Research On Inversion Of High Density Resistivity Method In Backpropagation Neural Network

In order to solve the practical problems such as energy exploration,environmental geological investigation,engineering geological exploration and water conservancy engineering,the high density resistivity method commonly used in electric method exploration is widely used because of its characteristics of high collection efficiency,large data information,stable and reliable data.However,with the further development of the high-density resistivity method,the actual exploration requires the sectional inversion accuracy of the visual resistance.BP neural network,with a series of advantages such as unique learning memory and nonlinear approximation,can solve a series of problems such as local optimal solutions and dependent model selection in the conventional high density resistivity method inversion,thus improving the accuracy of the visual resistivity inversion section and starting to play an increasingly important role in the field of electric method exploration.This paper first studies the basic principle and realization of high density resistivity and the mechanism of BP neural network.Then,the inversion code of high density resistivity method of BP neural network is completed using the Python program language.Secondly,we explore the influence of different grid parameters on the computing performance of the neural network,and establish the corresponding data training sample set to complete the learning,training,and prediction and calculation of the neural network.Finally,BP neural network is used to process high-density resistivity data of measured landfill field and analyze it.In the theoretical analysis of the paper,using the traditional high density resistivity Res2 dinv software and BP neural network,the neural networks have more accurate the ability to reflect morphological boundaries for anomalies,especially the anomalies close to each other in the horizontal or vertical direction.Moreover,neural networks also have better stability in stochastic noise experiments.In the actual production research stage,the BP neural network inversion method can provide more intuitive and accurate information on the main garbage distribution and leakage pollution in the landfill area.The BP neural network method can improve the accuracy of high density resistivity section inversion,which is important to improve the success rate of high density resistivity exploration.