基于WDNS-SWNPE算法的间歇过程故障检测

Fault detection of batch process based on WDNS-SWNPE algorithm

针对间歇过程数据的多模态与动态特性共存带来的故障检测问题,提出一种基于加权双近邻标准化(WDNS)的稀疏加权邻域保持嵌入(SWNPE)算法.首先,在寻找样本双层近邻的基础上加权得到加权双近邻集,用加权双近邻集信息标准化样本,将多模态数据处理为单一模态分布,消除多模态中心点差异,解决多模态特性;然后,考虑到NPE算法不能更好地处理动态特性带来的问题,利用反距离加权和局部最优稀疏表示的方法在NPE算法的基础上得到SWNPE算法,在处理数据动态特性的同时增强了噪声和离群点影响的鲁棒性;最后,采用加权双近邻标准化的SWNPE模型实现故障监控.青霉素发酵仿真检测效果对比表明:WDNS-SWNPE算法具有更高的检测率,提高了动态与多模态特性共存间歇过程的故障检测能力.

To solve the problem of fault detection caused by the coexistence of multimodal and dynamic characteristics of batch process data,a sparse weighted neighborhood preserving embedding(SWNPE) algorithm based on weighted double neighborhood standardization(WDNS) was proposed. First,based on finding the double nearest neighbors of the sample,the weighted double nearest neighbor set was obtained,the weighted double nearest neighbor set information was used to standardize the sample,and the multimodal data was processed into a single modal distribution,which could eliminate the difference of multimodal center points and effectively solve the multimodal characteristics.Then,considering that the NPE algorithm could not better deal with the problems caused by the dynamic characteristics,the SWNPE algorithm was obtained based on the NPE algorithm by using the inverse distance weighting and local optimal sparse representation. The SWNPE algorithm not only could deal with the dynamic characteristics of the data,but also could enhance the robustness of noise and outliers. Finally,the weighted double nearest neighbor standardized SWNPE model was used to realize fault monitoring. Comparison of detection results through penicillin fermentation simulation process,results show that the WDNS-SWNPE algorithm has a higher detection rate and improves the fault detection ability of batch processes with dynamic and multimodal characteristics.