Industrial Process Fault Monitoring Method Based on Optimized Dynamic Kernel Principal Component Analysis
A fault monitoring method based on artificial gorilla force optimization algorithm optimized dynamic kernel principal component analysis(GTO-DKPCA)is proposed for nonlinear multimodal data in modern industrial production processes.Using the autoregressive moving average time series model and kernel principal component analysis(KPCA)method,a DKPCA model is constructed to process batch data at each stage of the process.We construct an adaptive degree function based on normal and fault data features,and use the artificial gorilla force optimization algorithm to optimize the DKPCA kernel parameters to discover the optimal nonlinear features.Fault monitoring is carried out by calculating the Hotelling statistic T2 and the squared prediction error SPE statistic at each time point.The results of penicillin fermentation process indicate that the GTO-DKPCA method has better monitoring effect,adaptability,and accuracy than multi directional kernel principal component analysis(MKPCA)and multi dynamic kernel principal component analysis(BDKPCA).
dynamic kernel principal component analysis(DKPCA)artificial gorilla troop optimization(GTO)algorithmfault monitoringpenicillin fermentation
万方数据
维普
