Research on Abnormal Data Evaluation and Repair Method Based on TE-WAF Algorithm

In power systems, the accuracy, completeness and timeliness of data are crucial for equipment monitoring, fault diagnosis and decision support. However, outliers often appear in power data due to a variety of factors, such as equipment failure, communication outages, or external interference. These abnormal data may not only lead to wrong analysis results, but also affect the safety and reliability of the power system. Therefore, in view of the inaccuracy, incompleteness and untimely outliers in power data, this paper proposes an abnormal data evaluation and repair method based on TE-WAF algorithm. In this method, a comprehensive evaluation model of abnormal data is proposed by combining various identification models. Considering the characteristics of the three types of abnormal data, this paper evaluates the data in the order of incomplete, untimely and inaccurate data, and uses confusion matrix to evaluate the accuracy rate to verify the effect of the model. Then, weighted mean correction model and ARIMA prediction model are used to classify and correct the labeled abnormal data. By combining the smoothing characteristics of the former and the time series prediction ability of the latter, different types of abnormal data can be corrected more comprehensively, the limitations of a single model can be avoided, and the advantages of the two models can be maximized to improve the overall correction effect of abnormal data. Better data quality to support subsequent analysis and decision making. The experimental results show that the Triple evaluationweighted ARIMA Fusion (TE-WAF) algorithm proposed in this paper can effectively improve the Evaluation and correction effect of power data outliers, so as to provide reliable data guarantee for the stable operation of power system.