Search for the Single-type Load Schedules of the Power Facility

Abstract

One of the major necessities when planning and analyzing the operating modes of a power system is the forecasting of load rates and schedules. This paper presents the algorithm of searching for the single-type load schedules based on data mining methods. Averaged half-hour schedules of active and reactive power consumption were used as initial data. The proposed algorithm is primarily aimed at preliminary processing and grouping of the measurements when solving the task of identifying static voltage characteristics of load and also allows carrying out researches in the field of searching for single-type load curves to perform design and long-term calculations. The result of data processing is an array of labels for the daily schedule membership in a particular group. The proposed algorithm is based on the k-means method. The algorithm is tested on the real data collected from the substation feeding the assembly plant. In case the initial data is the sample of load schedules for several years, it is possible to set the task of forecasting the configuration of the load schedule. The initial data to solve this task will be a time ordered array of labels for the daily load schedule membership in a particular type.