Due to the influence of component performance decline and operation condition change,nuclear power plants(NPPs)show obvious time variability during operation,which leads to the failure of fault detection model.In order to improve the performance and in-service adaptability of traditional fault detection methods in time-varying industrial processes,this paper proposes a long-term fault detection strategy for NPPs based on kernel principal component analysis(KPCA)and moving window.In this method,the KPCA fault detection model is automatically updated by moving window technology,which solves the time-varying problem of signals in the detection process.The moving window KPCA method is applied to the long-term monitoring of the reactor coolant pump in a nuclear power plant.The results show that the moving window KPCA method has good performance in fault detection rate and false alarm rate under normal and abnormal conditions.
Nuclear power plantKernel principal component analysis(KPCA)Moving windowTime-varying processFault detection
维普
万方数据
