基于LMD和模式识别的矿山微震信号特征提取及分类方法
发布时间：2018-08-01 18:54
【摘要】：针对岩体破裂信号与爆破振动信号难以自动识别的问题,提出了基于局部均值分解(LMD)和模式识别的矿山微震信号特征提取及分类方法。首先采用LMD对微震信号进行自适应分解得到乘积函数(PF)分量,再利用相关系数和方差贡献率筛选得到PF主分量,进而计算各主分量的相关系数和能谱系数,并以此作为模式识别的特征向量。结果表明:LMD、经验模态分解(EMD)和离散小波变化(DWT)的主分量分别为PF1~PF6,IMF1~IMF6和D2~D7,其中IMFi(i=1,2,…,6)为EMD分解的本征模态分量,Dj(j=2,3,…,7)为DWT分解的细节分量;LMD主分量分类识别结果整体上优于EMD和DWT主分量分类识别结果;能谱系数分类结果整体上优于相关系数分类结果,人工神经网络(ANN)和支持向量机(SVM)识别效果明显优于逻辑回归(LR)和Bayes判别法识别结果,且基于LMD能谱系数的SVM分类准确率达到了93.0%。
[Abstract]:In view of the problem that the rock burst signal and the blasting vibration signal are difficult to recognize automatically, a method based on local mean mean decomposition (LMD) and pattern recognition is proposed to extract and classify the characteristics of the micro earthquake signals. First, the product function (PF) component is obtained by LMD adaptive decomposition of the microseismic signal, and the correlation coefficient and the variance contribution rate sieve are used. The principal components of PF are selected and then the correlation coefficients and the spectral coefficients of the main components are calculated and used as the eigenvectors of the pattern recognition. The results show that the main components of the LMD, the empirical mode decomposition (EMD) and the discrete wavelet change (DWT) are PF1~PF6, IMF1~IMF6 and D2~D7 respectively, and IMFi (i=1,2,...) 6) the eigenmode component of the EMD decomposition, Dj (j=2,3,...) 7) is the detail component of DWT decomposition; the LMD principal component classification recognition results are better than the EMD and DWT principal component classification recognition results. The classification results of the energy spectrum coefficient are better than the correlation coefficient classification results as a whole. The recognition results of the artificial neural network (ANN) and the support vector machine (SVM) are obviously better than the logical regression (LR) and Bayes discriminant recognition results. The SVM classification accuracy of LMD spectrum coefficient reached 93.0%.
【作者单位】： 黑龙江科技大学黑龙江省普通高校采矿工程重点实验室;黑龙江科技大学矿业工程学院;
【基金】：黑龙江省普通高等学校采矿工程重点实验室开放课题资助项目(2014KF04) 黑龙江省自然科学基金面上资助项目(E2016061)
【分类号】：TD311;TN911.7
本文编号：2158515
[Abstract]:In view of the problem that the rock burst signal and the blasting vibration signal are difficult to recognize automatically, a method based on local mean mean decomposition (LMD) and pattern recognition is proposed to extract and classify the characteristics of the micro earthquake signals. First, the product function (PF) component is obtained by LMD adaptive decomposition of the microseismic signal, and the correlation coefficient and the variance contribution rate sieve are used. The principal components of PF are selected and then the correlation coefficients and the spectral coefficients of the main components are calculated and used as the eigenvectors of the pattern recognition. The results show that the main components of the LMD, the empirical mode decomposition (EMD) and the discrete wavelet change (DWT) are PF1~PF6, IMF1~IMF6 and D2~D7 respectively, and IMFi (i=1,2,...) 6) the eigenmode component of the EMD decomposition, Dj (j=2,3,...) 7) is the detail component of DWT decomposition; the LMD principal component classification recognition results are better than the EMD and DWT principal component classification recognition results. The classification results of the energy spectrum coefficient are better than the correlation coefficient classification results as a whole. The recognition results of the artificial neural network (ANN) and the support vector machine (SVM) are obviously better than the logical regression (LR) and Bayes discriminant recognition results. The SVM classification accuracy of LMD spectrum coefficient reached 93.0%.
【作者单位】： 黑龙江科技大学黑龙江省普通高校采矿工程重点实验室;黑龙江科技大学矿业工程学院;
【基金】：黑龙江省普通高等学校采矿工程重点实验室开放课题资助项目(2014KF04) 黑龙江省自然科学基金面上资助项目(E2016061)
【分类号】：TD311;TN911.7
【相似文献】
相关期刊论文 前5条
1 靳奉祥,王同孝,独知行;变形观测中的模式识别问题[J];中国有色金属学报;1997年03期
2 陈志强;周科平;;基于模式识别的土石坝边坡稳态预测[J];采矿技术;2010年02期
3 陈遵德;测井数据模式识别中的信息优化方法[J];测井技术;1998年06期
4 卢宗华;巷道围岩类型模式识别的物元模型[J];山东科技大学学报(自然科学版);2003年01期
5 ;[J];;年期
相关硕士学位论文 前1条
1 荆晓亮;煤与瓦斯突出灾害信息模式识别与集成决策[D];辽宁工程技术大学;2012年
，本文编号：2158515
本文链接：https://www.wllwen.com/kejilunwen/xinxigongchenglunwen/2158515.html
