Data Quality Evaluation And Anomaly Detection Research Based On The Statistical Data Of Power System

With the development of intelligent power systems, power grids accumulated a largeamount of statistical data. Statistical data quality issues on abnormalities, redundancy andomission is becoming more and more prominent. At the same time, the quality of data, notonly affects the correctness and rationality of the power statistical analysis, but also directlyaffects the safety and reliability of power system operation. Therefore, how to improve thequality of electric statistical data and realize the high quality data service has become anindispensable part. This paper research the anomaly detection and data quality evaluationresearch of electric power statistical data.Firstly, based on the electric power statistical data analysis, an electric power statisticaldata evaluation model is established which realize statistical data quality evaluationquantization. And the data quality evaluation process as follow: first, determine theassessment data object; and then select the evaluation index according to the data qualityevaluation index of electric power demand, combined with statistical meaning to designevaluation rule. Then uses AHP to determine the weights of evaluation indexes, and to givethe index expected value to each of the evaluation, finally calculated the evaluation indexscore by the qualified percentage of each evaluation index. Through example analysis andproves that the electric power statistical data quality assessment method can scientifically andaccurate quantify the data quality.Secondly, in the electric power statistical data evaluation model built above, due to thedifferent period, different users in the process of statistical data quality assessment may differon the degree of each evaluation index. In order to distinguish the importance of differentevaluation index, use AHP to give weights for each indicator. Considering the relationshipbetween the evaluation indexes and form a system of class hierarchy, using the1-9ratio scalemethod built judgment matrix, and verify the consistency of judgment matrix. Finally theweights of each evaluation index is calculated, at the same time illustrate the evaluationprocess with the actual data.Finally, put forward the assessment method of correctness, which is the most importantevaluation indicators, we also can call it anomaly detection. We detect the outliers by threecases. One case is that the evaluation object is a single statistical indicator; one case is thatthere are direct logical relationships between statistical indicators; the third case is there areno direct logical relationships between statistical indicators. Comprehensive understand the situation of data anomaly to improve the abnormality detection accuracy of the results. Andstatistical data of a domestic network take an anomaly detection based on the method, theresults show that the detection method can detect the anomaly data scientifically and verifythe practicality and effectiveness of the detection method.This paper presents a statistical data anomaly detection method of power system and dataquality assessment models and methods, which can detect and assess electric power statisticaldata to provide technical support for operational status of the power system.