基于动态稀疏主成分分析的热工过程故障检测方法

Fault Diagnosis Method for Thermal Processes Based on DSPCA

针对主成分分析(PCA)法和稀疏主成分分析(SPCA)法检测动态数据敏感度不足及动态主成分分析(DPCA)法噪声过多的问题,提出了一种新的用于热工过程故障检测方法,即动态稀疏主成分分析(DSPCA)法。分别采用PCA法、DPCA法、SPCA法、DSPCA法进行故障检测,并将DSPCA法应用到高压加热器的故障检测中。结果表明:DSPCA法检测正确率更高,该方法对热工过程数据具有较强的故障敏感性,对高压加热器管系泄漏故障有较高的检测正确率,适用于工业过程的实时故障检测。

To solve the problems of insufficient sensitivity of principal component analysis(PCA)and sparse principal component analysis(SPCA)in detecting dynamic data as well as the excessive noise produced by dynamic principal component analysis(DPCA),a new fault diagnosis method for thermal processes,namely dynamic sparse principal component analysis(DSPCA),was proposed.Fault detections were conducted by PCA,DPCA,SPCA and DSPCA,with DSPCA applied to the fault detection of high-pressure heaters.Results show that the DSPCA method has higher detection accuracy,with strong fault sensitivity to the data of thermal processes,especially to the pipeline leakage fault of high-pressure heaters,which therefore is suitable for real-time fault detection of industrial processes.