Characteristic Analysis and Fast Inversion of Array Lateral-Logging Responses

摘要

The characteristic between responses of different models and parameters are analyzed, which is crucial to the inversion problem of logging and useful for logging engineers. The inversion of the multi-parameters of the array laterallogging is a nonlinear and multi-solution problem. To avoid the defects of the traditional inversion algorithms and enhance the inversion speed, the three-layer back propagation (BP) neural network is built to inverse formation parameters. The inputs of the BP neural networks are the logging responses and the target outputs are the formation parameters. The training results of different networks are given and compared, and the inversion results calculated with the trained nets are close to that of company, it only takes 3 s to inverse 100 groups of parameters.