Intelligent temporal synchronization enhanced cointegration framework for non-stationary industrial process monitoring

Document Type

Article

Publication Date

3-1-2026

Abstract

Ensuring safe and reliable operation constitutes a fundamental requirement in modern industrial production, which can be achieved through effective process monitoring capable of identifying abnormal deviations from normal operating behavior. However, conventional data-driven multivariate statistical process monitoring (MSPM) methods rely on the assumption that the process is stationary, which is often violated in practice due to operational changes, stochastic disturbances, and external factors. Cointegration analysis (CA) offers a promising solution for this issue but its effectiveness is limited by variable time delays (VTD) caused by sensor latency, signal transmission, and spatially distributed equipment. To address this challenge, an intelligent temporal synchronization enhanced cointegration (ITSEC) framework for non-stationary industrial process monitoring is proposed, which integrates two key components. First, a clustering assisted time lags identification strategy is developed to intelligently synchronize the relevant variables. Then, synchronized time series is applied to improve the identification of long-term equilibrium relationships extracted by CA. The effectiveness of ITSEC is validated through a numerical simulation case and two real industrial cases. Comparative results show that ITSEC achieves a superior balance between high fault detection rates and low false alarm rates, proving its effectiveness and potential for practical deployment in complex industrial environments.

Publication Source (Journal or Book title)

Process Safety and Environmental Protection

This document is currently not available here.

Share

COinS