Based On FastICA And Wavelet Threshold Joint Algorithm For Transient Eletromagnectic Signal Noise Reduction

The People use transient electromagnetic method that belongs to geophysical exploration to detect the coal goaf, it is one of the most effective methods. It emits the current through the grounding electrodes or not grounding loop line, which set up a magnetic field in the underground, and then in the intermittent of shutting off the current, secondary field signal is acquired, it has high resolution in small inhomogeneous geological object. There are convenient and economic advantages, but it was more susceptible to the disturbance of noise, such as the high voltage lines, radio waves and the instrument itself, the noise can lead to the distortion of signals, and the signals can not explore so deep in the underground as we expected.This paper is based on the principle of using transient electromagnetic method and the characteristics of signals attenuation. Firstly probes the noise sources which polluted transient electromagnetic signals and noise characteristic. Summarizes some manners that the previous scholars brought forward for reducing the noise, such as three approximate nonlinear index noise removing, the superposition of signal, wavelet analysis, but so far an effective denoising manner has not yet appeared. And then puts forward a research method which connects FastICA with wavelet threshold to reduce the signal noise. And introduces the theory of the ICA and wavelet analysis,the main content contains the FastICA and wavelet threshold, FastICA is the most effective and fastest algorithm of ICA, wavelet threshold is also one of the wavelet analysis algorithm, comparing every layer’s coefficient which was decomposed by wavelet with a certain threshold coefficient. After that, rebuild a new signal. Finally, simulation experiment and field test demonstrate the feasibility that using firstly FastICA and then wavelet threshold reduce the noise of transient electromagnetic signals.,to judge with the reducing effection of the different signal-to-noise ratio signals。This paper focuses on the argument that whether it is possible to reduce transient electromagnetic signal noise by the union algorithm of FastICA and wavelet threshold, after all, the combination algorithm was commonly used to separate the noise from the signals. Achieving simulation experiment by means of Matlab software, make use of Matlab FastICA toolbox and call the wavelet threshold function, in strictly accordance with the FastICA calculation steps, choosing suitable wavelet threshold parameters to calculate. Obtaining some graphics contrast with separately using FastICA, wavelet threshold, the three related noise reduction method, and evaluate the relevance. confirmed that joint calculation of the noise control effect is superior to other methods:1. the amplitude of data has been changed after FastICA calculation, but the shape of curve after joint calculation was more close to ideal curve:2. After wavelet threshold calculation, the curve is more smooth,easier to recognize and further judgment. The combined algorithm is attempted to apply in the field practical work. the field experiment completely comply with the existing transient electromagnetic work methods and data processing. On multi-channel profiles finding the position of possible coal goaf is perfectly fit for the practical ones. further proved the feasibility that take advantage of combined algorithm to reduce noise. So the joint calculation for coal goaf detecting, can improve the signal utilization and SNR, and improve the detection accuracy, which provides the guiding significance. For the transient electromagnetic method in detecting metal deposits, groundwater, etc of data processing provides reference.