Data Mining Applied to Customer Categorization Based on Load Profiling

Abstract

Load Profiling, a method where load consumption patterns of different electricity consumers are identified using the daily/monthly load curves is used in Distribution System planning activities like peak load management and time of use tariff. The load profiling identifies various customers with similar load patters and groups them into clusters. This method of customer categorization helps the utilities by eliminating the tiresome task to collect load information data of individual customers’ continuously over time and analyze each of them to be applied for planning activities. Instead with defined categories of customers exhibiting similar consumption patterns, the utilities can then efficiently plan the distribution of power without any difficulty. Efficient distribution system load data processing and analysis arises as one of the main concerns in Load Profiling. Data mining a process to derive interesting and intelligent knowledge from large databases and hence analyze the data to obtain patterns of similarity or dissimilarity can be employed for data processing and analysis. Thus this paper aims to utilize the Tadpole method in Dynamic Time Warping of Data Mining implemented in the R-tool to effectively process and analyze load data and hence obtain different categories of customers. This will in-turn aid in efficient distribution system planning with regards to demand side management programs.