Research On A New Method Of Transient Electromagnetic Denoising

Transient electromagnetic method is a commonly used method in geological exploration,which is applicable to a variety of topographic environments and has a sensitive response.It is widely used in the exploration of hydrogeology,underground roadway,coal field,oil and gas field,and can effectively detect coal goaf,mine water,oil and gas field and metal mineral.In the process of collecting transient electromagnetic data in the field,the signal will be interfered by the noise,resulting in data interpretation errors and low inversion accuracy.Therefore,to find a proper denoising method can effectively improve the interpretation accuracy of the data and avoid obtaining the wrong geological information.Transient electromagnetic signal is easy to be interfered by random noise,atmospheric noise,and human interference.The traditional denoising methods can not meet the needs of denoising amount and effect.Many methods,such as wavelet threshold method,empirical mode decomposition method,singular value decomposition method,can effectively suppress one of the above noise.These denoising methods are summarized and their suitable range are found out in this paper.At the same time,according to the derivation formula of transient electromagnetic signal,the ideal transient electromagnetic signal is simulated,and the white noise is used to simulate the random noise,the alpha stable distribution is used to simulate the atmospheric noise,the 50 Hz sine wave and its harmonics are used to simulate the human noise,which are used for the denoising experiment analysis.The main contents of this paper are as followed:1.This paper discusses a denoising method of transient electromagnetic signal based on the principle of minimum mean square error.The transient electromagnetic signal with random noise and human noise is processed in frames,and the short-time Fourier transform is applied to each frame to calculate the short-time spectrum amplitude.Then,the prior SNR and the posterior SNR are estimated,and the mean square error between the estimated amplitude and the actual amplitude is minimized.Finally,the phase information of the noisy signal is combined to obtain the estimated enhanced signal.According to the experiment,this method can effectively suppress the noise.2.Most of the denoising methods of transient electromagnetic signal can only deal with single type noisy signal,so,in the complex environment,the denoising effect is general,sometimes even obvious distortion appears.This dissertation researches a denoising method based on SVD and sine wave estimation.In this method,the singular value of the noisy signal is decomposed first,then the sine wave of the initial denoising signal is estimated and eliminated.This method has a good denoising effect.3.Because the noise environment of the collected transient electromagnetic signal is different,the denoising effect of the same algorithm is different too.This paper researches the denoising effect of each method in several noise environments,and the classification technology based on convolution neural network is used to identify the noise of noisy signals.Through the recognition results,the corresponding method is selected to achieve the optimal denoising.