An improved dynamic latent variable regression model for fault diagnosis and causal analysis

Abstract

The advancement of industrial techniques has imposed a high demand for powerful machine learning algorithms to model the increasingly complicated relations in the data. Among them, dynamic models are widely studied to capture the inevitable temporal relations. However, most existing methods only focus on the dynamics between input and output data, failing to exploit other valuable information in the output. In this article, an improved dynamic latent variable regression (LVR) method is proposed to capture both autocorrelations and cross-correlations between input and output with an autoregressive exogenous model, which is referred to as dynamic regularized LVR with auto-regressive exogenous input (DrLVR-ARX). Further, a DrLVR-ARXbased fault detection and diagnosis framework is designed to identify the root causes of a detected fault. The framework systematically integrates reconstruction-based contribution, time-domain Granger causality, and conditional spectral Granger causality to determine and locate the assignable causes.The effectiveness of the proposed algorithms is demonstrated with two industrial processes.

Variable Time Delay-Based Granger Causality Approach Integrated with Dynamic Coupling Analysis for Root Cause Diagnosis in Chemical Processes

“…Efficiently pinpointing the root cause and fault propagation path is crucial for effective maintenance, which enables operators to swiftly implement corrective actions and prevent potential major accidents resulting from fault propagation [2]. Therefore, RCD and FPPI have received increasing attention from both academia and industry [3], [4].…”