基于CEEMDAN-BO-SVM的选煤设备故障诊断方法  

Fault Diagnosis Method of Coal Preparation Equipment Based on CEEMDAN-BO-SVM

作　　者：谢尚海 李敬兆[1] 王国锋 XIE Shang-hai;LI Jing-zhao;WANG Guo-feng(School of Electrical and Information Engineering,Anhui University of Science and Technology,Huainan 232001,Anhui,China;Huaihe Energy(Group)Co.,LTD.,Huainan 232001,Anhui,China)

机构地区：[1]安徽理工大学电气与信息工程学院,安徽淮南232001 [2]淮河能源(集团)股份有限公司,安徽淮南232001

出　　处：《兰州文理学院学报（自然科学版）》2024年第3期58-64,共7页Journal of Lanzhou University of Arts and Science(Natural Sciences)

基　　金：国家自然科学基金项目(51874010,61170060)。

摘　　要：针对选煤设备故障诊断准确率低且所需特征信号难以采集和提取等问题,提出一种基于自适应噪声完备经验模态分解(CEEMDAN)及黑猩猩算法(BO)优化支持向量机(SVM)的音频信号故障诊断方法.对设备的原始音频信号进行CEEMDAN分解后得到一系列本征模态分量(IMF),计算各IMF的峭度值-相关系数,依据筛选准则优选有效IMF分量.提取有效分量的能量系数及波形系数,组成故障诊断特征集,使用BO-SVM进行故障诊断.实验结果表明,本文方法的故障诊断平均准确率为96.8%,在音频信号特征提取及故障诊断领域有一定的优势,具有一定的应用价值.To solve the problems of low accuracy in fault diagnosis of coal preparation equipment and difficulty in collecting and extracting the required characteristic signals,this paper presents an audio signal fault diagnosis method based on CEEMDAN-BO-SVM.After CEEMDAN decomposition of the original audio signal of the device,a series of IMFs are obtained,the kurtosis-correlation coefficients of each IMF are calculated,and the effective IMF components are selected according to the selection criteria.The energy and waveform coefficients of the effective components are calculated to form a feature set for fault diagnosis,and the BO-SVM is used for fault diagnosis.Experimental results show that the average fault diagnosis accuracy of the proposed method is 98.5%.The comparative study verifies the advantages of the proposed method in the field of audio signal feature extraction and fault diagnosis,and has certain practical application value.

关 键 词：音频信号 故障诊断 自适应噪声完备经验模态分解 特征提取 支持向量机 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]