Technology for Power Outage Research and Judgment-dependent Data Feature Noise Analysis

INTRODUCTION: Power grid blackouts occur frequently, which significantly impacts social impact. Because these accidents are dynamic and random, predicting and evaluating them is challenging. OBJECTIVES: To explore the complexity of the power grid itself, analyzes the critical changes of the self-organizing model during power grid fault, extracts the data characteristics related to the steady-state maintenance of abnormal systems, and puts forward an effective outage prediction model. METHODS: Starting with cluster analysis, The authors can reduce data fluctuation and eliminate noise interference to optimize data. The evaluation indexes of initial fault occurrence possibility and fault propagation speed in the power grid are constructed. RESULTS: The validation of the outage forecasting model has produced promising results, achieving 96.4% forecasting accuracy and a meager error rate. In addition, the evaluation index developed in this study accurately reflects the possibility and spread speed of power outage accidents. CONCLUSION: The research proves the feasibility of establishing an outage prediction model based on the power grid system data characteristics. The model has high accuracy and reliability and is a valuable tool for power outage research and judgment.