Study On Airborne Transient Electromagnetic Imaging Method Based On Wavelet Artificial Neural Network

Airborne Transient Electromagnetic Method(ATEM)refers to a geophysical detection method based on electromagnetic induction with aircraft as a vehicle.It is one of the commonly used measurement methods for airborne geophysical prospecting.This article is based on the national key research and development plan of "Research and Software Development of Data Processing Methods for Airborne Transient Electromagnetic Method and Ground Electrical Method",using a kind of artificial neural network(wavelet neural network)to study the ATEM imaging,and the the transient quasi-resistivity imaging based on wavelet neural network is realized.This paper introduces the basic principles of the airborne transient electromagnetic method and derives its one-dimensional forward modeling formula in detail,establishes a variety of typical layered geoelectric models,uses the one-dimensional forward modeling program to obtain the time-domain airborne transient electromagnetic response,formed a double-hidden-layer wavelet neural network.The trained wavelet neural network is used to image and explain the airborne transient electromagnetic response of various layered geoelectric models,and the imaging results are compared and analyzed.The imaging characteristics of the two neural networks are expected to be compared.According to the results:Compared with the traditional apparent resistivity translation algorithm,the pseudo-resistivity depth curve of the wavelet neural network is closer to the model,and it is more sensitive to low resistivity anomalies.And thanks to the characteristics of the wavelet function,the curve is closer to the established model at the abrupt formation,and the false anomaly is reduced to a certain extent,which can better reflect the underground electrical characteristics.Compared with BP neural network,wavelet neural network has a better fitting degree to the model.Therefore,wavelet neural network is a kind of pseudo-resistivity imaging method with better effect.In the data processing part of the measured airborne transient electromagnetic data,because the wavelet neural network is superior to the BP neural network in all aspects,only the processing effect of the wavelet neural network algorithm and the all-area apparent resistivity algorithm is compared,the trained neural network is used to process and interpret the measured data,the pseudo-resistivity and depth data calculated by the wavelet neural network and the apparent resistivity and depth data calculated by the all-region apparent resistivity algorithm are obtained,and then by drawing a quasi-resistivity-depth profile,it is found that the two algorithms are in good agreement with the real situation,and the results calculated by wavelet neural network reflect the low resistance more obviously,and the pseudo-resistivity range is larger and the effect is better,the feasibility and practicability of the application of neural network in airborne transient electromagnetic imaging are verified.