more accurate and the performance is better.