A Reciprocal Information Entropy Causal Inference Method for Exploring the Cause-effect Relationship in Power System Operation Data

With the massive power system operation data acquisition, the data-based analysis method plays an increasingly important role in operational analysis of power systems. Most of the existing data-based analysis methods mainly focus on correlations between data sequences. Any two variables are mutually correlated. But the asymmetric causality is generally presented between two correlated variables. Revealing the cause-effect relationship between operation variables can bring us more insight into the operation regularity of power system. Many achievements have been made in causal inference studies in recent years. In this paper, the asymmetrical cause-effect relationships in correlated variables of power systems were revealed physically. A new causal inference method-reciprocal information entropy causal inference (RIECI) method was proposed. The causal strength could be correctly reflected by the proposed causal index of RIECI. The validation in power system examples shows that the RIECI method can effectively reveal the cause-effect relationships in power system operation data. Comparing with the mutual correlations, the cause-effect relationship is helpful for soundly recognizing the operation mechanism and correctly regulating the operation state of the power systems based on the operation data. © 2022 Chin.Soc.for Elec.Eng.