Method for identifying a microearthquake event with low signal-to-noise ratio based on multi-scale permutation entropy

Abstract

The invention discloses a method for identifying a microearthquake event with a low signal-to-noise ratio based on multi-scale permutation entropy. The method is performed by a computer and comprises steps of: processing acquired microearthquake data to convert the microearthquake into time series data, coarse graining a time series to obtain a multi-scale time series, and computing multi-scale time series permutation entropy; training a least squares support vector machine (LS-SVM) on the basis of the multi-scale time series permutation entropy, and identifying a signal to be identified by using the trained LS-SVM. The method may analyze signal features by applying multiple scales and the permutation entropy to the microearthquake signals, accurately expresses the waveform characteristics in multiple dimensions of the microearthquake signals, and is beneficial to discrimination between the microearthquake event and a noise event. The method extracts characteristic data of the microearthquake signals and the noise signals by using the multi-scale permutation entropy, trains the characteristic data by using the LS-SVM to obtain the LS-SVM, and accurately classify the microearthquake event with a low signal-to-noise ratio and the noise event.

Description

technical field [0001] The invention relates to a method for identifying microseismic events with a low signal-to-noise ratio. Specifically, it is a method of using multi-scale permutation entropy to represent the characteristics of microseismic waveforms, and then using a support vector machine to determine whether the waveform belongs to a microseismic event. It belongs to information processing technology field. Background technique [0002] In the process of coal mining, microseismic signals are generated by coal and rock mass fracture under pressure. The microseismic signal generated by a rupture is called a microseismic event. Collecting as many microseismic events as possible and performing positioning and energy calculation on them can directly reflect the surrounding environment. The distribution of dynamic stress in rock mining can reveal the spatiotemporal evolution of the stress field, which is of great significance for early warning of coal-rock dynamic disaster...

AI Technical Summary

Problems solved by technology

[0003] The STA / LTA (Short Time Average / Long Time Average) method is the most commonly used microseismic event identification method. This method constructs a characteristic function based on the signal energy change in the time domain, and uses the ratio of STA to LTA to identify the first arrival of the seismic phase. This method is simple to implement , the calculation speed is fast, but the interference signal will be misjudged as a microseismic event, and the anti-noise performance of the algorithm is poor; the AIC (Akaike Information Criterion, AIC) method is another commonly used microseismic event identification method. The statistical difference of the waveform data of the seismic phase is given, and the AIC criterion for judging the first arrival of the seismic phase is given. The AIC function has a sharp local minimum peak when the first arrival of the seismic phase arrives, and the accuracy is high. However, the AIC function is greatly affected by the time window. Different The location of the minimum value under the time window is also different, so it is only applicable to the situation where the approximate location of the first arrival of the seismic phase is known; at present, the identification of microseismic events is mainly based on manual judgment with the assistance of the above algorithm software, and the processing of data The efficiency is low, and the identification standards are not uniform, and it is easy to misjudgment

Images