Research On Spatial Load Forecasting Method Based On Kernel Density Estimation

Spatial load forecasting(SLF)is a prediction of spatio-temporal characteristics of electric load in planning area.It is a complex problem involving many factors and high uncertainty.It not only predicts the future load,but also considers the spatial distribution of future loads.space power load forecasting is fundamental to the planning and design of power system.Therefore,high-precision spatial load forecasting results are of great significance to the planning and design of urban power network.In the field of SLF,the research on the prediction method has been the main task.Therefore,the influence of preprocessing of load data on the prediction accuracy is neglected.At the same time,the rigorous elaboration of the method of preprocessing the load data is lacking.In the process of SLF,the maximum load of the cell load in a certain period of time is generally used.However,due to many accidental factors,it will have an impact on the maximum cell load,if you directly use the affected cell load maximum performing SLF will lead to a decrease in prediction accuracy.Therefore,the study of load data preprocessing in the SLF is also very important.Firstly,this paper introduces the research status of SLF and data preprocessing in detail;Second,in the power geographic information system environment,cells are generated based on the power supply range and the cell load is obtained,based on the analysis of the fluctuation characteristics and differences of each cell load,all the abnormal data in the cell load are divided into four categories,and establish a classification index system with cell load with abnormal data;Again,the kernel density estimation theory is introduced into the SLF and a kernel density estimation method for determining the maximum cell load is proposed,this method calculates the nuclear density estimation curve of the cell load with abnormal data,and proposes the idea of obtaining a reasonable maximum value of the cell load by truncating the nuclear density to estimate the tail area of the curve,the computational models of the follow-up thresholds for truncating the various types of cell load nuclear density estimation curves containing abnormal data are also constructed,and follow-up truncations are implemented,and we analyzed the cell load of the class of abnormal data with load transfer;In addition,aiming at the problem of unclear display of low-dimensional space cell load data features,combined with the theory of phase space reconstruction,the low-dimensional data is mapped to high-dimensional space to better reflect the various types of cell load with abnormal data feature,with a density-based spatial clustering of applications with noise(DBSCAN),the reconstruction results are clustered,and the abnormal data is identified and rejected according to the clustering results,then a methodfor determining the maximum cell load based on phase space reconstruction and DBSCAN is proposed;Finally,the maximum cellular load determined by the two methods is predicted by the traditional SLF method.The validity of the two methods is verified by an example analysis,and the maximum cell load determined by the kernel density estimation method is more reasonable.