Inversion of Vertical Electrical Sounding Data Based on PSO-BP Neural Network

To address the issues of traditional linear inversion methods, such as their dependence on initial models and the high computational cost of Jacobian matrix calculations, this study conducts inversion research on vertical electrical sounding data based on the backpropagation (BP) neural network combined with the Particle Swarm Optimization (PSO) algorithm. First, two-layer and three-layer horizontally layered geoelectric models were constructed to generate the sample data required for neural network training. Secondly, the PSO-BP neural network model was employed to perform test inversions. The inversion results demonstrate that both neural network methods can successfully invert apparent resistivity data into corresponding geoelectric model parameters, thereby validating the correctness of the PSO-BP neural network inversion approach. Finally, the PSO-BP neural network method was applied to training and inversion of field-measured apparent resistivity data. A comparison between the inversion results of the PSO-BP neural network and those of the conventional BP neural network revealed that the PSO-BP neural network yields superior inversion results. This further confirms the reliability, effectiveness, and practical applicability of the proposed inversion method. The work presented in this study provides a novel approach and perspective for the inversion of vertical electrical sounding data.