摘要
在大数据背景下,基于机器学习的旋转机械故障诊断研究发展迅速。对于非线性非平稳的振动信号,提出了基于熵权法优化PSO-SVM故障诊断模型,强化特征提取能力。本文数据来自CWRU数据集和抽水蓄能电站实测数据,首先,分别对每个数据集的四种故障滑动采样、平滑降噪等预处理,其次,对故障样本VMD分解,利用样本熵、能量熵、模糊熵、功率谱熵构建特征向量,采用熵权法选取权值最大的特征向量作为EWM-PSO-SVM模型输入,得到诊断结果,同时与其他方法进行对比证实方法有效性与准确性。
Under the background of big data,the research on fault diagnosis of rotating machinery based on machine learning is developing rapidly.For nonlinear and non-stationary vibration signals,this paper proposes a PSO-SVM fault diagnosis model based on entropy weight method to strengthen the feature extraction ability.The data in this paper are from the CWRU data set and measured data of pumped storage power station.Firstly,the four kinds of fault sliding sampling,smoothing and noise reduction of every data set are preprocessed respectively.Secondly,the fault sample VMD is decomposed,and the sample entropy,energy entropy,fuzzy entropy and power spectrum entropy are used to construct the feature vector.The entropy weight method is used to select the feature vector with the largest weight as the input of the EWM-PSO-SVM model,and the diagnosis results are obtained.At the same time,the validity and accuracy of the method are verified by comparison with other methods.
作者
彭绪意
刘泽
吴中华
聂赛
章志平
姚婵
冯陈
张玉全
PENG Xuyi;LIU Ze;WU Zhonghua;NIE Sai;ZHANG Zhiping;YAO Chan;FENG Chen;ZHANG Yuquan(Jiangxi Hongping Pumped Storage Co.Ltd.,Yichun 330600,China;College of Energy and Electrical Engineering,HohaiUniversity,Nanjing 211100,China)
出处
《水电与抽水蓄能》
2024年第3期41-48,共8页
Hydropower and Pumped Storage
基金
国网新源控股有限公司科技项目资助“发电机转动部分状态预警及评估系统研究与应用”(SGXYKJ-2022-034)
国家自然科学基金重点项目“水力发电机组系统动态安全机制与运维调控”(52339006)
关键词
熵权法
粒子群算法
变分模态分解
支持向量机
故障诊断
动力学熵
entropy weight method
particle swarm optimization algorithm
variational mode decomposition
support vector machine
fault diagnosis
kinetic entropy
