基于改进的HHT-SVM微震信号特征提取及分类识别研究

Research on feature extraction and classification and identification of microseismic signals based on improved HHT-SVM

为了解决实际工程中微震信号与爆破信号在波形上具有较强相似性,难以识别的问题,提出基于改进HHT-SVM的微震信号识别方法。分析2种信号的瞬时能量变化规律,发现微震信号具有孕育速度慢、持续时间短及峰后衰减速度慢的特点,对其进行量化处理并提取特征,构建多维特征向量。对输入的特征向量经功率谱熵特征加权算法赋予权重,并使用GSWOA算法对SVM进行参数寻优,建立信号识别网络对微震信号、爆破信号进行识别。研究结果表明:方法的综合识别成功率达到96.250%,能够有效识别微震、爆破信号。研究结果可为后续利用微震信号进行灾害预警提供有力支撑。

To solve the problem that the microseismic signals and blasting signals in actual engineering have strong similarity in waveforms and are difficult to identify,an identification method of microseismic signals based on improved HHT-SVM was proposed.The instantaneous energy change laws of the two signals were analyzed,and it was found that the microseismic signals had the characteristics of slow gestation speed,short duration and slow decay speed after the peak,then the microseismic signals were quantized and the features were extracted to construct a multi-dimensional feature vector.The input feature vectors were given weights by the power spectrum entropy feature weighting algorithm,then the parameters of SVM were optimized by using GSWOA algorithm,and a signal identification network was established to identify the microseismic signals and blasting signals.The results show that the comprehensive identification success rate of this method reaches 96.250%,which can effectively identify the microseismic and blasting signals.The research results can provide strong support for the subsequent use of microseismic signals for disaster warning.