基于OVMD算法集成学习模型的火电厂关键辅机故障诊断  

Integrated Learning Model Based on OVMD Algorithm for Fault Diagnosis of Critical Auxiliary Machines in Thermal Power Plants

作　　者：周传杰 张林 陈节涛 张航 裴浩然 徐春梅 彭道刚 ZHOU Chuanjie;ZHANG Lin;CHEN Jietao;ZHANG Hang;PEI Haoran;XU Chunmei;PENG Daogang(Guodian Changyuan Hanchuan No.1 Power Generation Co.,Ltd.,Wuhan 431614,China;College of Automation Engineering,Shanghai University of Electric Power,Shanghai 200090,China)

机构地区：[1]国电长源汉川第一发电有限公司,湖北武汉431614 [2]上海电力大学自动化工程学院,上海200090

出　　处：《自动化仪表》2023年第4期43-47,共5页Process Automation Instrumentation

摘　　要：针对火电厂辅机设备运行工况复杂、系统非线性强、易受背景噪声干扰、故障特征难以提取等问题,提出1种基于最优变分模态分解(OVMD)算法的集成学习模型故障诊断方法。首先,使用OVMD算法对辅机的纵向与横向原始振动信号进行预处理,从中提取均方根、裕度、峰值、平均值、波形指标、方差等10个参数作为轻量级梯度提升机(LightGBM)的特征向量。然后,结合集成学习算法构造Bagging-LightGBM集成学习模型。试验结果表明:与单一的LightGBM分类器相比,Bagging-LightGBM集成学习模型对于火电厂辅机故障诊断性能更优。集成学习模型为火电厂辅机故障诊断研究提供了参考。An integrated learning model fault diagnosis method based on the optimal variational modal decomposition(OVMD)algorithm is proposed for the problems of complex operating conditions of auxiliary equipment in thermal power plants,strong system nonlinearity,susceptibility to background noise interference,and difficulty of fault feature extraction.Firstly,the OVMD algorithm is used to preprocess the longitudinal and transverse raw vibration signals of the auxiliary machine,from which ten parameters such as root mean square,margin,peak,mean,waveform index,variance,etc.are extracted as the feature vectors of the light gradient boosting machine(LightGBM).Then,the integrated learning algorithm is combined to construct the Bagging-LightGBM integrated learning model.The experimental results show that the Bagging-LightGBM integrated learning model has better performance for fault diagnosis of thermal power plant auxiliary machines compared with the single LightGBM classifier.The integrated learning model provides a reference for the research of auxiliary machines fault diagnosis in thermal power plants.

关 键 词：火电厂 关键辅机 最优变分模态分解算法 集成学习 轻量级梯度提升机 特征提取 故障诊断 

分 类 号：TH133[机械工程—机械制造及自动化]