Identification of abnormal data of substation line loss under nonparametric kernel density estimation
A nonparametric kernel density estimation method for identifying abnormal data in substation line loss systems is proposed to address the issue of accessing massive amounts of data,which makes it difficult to accurately identify abnormal data.Construct station area line loss anomaly data model based on Gaussian kernel,and combine it with nonparametric kernel density estimation method to analyze the positive and negative"rank sum"characteristics of the anomaly data.Calculate the maximum and minimum values of historical electricity data at the same time,and obtain the upper and lower bounds of the historical data domain window width.By combining the two-dimensional wavelet denoising method,the denoising threshold's own characteristics are processed with a semi soft threshold function to achieve denoising of line loss data in the power grid substation area.Design a nonparametric kernel density estimation waiting curve to identify abnormal line loss data.According to the experimental results,this method identified 2 abnormal line loss data in the summer and 9 abnormal line loss data in the winter,which is consistent with the actual number of abnormal line loss data and has a precise identification effect.Based on the identified abnormal data of line loss in the substation area,the power supply lines of the distribution network can be optimized and adjusted to reduce line loss and improve power supply quality.
nonparametric kernel densitysubstation line lossidentification of abnormal datarank sum
NETL
NSTL
万方数据
