Short-Term Load Forecasting Using Fuzzy Logic

Abstract

In this paper, fuzzy logic (FL) is applied to the problem of short-term load forecasting (next day) in electrical power systems. To achieve this, it is necessary to select the historical data to be used and pre-process them using the c-means method, grouping them according to power levels (MW) to define the number of membership functions (MFs) to the fuzzy system, which is very important for the calculation of the lowest forecast error; finally, the historical data are entered into the fuzzy system implemented in MATLAB. This methodology is applied to predict the daily electrical load (demand) of the Peruvian Electrical System using the historical data of the actual demand executed for the study period and by calculating the MAPE error. It is shown that the FL offers better results than the conventional methodology for the forecast of the electrical load.