Multi‐block principal component analysis based on variable weight information and its application to multivariate process monitoring

Abstract

The traditional principal component analysis (PCA)‐based process monitoring method builds a global statistical model and omits the mining of the local variable behaviours, which may degrade the fault detection performance. Considering this problem, this paper proposes a variable weight information‐based multi‐block PCA (VWI‐MBPCA) method. Firstly, a sequence hierarchical clustering algorithm is proposed to divide the full PCA component space into several sub‐blocks, where the components sharing similar variable weight information are gathered together and then the sub‐block T2 statistic is constructed for monitoring sub‐block components. Further, the variables with small weight information on each component sub‐block are extracted to build an additional sub‐PCA model, where the trueTˆ2 statistic is developed to compensate the sub‐block T2 statistic. In order to integrate the monitoring results of each sub‐block, Bayesian inference is applied to construct an overall T2 statistic. To identify the faulty variables, a multi‐block PCA contribution plot is designed by choosing some specific blocks to highlight fault information. Finally, simulations on a numerical example and the benchmark Tennessee Eastman (TE) process are used to demonstrate the strengths of the proposed method.

Multimode Process Monitoring Method Based on Multiblock Projection Nonnegative Matrix Factorization

“…Some fault information are reflected in theX space, and some fault information are reflected in theX space. In order to monitor the fault more accurately, two monitoring statistics are usually monitored at the same time in many literatures [18][19][20][21][22][23]. N 2 lim and SPE lim are confidence limits of statistics N 2 and SPE.…”

“…Tong et al introduced an improved multiblock principal component analysis (MBPCA) algorithm to extract the specificity of each block and the block fraction of correlation between different blocks [21]. Wang and Deng considered the local variable behavior of process data and proposed a multiblock PCA process monitoring method based on variable weight information [22]. Wang et al proposed an adaptive partitioned nonnegative matrix factorization algorithm based on nonfixed subblock NMF model for fault monitoring in chemical processes [23].…”

Distributed process monitoring framework based on decomposed modified partial least squares

“…Therefore, the monitoring performance of large‐scale processes is degraded. In order to perform effective process monitoring for large‐scale processes, the distributed process monitoring framework is commonly used; this framework uses more local information and reduces the complexity of the model . In recent years, data‐driven approaches have been widely used in large‐scale plant‐wide processes, in which the most common approaches are multiblock PCA methods, multiblock PLS methods, and multi‐level modelling methods…”

“…In order to perform effective process monitoring for large-scale processes, the distributed process monitoring framework is commonly used; this framework uses more local information and reduces the complexity of the model. [4][5][6] In recent years, datadriven approaches have been widely used in large-scale plantwide processes, in which the most common approaches are multiblock PCA methods, multiblock PLS methods, and multilevel modelling methods. [7] Since product quality is an important economic indicator in industrial processes, quality-related fault detection, one of the monitoring tasks, has recently become a common research topic.…”

A Novel Decentralized Weighted ReliefF-PCA Method for Fault Detection

“…To solve the above problems, Ge and Song [16], Tong et al [17] made a block processing of variances from the perspective of data statistical property; Tong et al [18], Wang and Deng [19] made a weight processing of variances by using the relationships between variables. Deng and Deng [20], Xiaogang et al [21] proposed several strategies of integrating weighting.…”