Multistage Consistency-Constrained Dynamic Causal Learning for Evolution-Revealed Root Cause Analysis

doi:10.1109/TIM.2025.3644559

Multistage Consistency-Constrained Dynamic Causal Learning for Evolution-Revealed Root Cause Analysis

2025-01-01

0

PRE

AI

DOI：10.1109/TIM.2025.3644559

Original

Original request for help

Save

Abstract

En 中文

Exploring the causality among variables is crucial for uncovering the fault’s root cause. However, the existing methods are difficult to achieve a fine characterization of the causality evolutionary process. To this end, this article proposes a multistage consistency-constrained dynamic causal learning for root cause identification. First, the DenStream-based change-point detection (CPD-DenStream) is proposed to objectively describe the evolution process of faults, which lays the foundation for the following causal learning and root cause analysis (RCA). Then, the designed causal Transformer introduces a variable-level attention mechanism, which can accurately capture the causality among variables through continuous dynamic learning. Subsequently, the F-Diff causal loss and interaction redundancy-based dynamic causal learning are combined to guide the learning and adjustment of causality, making them more compatible throughout the entire evolution process of faults. Finally, the proposed method is evaluated on Tennessee Eastman process and the real-world coal mill dataset, achieving a minimum improvement of 8.7% in ${F}1$ -score, 9.8% in precision, and 8.4% in recall compared with baseline methods. These results demonstrate the superiority of the proposed method in capturing multistage causality, thereby advancing the practical implementation of RCA systems.

Keywords:

Dynamic causal

fault evolution

interaction redundancy

multistage consistency

root cause analysis (RCA)

Multistage Consistency-Constrained Dynamic Causal Learning for Evolution-Revealed Root Cause Analysis

Abstract

Journal

Organization