基于EEMD-KPCA和KL散度的垮落煤岩识别  

Caving coal-rock identification based on EEMD-KPCA and KL divergence

作　　者：李一鸣 白龙[1,2] 蒋周翔 高宏 黄小龙[1] 刘相权[1] 黄民[1] LI Yiming;BAI Long;JIANG Zhouxiang;GAO Hong;HUANG Xiaolong;LIU Xiangquan;HUANG Min(School of Mechanical Electrical Engineering,Beijing Information Science&Technology University,Beijing 100192,China;Key Laboratory of Modern Measurement&Control Technology,Ministry of Education,Beijing Information Science&Technology University,Beijing 100192,China)

机构地区：[1]北京信息科技大学机电工程学院,北京100192 [2]北京信息科技大学现代测控技术教育部重点实验室,北京100192

出　　处：《煤炭学报》2020年第2期827-835,共9页Journal of China Coal Society

基　　金：北京市自然科学基金京津冀基础研究合作专项资助项目(J170004);北京市属高校高水平创新团队建设计划资助项目(IDHT20180513);北京市教委科研资助项目(KM201811232001)

摘　　要：综放开采中关键工艺放煤自动化的实现,可以使放煤工人远离工作面,远程控制顶煤的放落,从而保障放煤工人的健康问题。而顶煤放落过程中垮落煤岩的实时有效识别是放煤自动化的理论基础。针对垮落煤岩识别的实时性和综放开采的效率问题,基于综放开采现场采集的垮落煤岩冲击液压支架后尾梁的振动信号,提出了一种基于EEMD-KPCA和KL(Kullback-Leibler)散度的垮落煤岩识别方法。该方法首先对振动信号进行集合经验模态分解(Ensemble Empirical Mode Decomposition,EEMD)得到一组固有模态分量(Intrinsic Mode Functions,IMFs),分别计算各个IMF的能量,峭度和样本熵,构造表征垮落煤岩的特征向量;然后利用核主成分分析(Kernel Principal Component Analysis,KPCA)对特征向量进行降维,分别以各个IMF能量、IMF峭度、IMF样本熵构成的向量的KPCA低维特征来表征垮落煤岩;最后计算"未知样本"不同特征向量与垮落煤岩两类样本与之对应的特征向量的KL散度值,通过比较KL散度值来实现垮落煤岩的实时识别,并比较表征垮落煤岩不同特征向量的特征提取耗时和识别有效性。然后利用BP神经网络验证基于EEMD-KPCA的特征向量的有效性。实验结果表明:基于EEMD-KPCA和KL散度的识别方法可以实现垮落煤岩的实时识别,且大大降低了传统垮落煤岩识别方法对综放开采效率的影响;由EEMD分解后的各个IMF的能量和峭度构成的向量的KPCA低维特征最有效。The realization of automatic top coal caving(key technology in a fully mechanized caving mining face) can enable the coal miners to be away from the working face and to remotely control the top coal caving,thereby protecting coal miners.Moreover,the effective and real-time caving coal-rock identification in the process of top coal caving can provide a theoretical basis for the control of top coal caving.Considering the real-time recognition of caving coal-rock and the efficiency of the fully mechanized caving mining,a new caving coal-rock identification method based on EEMD-KPCA and KL divergence is proposed.The method is based on the vibration signals caused by the impact of caving coal-rock and hydraulic support tail beam on the scene.Firstly,the vibration signals are decomposed by Ensemble Empirical Mode Decomposition(EEMD) and a number of Intrinsic Mode Functions(IMFs) are obtained.To construct the eigenvectors characterizing the caving coal and rock,the energy,kurtosis and sample entropy of each IMF are calculated respectively.Then the Kernel Principal Component Analysis(KPCA) is employed to reduce the dimensionality of the feature vectors.The caving coal-rock is characterized by the KPCA low-dimensional feature of the vectors composed of the energy of IMFs,the kurtosis of IMFs,the sample entropy of IMFs respectively.Finally,the values of the KL divergence between the eigenvector of ’unknown sample’ and the eigenvector of caving coal sample or caving rock sample are calculated,the real-time recognition of caving coal-rock is achieved by comparing the values of the KL divergence,and the feature extraction time consuming and recognition effectiveness of different features characterizing caving coal-rock are compared.Then BP neural network is used to verify the validity of the eigenvectors based on EEMD-KPCA.The results show that the identification method based on EEMD-KPCA and KL divergence can realize the real-time recognition of caving coal-rock,and greatly reduce the influence of the traditional identif

关 键 词：综放开采 垮落煤岩识别 EEMD KPCA KL散度 

分 类 号：TD823.97[矿业工程—煤矿开采]