Multimode Process Monitoring Using Variational Bayesian Inference and Canonical Correlation Analysis

Distributed process monitoring based on canonical correlation analysis with partly-connected topology

A multimode process possesses dynamic behaviors, especially in between-mode transitions. Motivated by that dynamic models are well-suited for interpretation of process dynamics, we initially propose a dynamic model-level method for multimode process monitoring to detect abnormal operations quickly and accurately. It mainly includes three parts. First, a piecewise affine (PWA) model is identified offline to represent different dynamic patterns of a multimode process, which can avoid model redundancy compared with existing achievements. Owing to the utilization of the PWA model, the proposed method is still valid when unknown transitions occur. Second, in online monitoring, local autoregressive exogenous (ARX) models, obtained by just-in-time learning (JITL), can sufficiently extract local dynamic patterns so as to facilitate evaluating dynamic changes in the process. Finally, a novel measurement, <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\omega $ </tex-math></inline-formula> -metric, is devised to analyze the dynamic variations in the process by directly measuring the similarity between local models and the PWA model, which is more beneficial for detecting abnormal operations than data distances. The superiority of the proposed method is validated by a continuous stirred tank reactor (CSTR) and a real air separation process compared with the existing multimode process monitoring approaches.

Modern industrial processes increasingly prioritize demands for safety and reliability, spurring substantial research on process monitoring models. Among existing research subjects, concurrent multimode operating conditions are vital for effective process monitoring. This work proposes an efficient dimensionality‐reducing Gaussian mixture‐based reconstruction approach for multimode industrial process monitoring. The t‐SNE method is first employed to transform high‐dimensional data into a lower‐dimensional space that retains critical operational information. Using these reduced dimensions, a robust Gaussian mixture model is established to partition the operation data into different modes. Furthermore, the original data are assigned to the corresponding operating modes, and local variational autoencoder (VAE) reconstruction models are established, respectively. For each VAE model, two statistics are designed, termed and , to detect abnormalities. The proposed method is applied to a three‐phase flow facility, and the superiority over the comparison methods is proved.

Enhanced robust multimode process monitoring under dirty data via difference-based decomposition of matrix

Spatio-temporal information analytics based performance-driven industrial process monitoring framework with cloud–edge-device collaboration

Multimode Process Monitoring and Fault Diagnosis Based on Tensor Decomposition

Previous chapter Next chapter Full AccessProceedings Proceedings of the 2016 SIAM International Conference on Data Mining (SDM)GSpartan: a Geospatio-Temporal Multi-task Learning Framework for Multi-location PredictionJianpeng Xu, Pang-Ning Tan, Lifeng Luo, and Jiayu ZhouJianpeng Xu, Pang-Ning Tan, Lifeng Luo, and Jiayu Zhoupp.657 - 665Chapter DOI:https://doi.org/10.1137/1.9781611974348.74PDFBibTexSections ToolsAdd to favoritesExport CitationTrack CitationsEmail SectionsAboutAbstract This paper presents a novel geospatio-temporal prediction framework called GSpartan to simultaneously build local regression models at multiple locations. The framework assumes that the local models share a common, low-rank representation, which makes them amenable to multi-task learning. GSpartan learns a set of base models to capture the spatio-temporal variabilities of the data and represents each local model as a linear combination of the base models. A graph Laplacian regularization is used to enforce constraints on the local models based on their spatial autocorrelation. We also introduce sparsity-inducing norms to perform feature selection for the base models and model selection for the local models. Experimental results using historical climate data from 37 weather stations showed that, on average, GSpartan outperforms single-task learning and other existing multi-task learning methods in more than 65% of the stations, which increases to 81% when there are fewer training examples. Previous chapter Next chapter RelatedDetails Published:2016eISBN:978-1-61197-434-8 https://doi.org/10.1137/1.9781611974348Book Series Name:ProceedingsBook Code:PRDT16Book Pages:1-867

Canonical correlation analysis is a useful multivariate statistical method to assess linear relationships between groups of variables. In this paper, we first build a regional science and technology (S&T) input and output indicators system. Then we take the regional S&T of Jiangsu Province, China, as a typical case study. Second, canonical correlation analysis is used on two group variables, the regional S&T input and output indicators, with the data derived from thirteen cities of Jiangsu Province, and the sample period is from 1996 to 2005. Furthermore, we analyze the S&T input and output relationship by extracting two pairs of canonical variables, and reveal the internal links of dominant indicators between the two group variables. The results would give us better understanding of S&T input and output relationship. Finally, we discussed the S&T policy implications for Jiangsu Province based on the quantitative results.

Multimode process identification and monitoring based on hierarchical fluctuation window strategy

The local linear modelling approach is capable of modelling complex nonlinear processes. The transparency of the model architecture permits the incorporation of prior knowledge in different ways and the integration of process knowledge and experimental data has been demonstrated in detail for the modelling of the heat exchanger. Utilising an arbitrary local modelling strategy does not automatically guarantee a transparent model. The employed LOLIMOT algorithm has proven to yield transparent and well interpretable models. The appropriate choice of an excitation signal is vital for a good model. Here, the transparency of the local structure again aids the engineer in the design and the assessment of the excitation signal. The interpretability of the local structure offers simple ways for the validation of the identified model based on local characteristic values as process gains or time constants. The obtained local linear models can further be utilised in model-based control approaches as well as process monitoring or fault detection.

Key principal components with recursive local outlier factor for multimode chemical process monitoring

Canonical correlation analysis (CCA) has attracted increasing attention in the field of fault detection because it provides an effective way to explore the relationship between the input and output data. This paper develops a novel rank constrained CCA (RCCCA) framework, which is the first approach that takes the rank prior information into consideration. Technically, the rank constrained optimization is able to capture the global structures of variables, and thus improve the performance of fault detection. In order to solve RCCCA, an alternating minimization algorithm is designed, which aims to preserve the maximum correlation with the low-rank learning. A fault detection residual is then generated, and the test statistic is constructed to determine whether a fault occurs. The RCCCA-based fault detection is finally tested on a numerical example and the Tennessee Eastman benchmark process. Monitoring results indicate the efficiency and feasibility of the proposed method.

A new modeling and monitoring approach for multi-mode processes is proposed. The method of similarity measure(SM) and kernel principal component analysis (KPCA) are integrated to construct SM-KPCA monitoring scheme, where SM method serves as the separation of common subspace and specific subspace. Compared with the traditional methods, the main contributions of this work are: 1) SM consisted of two measures of distance and angle to accommodate process characters. The different monitoring effect involves putting on the different weight, which would simplify the monitoring model structure and enhance its reliability and robustness. 2) The proposed method can be used to find faults by the common space and judge which mode the fault belongs to by the specific subspace. Results of algorithm analysis and fault detection experiments indicate the validity and practicability of the presented method.

Hidden Markov model-based approach for multimode process monitoring

Due to the existence of multiple operating modes, traditional fault detection techniques are ill-suited for complex industrial processes. Although there are more and more literature studies concerning this problem, only a few of them are based on the hidden Markov model (HMM). However, there is no exploration to concern the unknown mode in the industrial process based on it. This article proposes a novel process monitoring approach based on moving window HMM (MVHMM) for real-time multimode process monitoring with unknown mode. First, a hidden Markov model is built by training set. Instead of just considering the posterior probability of one single sample, the moving window is introduced to utilize the independence of samples for improving the accuracy of online mode identification. In addition, an MVHMM-based threshold statistic is defined to identify the unknown mode. Also, various known modes which include stable modes and transitions are separated on the basis of the Viterbi algorithm. Second, a new mo...