Diagnosis of critical events in converter stations based on association-causality analysis

Abstract As the core hub of High-Voltage Direct Current (HVDC) transmission systems, converter stations rely on millisecond-level Sequential Event Records (SER) for fault diagnosis. However, SER data are characterized by massive volume, intricate event associations, and hidden causal relationships. Traditional operation and maintenance (O&M) diagnosis methods mainly depend on expert experience and statistical analysis of historical data, which are easily affected by redundant rules and confounding variables, making it difficult to effectively capture both event associations and causality. To address these challenges, this study proposes an Association–Causal–based diagnostic method for converter station O&M events. The proposed method integrates association rule mining to identify key event sets, introduces redundant rule pruning and confounding variable control to eliminate spurious causality, and finally employs OR-based statistical causal analysis to uncover potential causal chains among events. This paper integrates Association–Causal analysis with SER fault diagnosis and establishes a complete methodological framework based on Apriori-based frequent event mining, redundant rule pruning, confounder control, and OR-based causal statistics. The proposed approach significantly improves diagnostic accuracy and robustness.