| In the context of smart grid construction,large-scale grid interconnection and lower cost of data collection as well as the increasingly wide variety of accessible data have made the trends of big data in smart grid increasingly apparent.With the development of big data in smart grid,power supply agents are able to have positive interaction with users.Big data can be used to analyze the constantly changing load characteristics.According to the power utility characteristics of every user,data mining technology can be fully used to analyze the power utility potential of users in terms of peak load shifting and to reduce the operation risk and cost of power system.In this paper,grid in Shanghai Pudong is chosen as the research object and outlier data in terms of load and relevant data are processed.First,monthly and daily load characteristics in terms of the overall load in this area are analyzed,in order to understand the regular patterns of power utility in this area.And annual load rate of1044 transformer areas are analyzed.Second,based on the analysis of load characteristics,regression analysis is used to quantitatively analyze the influences that temperature has on the overall load of all the days and weekdays in this area.Third,K-means clustering algorithm is used to reasonably categorize the daily load of typical days in summer in 12729 transformer areas and transformer areas with identical or similar power utility mode go into the same category.The power characteristics of every category clustering center is put forward.Last,a regression-based baseline load prediction model is put forward that uses a time-of-week indicator variable and a piecewise linear and continuous outdoor air temperature,in order to calculate the 15-min-interval baseline load data of the transformer areas.The method could characterize facility electricity loads and demand response behavior.At the same time,the method could be translated into easy-to-use tools for users. |
PDF Full Text Request