基于波形特征的矿山微震监测事件分类方法研究

Study on classification method of mine microseismic monitoring events based on waveform features

目的:针对微震监测技术应用中出现的微震爆破信号分类困难、人工分类工作量大、分类准确率低等问题,本文创新性地提出了支持向量机的波形自动分类技术,旨在解决微震监测系统现有应用难题。方法:针对BSN微震监测系统采集的微震信号和人工爆破信号进行了时域和频域的分析,获得了信噪比、P波到时、首次峰值、最大峰值、主频及P波主频等9个波形特征,经过分析对比后挑选出了5个典型特征,最后采用SVM方法对上述典型特征进行训练,建立了微震和爆破信号自动分类模型,并在广西某金属矿山进行了实际现场应用分析。结果:微震准确率为80.28%,爆破准确率为86.36%。结论:本文研究的分类模型的识别和分类方法使分类准确率达到了80%以上,说明成果能够协助矿山人员减少部分波形分类工作,有助于推动全自动矿山微震监测系统发展。意义:本文提供的分类方法研究思路,能够实际解决微震监测系统在现场应用中的难题,且其中的分析方法不仅可在微震处理中应用,还可进一步用于微震分析,对实现矿山微震实时监测服务具有重大意义。

Purpose:In response to the difficulties in classifying microseismic blasting signals,heavy manual classification workload,and low classification accuracy in the application of microseismic monitoring technology,this article innovatively proposes the waveform automatic classification technology of support vector machine,aiming to solve the existing application problems of microseismic monitoring systems.Method:This article conducted time-domain and frequency-domain analysis on the microseismic signals and artificial blasting signals collected by the BSN microseismic monitoring system.Nine waveform features,including signal-to-noise ratio,P-wave arrival time,first peak,maximum peak,main frequency,and P-wave main frequency,were obtained.After analysis and comparison,five typical features were selected.Finally,SVM method was used to train the above typical features,and an automatic classification model for microseismic and blasting signals was established.The actual on-site application analysis was conducted in a metal mine in Guangxi.Result:The microseismic accuracy is 80.28%,and the blasting accuracy is 86.36%.Conclusion:The recognition and classification methods of the classification model studied in this article have achieved a classification accuracy of over 80%,indicating that the results can alleviate the labor intension of processing personnel and promote the development of fully automatic mining microseismic monitoring systems.Significance:The classification method research ideas provided in this article can effectively solve the difficulties in on-site application of microseismic monitoring systems,and the analysis methods can not only be applied in microseismic processing,but also further used in microseismic analysis,which is of great significance for achieving real-time monitoring services for mining microseismic.