The Research And Applications About Two-dimensional Resistivity Imaging Survey Inversion Theory On BP Artificial Neural Network

A large number of small and medium-sized earth-rock dams were built since theestablishment of our country, after many years after the operation, many of themappeared different degree of diseases, for example leakage behaved most outstanding.two-dimensional resistivity imaging survey is a kind of effective means that candetect the earth dam leakage. Through inverse the measured potential conductivity,leakage was identified from the conductivity of the abnormal. At present thetwo-dimensional resistivity imaging survey in data inversion was mostly based on thedamping least square method, and the inversion precision is not high, the actual useeffect is not good. This paper on how to improve two-dimensional resistivity imagingsurvey to the research on the inversion precision.BP artificial neural network is a kind of error back propagation algorithmaccording to the training of the multilayer network, which can learn and store a lot ofnonlinear input-output model mapping relation, its learning rule was the steepestdescent method, through the back propagation to constantly adjust the networkweights and threshold, make the error smaller squares of the network. BP artificialneural network was used into high density resistivity method to study the backanalysis in this paper. the network layer and parameter was selected through theneural network design, through the sample design and finite element analysis to set upnetwork training input and output sample, and then train the BP artificial neuralnetwork. Putting the experimental results into the trained neural network, can getearth dam filling the conductivity of the material, and eventually find leak.In this paper an earth-rock dam was took as the research object, thetwo-dimensional resistivity imaging survey was carried out to measure, the BPartificial neural network nonlinear inversion method was applied for inversionanalysis, and the result was compared with the results that use conventional inversionmethod. The analysis results show that the BP artificial neural network based on theinversion results its accuracy was higher and speed is faster than the conventional inversion. The inversion results was compared by different two-dimensionalresistivity imaging survey devices in this pape too, and the advantages anddisadvantages of different measurement device and applicability was summed up, andcan be used for similar engineering leakage detection for hidden trouble when choosemeasuring device as the basis and the reference. The research results enriched andperfected the high density resistivity method to detection technology, promoted thetwo-dimensional resistivity imaging survey in the application of earth dam leakagedetection, it has certain theoretical significance and practical value.