A method and system for determining magnetotelluric time domain impedance

Abstract

The present invention relates to a method and system for determining the magnetotelluric time-domain impedance. By equating the optimization problem, the operation of inverting the matrix is ​​avoided, and the noise type and distribution in the signal are not assumed, so there is no need to pass Bayesian estimation suppresses the impact of spike noise, simplifies the impedance estimation process, and reduces the amount of calculation; and, the present invention uses the interleaved multiplier iteration method to speed up the update of iteration parameters and reduce the number of variables in each iteration, compared with The steepest descent iteration used by the traditional method converges faster. In addition, the present invention uses an improved algorithm to transform the time-domain impedance estimation into an unconstrained convex optimization problem, and the solution to the convex optimization problem obtains the global optimal value, which avoids the steepest descent iteration using local optimization in the traditional method from falling into a local minimum The fitting error is reduced, and the convergence result is more reliable.

Description

technical field [0001] The invention relates to the technical field of magnetotelluric detection, in particular to a method and system for determining the impedance of the magnetotelluric time domain. Background technique [0002] Magnetotelluric sounding is an important geophysical exploration method. This method uses natural electromagnetic field signals measured on the surface to estimate impedance, and can invert to obtain underground electrical structure information. According to different processing objects, impedance estimation is divided into time domain and frequency domain impedance estimation. Among them, the time-domain impedance estimation is less sensitive to some noise and has lower requirements on the stability of the electromagnetic field. Therefore, this method can achieve better results than the frequency-domain impedance estimation in the case of strong electromagnetic interference in the survey area. However, in practical applications, due to the large ...

AI Technical Summary

Problems solved by technology

However, when the time series to be processed is long, Bayesian estimation of noise, noise suppression using Bayesian estimation results, and the calculation of adaptive step size require a large amount of calculation

At the same time, the objective function of the measured data is often complex and cannot satisfy the assumption of the negative gradient direction (ideal direction of descent) everywhere. The convergence speed of the steepest descent method will therefore slow down, and it is difficult to avoid the convergence falling into a local minimum.

[0004] In order to solve the problems of the current time-domain impedance estimation algorithm in the actual measurement data processing, such as large amount of calculation (low speed), difficult convergence, large fitting error, etc., it is urgent to provide an improved method for determining the magnetotelluric time-domain impedance in this field. method or system

Images