基于PCA的火电厂湿法烟气脱硫系统的传感器故障诊断

Study on sensor fault diagnosis of wet flue gas desulfurization system in thermal power plant based on PCA

提出一种基于主元分析(PCA)的故障诊断方法,用于解决火电厂湿法烟气脱硫系统的传感器故障诊断问题.该方法利用PCA建立故障诊断模型,通过计算平方预报误差、传感器识别指数、故障重构值,对传感器故障进行检测、识别及恢复.利用华能福州电厂湿法烟气脱硫系统的采集数据进行传感器的完全失效、偏差、漂移与精度等级下降等4种类型故障的实验验证,结果表明,该方法对湿法烟气脱硫系统的传感器故障具有良好的诊断效果及恢复能力.

A fault diagnosis method using principal component analysis （PCA） is proposed to solve sensor fault diagnosis of wet flue gas desulfurization system in thermal power plant. Based on PCA model, the sensor faults are detected, identified and recovered by calculating square prediction error, sensor validation index, reconstruction index. Employing the actual data from wet flue gas desulfurization system of Huaneng Fuzhou power plant, the PCA model is proved effectively enough to detect and identify the complete invalidation fault, fixed bias fault, drift bias fault and accuracy decrease fault of sensors. The results show that the PCA approach has a good ability in fault diagnosis and recovery.