A New Method of Discovering Intrinsic Rules in the Operation of Large Power Grid

This article introduces a new method for discovering intrinsic rules in large-scale power grids, including feature extraction, causal association analysis, and interactive visualization. First, a hybrid feature selection method (ReliefF-CUTS+) is designed, combining ReliefF with CUTS+(Cau<tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$s$</tex>al discovery from irregUlar Time-Series +) to extract key features while preserving causal relationships. Second, an optimized association rule mining approach is developed, incorporating temporal state encoding and interval dynamic partitioning to enhance the efficiency of FP-Growth algorithm by constraining rule consequents. Finally, a dynamic visualization system with color encoding is constructed to intuitively display feature evolution patterns and causal interactions through interactive charts (e.g., dynamic histograms and ring-shaped correlation graphs). This method forms a closed loop from feature extraction, rule mining to visualization, which has both theoretical innovation and engineering practicability, enhances users' understanding of system state and association rules, and improves decision-making efficiency.