基于改进PCA联合SVM的电厂设备故障诊断方法
投稿时间:2022-08-30  修订日期:2022-09-19  点此下载全文
引用本文:
摘要点击次数: 34
全文下载次数: 0
作者单位邮编
刘兴彦 杭州华电半山发电有限公司 310000
郑迎九 杭州华电半山发电有限公司 
基金项目:中国华电集团有限公司项目,课题名称:杭州华电半山发电有限公司数字电厂建设信息平台项目(课题编号:JG0120190686)
中文摘要:转子系统是燃气轮机极为重要的组成部件,对其进行故障诊断与分析对燃气轮机的安全稳定运行具有重要意义。转子故障信号为典型的非线性、非平稳和微弱性时间序列。提出一种基于改进主成分分析(Improved Principal Component Analysis, ImPCA)的燃气轮机转子故障诊断方法。首先针对传统PCA主分量个数确定难题,将贝叶斯理论引入PCA,构建贝塔先验主成分分析模型对转子故障信号进行自适应分解,将其转化为少数几个主分量(Principal Component, PC)之和的形式,然后将PC对应的大特征值作为特征向量并构建SVM分类器进行分类,实现对“不平衡故障”,“动静件碰磨故障”和“不对中故障”三种燃气轮机转子故障的有效分类诊断。基于实际数据的实验结果表明,所提方法能够获得优于97.2%的诊断正确率,并且具有噪声稳健性,适用于实际工程应用场景。
中文关键词:故障诊断  主成分分析  燃气轮机  贝叶斯理论  特征提取
 
Fault diagnosis method of power plant equipment based on Improved PCA and SVM
Abstract:Rotor system is a very important component of gas turbine. Fault diagnosis and analysis is of great significance to the safe and stable operation of gas turbine. Rotor fault signals are typical nonlinear, nonstationary and weak time series. A fault diagnosis method for gas turbine rotor based on improved principal component analysis (IMPCA) is proposed. Firstly, aiming at the problem of determining the number of principal components of traditional PCA, Bayesian theory is introduced into PCA, and a beta priori principal component analysis model is constructed to adaptively decompose the rotor fault signal, which is converted into the form of the sum of a few principal components (PCS). Then, the large eigenvalues corresponding to PCs are used as feature vectors and SVM classifier is constructed to classify the "unbalanced fault", Effective classification and diagnosis of three kinds of gas turbine rotor faults, i.e. "rubbing fault of moving and stationary parts" and "misalignment fault". The experimental results based on the actual data show that the proposed method can obtain a diagnosis accuracy rate better than 97.2%, and has noise robustness, which is suitable for practical engineering application scenarios.
keywords:Fault diagnosis  Principal component analysis  Gas turbine  Bayesian theory  Feature extraction
查看全文   查看/发表评论   下载pdf阅读器