Application Research On Leakage Hazard Detection Of Earthrock Dam Based On High-Density Resistivity Method

Dam is an important system of flood control.References indicate that hidden problems accumulated over time would result in significantly severe accidents if losing attentions on them.Thus,it is increasingly important to precisely detect and recognize hidden problems of dam.Currently,high-density electrical method is one of the most common methods,of which the detection precision is closely relative to parameter setup,the result approximation is poor due to the high dependence on artificial experiences,and the detection efficiency is low.Therefore,based on the finite difference method,in this study we used the dipole-dipole method,Winner method,and Schlumberger method,respectively,to simulate the geoelectric leakage model of rock filled dam,which showed positive results.Furthermore,the precision of highdensity electrical method was effectively improved by the comparison and analysis of modeling tests and simulations,and the intelligent recognition of tested apparent resistivity maps through machine learning.The improvement was verified via engineering projects.The conclusions in this study are as follows:(1)In the simulations of high-density electrical method,the finite difference method can be applied in multiple detection methods well,of which the results were satisfactory in different practical scenarios.While fixing the length of measurement line,the shorter the distance between electrodes and the more the quantities of electrodes,the richer and more precise the geological data detected via of high-density electrical method.(2)Through modeling tests of hidden problems in rock filled dam,we found that all of the dipole-dipole method,Winner method,and Schlumberger method can detect the existence of leakage problems in the detection processes of different scales and locations.Compared to Winner method and Schlumberger method,the dipole-dipole method was better at detecting the existence and locations of hidden problems.However,all of the three methods failed to deliver reliable data with respect to leakage scales;In terms of the detections of hidden problem locations,the effect of the middle position in measurement line is better than that of the side position;among the three methods,the dipole-dipole method is better at detecting low-resistance media,while Schlumberger method is better at detecting high-resistance media.From a macro perspective in terms of detection effects,the preferred priority in the detection of hidden problems of dam leakage is: the dipole-dipole > Schlumberger > Winner.(3)Based on machine learning and Faster R-CNN model,we established a learning model of the intelligent leakage recognition by high-density electrical method.Through the study and training of leakage characterizations in the established leakage problems database of high-density electrical method,we found that the model established in the intelligent recognition of hidden leakage problems of high-density electrical method can significantly improve the precision and efficiency of detecting hidden leakage problems,which makes the discrimination results more precise and objective.(4)In engineering projects,by comparing the results of simulations and tests,applying the method of elongating the measurement line can effectively avert the drawback of the high-density electrical method for its poor detection on the side positions of the measurement line.Meanwhile,the three aforementioned detection methods had been verified effectively: the dipole-dipole method can be used for detecting hidden problems of dams at any time;and Schlumberger method is better used for the detection of high-polymerization grouting effect due to its satisfactory detection on high-resistance media.