An Identification and Correction Method of Abnormal Data of Electricity Market Consumers Based on Nonparametric Kernel Density Estimation

With the gradual development of electricity spot markets in China, the settlement link of transactions in electricity market makes a higher request for the accuracy of power users’ electricity consumption data. Firstly, to cope with the deficiency and abnormity of time-of-use type of electrical quantity data caused by metering device and so on, the nonparametric kernel density estimation (KDE) method was adopted to optimally select the window width, and according to the historical electric quantity data of users in electricity market their daily time-of-use type of electrical quantity characteristic curves was extracted. Secondly, combining the extracted characteristic curves with the information contained in the historical electric quantity, the feasible domain matrix of users’ electric quantity data was obtained and applied to the identification of abnormal data. Thirdly, according to the magnitude of consumed electric quantity of the abnormal data interval, the data from scaled characteristic curves was taken as the modification value of continuous defective data. Finally, the effectiveness and veracity of the proposed method were verified by use of the data of the metered electric quantity of the consumers in the electricity market of a certain province. The results of the computing example show that the abnormal data can be effectively processed by the proposed method, and comparing with other methods, the proposed method possesses the highest accuracy in the correction process of continuous multi-point electric quantity data anomalies, thus the proposed method could apply in practice.