Research On Noise Reduction And Reconstruction Method Of Microseismic Data

In recent years,with the deepening of coal mining,the stress of mining sites has increased,and various power disasters have become more serious and complicated than before,seriously threatening the safe production of coal mining enterprises.At present,high-precision microseismic monitoring has become an effective way to predict and forecast the dynamic disasters such as ground pressure,coal mine water inrush,coal and gas outburst.However,due to the extremely complicated underground monitoring environment,the signals picked up by the microseismic detector not only face the problem of lacking of effective data,but also are highly susceptible to contamination by external environment or random noise inside the instrument,seriously interfering with the processing and interpretation of subsequent microseismic data.It is unable to provide strong support for mine dynamic disaster prediction and early warning technology.In view of these problems,this paper focuses on the method of random noise suppression in microseismic signals and the high-precision microseismic signal recovery and reconstruction algorithm,including the following three aspects:(1)Study the suppression method of random noise in microseismic signals.The noise in the microseismic signal has obvious random non-stationary characteristics,which greatly interferes with the subsequent work such as precise picking of the microseismic signal.The traditional noise reduction methods are mostly based on the premise that the signal obeys the Gaussian distribution.However,due to the non-stationary characteristics of the microseismic signal with high noise,short duration,fast mutation,etc.,the effect of traditional noise reduction method is not ideal.Based on this,this paper chooses the Empirical Mode Decoposition(EMD)method as the main method of microseismic noise reduction processing to make up for the defects of traditional methods in non-stationary signals.(2)An improved EMD-based microseismic signal denoising algorithm is proposed.After decomposition,the traditional EMD noise reduction method abandons the components with more noise in the high frequency region and reconstructs the residual components to achieve noise reduction.But discarding the high-frequency components directly can easily cause the loss of some high-frequency effective signals,which leads to the signal after noise reduction can not well retain the high-frequency characteristics of the original signal.In this paper,a more effective EMD denoising algorithm is proposed based on the wavelet threshold algorithm.The high-frequency noisy signal components are processed by the wavelet threshold and then reconstructed,which not only suppresses the random noise effectively,but also retains the effective information of the original microseismic signal to the maximum extent.(3)Aiming at the problem of traditional compressed sensing(CS)algorithm,due to poor sparse representation of microseismic signals,the signal recovery and reconstruction effect is not good.A signal recovery improvement strategy based on compressed sensing theory is proposed.Starting from the basic knowledge of the theory of compressed sensing theory,this paper fully considers the nonlinearity and complexity of microseismic data,attempts to start from the characteristics of the signal itself,and takes the breakthrough of the traditional compressed sensing signal reconstruction method as a breakthrough to establish an improved One-dimensional microseismic data recovery reconstruction model to compensate for the sparseness of microseismic signals and achieve high-precision reconstruction.(4)Design a comparative experiment.By observing the microseismic waveform and spectrum distribution after the experiment,and comparing the evaluation indexes such as signal-to-noise ratio,standard deviation,energy ratio and correlation coefficient after the experiment,the advantages of the above algorithm in suppressing random noise and recovering and reconstructing missing signals are verified.