1D intelligent inversion of electromagnetic wave logging data based on convolutional neural network

With the increase of the application of highly deviated wells and horizontal wells, the accurate prediction of formation boundary and resistivity becomes particularly important, but the logging environment of horizontal wells is relatively complex, it is difficult to predict the formation information accurate. In this paper, a new method based on convolutional neural network (CNN) is proposed to solve the problem of resistivity inversion. This method establishes the neural network model, uses Adam optimization algorithm to optimize the network parameters, enhances the nonlinear approximation ability of the network by adding activation function, and uses batch normalization to speed up the rate of training. The data set is calculated by the forward model and processed for the training, verification and test of neural network. Finally, relative noise is added to the test data set, and the network is applied to the test set with noise. The results show that the model has strong robustness. In addition, compared with the commonly used nonlinear iterative method, it not only overcomes the disadvantage of its dependence on the initial value, but also greatly reduces the time of inversion calculation.