Research On Localization Algorithm Of Microseismic Focal Source Based On Local Sensitive Hash

Mine microseismic is a dynamic phenomenon of rapid fracture of coal and rock mass caused by human mining behavior,which poses a huge threat to the safety of mine production.Therefore,it is very important to quickly find the location of the earthquake source and take corresponding measures.This paper introduces the search engine technology into the localization of microseismic sources.This method is completely different from the traditional microseismic localization method.It uses a new method to quickly determine the location of microseisms and the focal mechanism.A waveform that is very similar to the newly input waveform is retrieved from the waveform database.The source location of the newly input waveform is the location of the microseismic event in the waveform library that is very similar to the newly input waveform.The most classic nearest neighbor search algorithm is Local Sensitive Hash(LSH).LSH has a rigorous theoretical proof,which can achieve more rapid and accurate focal location.In this paper,a local sensitive hash search engine model based on P stable distribution is used to locate the source of the experimental data and the measured data.In order to improve the recall rate and localization results of LSH,an optimized hash function method-data-oriented LSH and multi-Probe LSH is used to search the database.The main research contents are as follows:(1)The principle of microseismic signal generation in coal rock strata and the reason for microseismic events in coal rock masses are explained.The propagation characteristics of microseismic signals in coal and rock formations are introduced,the methods of identifying P-waves and S-waves are described,and the principle and simplex method of traditional microseismic positioning algorithms are explained.(2)Structural design of microseismic search engine.The platform of the microseismic monitoring system for coal mines was introduced,and the information of the microseismic waveform data and the location of the earthquake source were obtained.Because the complex environment of the coal mine makes the collected signal contain noise,it is proposed to use EEMD combined with wavelet denoising to perform noise reduction processing on coal mine microseismic signals.The experimental microseismic model was set up,and then the acceleration signal was collected from the microseismic model using the DHDAS dynamic acquisition system.The preparation and establishment of the search engine database,the measured waveform database after the denoising and interception of the measured waveform data and the corresponding source location,and the experimental waveform data after the interception of the experimental waveform data and the corresponding source location.The LSH structure and mathematical model based on P-stable distribution are introduced and improved on this basis to obtain an optimized algorithm,multi-probe LSH.Secondly,based on the characteristics of microseismic data,a multi-probe LSH microseismic localization model was established,and the effects of the number of hash tables,the size of the hash function family,and the interval size on the LSH microseismic localization model were determined.Finally,the parameters of the LSH were determined.,Locate experimental microseismic data and measured coal mine microseismic data.(3)Design of improved multi-probe LSH microseismic search engine model.For the multi-probe LSH randomly selecting the projection vector of the hash function will cause problems such as low recall rate and insufficient accuracy of the retrieval results,the PCA algorithm is used to determine the projection vector,and the structure and mathematical model of the improved multi-probe LSH are introduced.An improved multi-probe LSH microseismic positioning model was established.The improved method increased the recall rate and retrieved more accurate positioning results.