Novel Distributed Alarm Visual Analysis Using Multicorrelation Block-Based PLS and Its Application to Online Root Cause Analysis

Abstract

Process monitoring and alarm systems play an important role in production safety. Distributed alarm systems have been utilized to effectively reduce the burden of data calculations, contributing to fast-track root causes of alarms. However, the traditional distributed contribution analysis method cannot be well applied to online monitoring in time. In addition, the correlation-based variable reconstruction method cannot accurately explain the actual process. To solve this problem, a novel distributed alarm analysis model based on multicorrelation blocks and an improved contribution graph is proposed. The proposed multicorrelation blocks are based on partial least squares. The computational burden is reduced, but accuracy is ensured by multicorrelation block partial least squares. In addition, an improved alarm analysis strategy using the contribution value is adopted, so that the results of the distributed contribution analysis method can be used for essential variable analysis online. Simulation results of a case study using the Tennessee Eastman process verify the effectiveness and practicability of the proposed method and indicate that the online analysis performance in terms of actual industrial process monitoring and root cause analysis is improved.

Integrated Diagnostic Framework for Process and Sensor Faults in Chemical Industry

“…However, the analysis method based on the kernel function does not need to calculate the eigenvector as the PCA method but to convert it into the eigenvalue and eigenvector of the kernel matrix. Thus, it avoids the calculation for obtaining the eigenvector in the high-dimensional space and converting it into projection, solving the linear combination of kernel functions, and by capturing the data dynamic matrix [ 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 ], it also solves the dynamic matching problem of the PCA model. Hence, the calculation is greatly simplified.…”

“…When the system detects that a fault has occurred, it needs to locate the current fault location and fault variables, and then determine whether the fault is a process fault or a sensor fault. Currently, the most widely used method for fault variable location is the contribution graph [ 26 ]. The method is based on quantifying the contribution of each process variable to a single principal component score.…”

Investigations of mixing and heat transfer in a structured tube‐in‐tube millireactor by numerical, experimental and statistical methods

“…PLS's excellent information extraction capability has been gradually applied to the process control of batch/semi-batch processes to ensure the final product quality. [38][39][40] However, there is still a blank in applying PLS in flow chemistry.…”

“…Partial least square regression (PLS) is one of the methods used for multivariate statistical analysis, which is regarded as the extension of principal component analysis and canonical. PLS's excellent information extraction capability has been gradually applied to the process control of batch/semi‐batch processes to ensure the final product quality 38–40 . However, there is still a blank in applying PLS in flow chemistry.…”

Online Distributed Process Monitoring and Alarm Analysis Using Novel Canonical Variate Analysis with Multicorrelation Blocks and Enhanced Contribution Plot

“…Block segmentation is an important step in the design of a distributed alarm system. − Process knowledge can be used to separate one system into blocks. In each subblock, latent variable space was built for independent alarm analysis .…”