Short Term Weather -Dependent Load Forecasting using Fuzzy Logic Technique.

Electric load forecasting has become a very decisive in power system studies for its systematic functioning. Precise short term load forecasting aids the electric utilities for the determination regarding unit commitment, reducing spinning reserve capacity, maintenance schedules and other optimal energy exchange plans properly. Conventional methods suffer from the problem of complexity of the estimation procedure and substantial database support requirements. In this paper, a feasibility study of the implementation of fuzzy logic model for short term load forecasting is carried out. The proposed methodology uses fuzzy reasoning decision rules that capture the nonlinear relationships between inputs and outputs. The input data includes historical load (hourly & daily), physical variables hourly data like temperature, humidity and wind speed. The proposed fuzzy based logical model has been used to estimate the daily peak load forecasts over one week period comprising four working days and two days of weekend. The model is tested with two different set of membership functions namely triangular and trapezoidal with different amounts of function overlapping on the actual data obtained from the state load dispatch centre. Test results for daily peak load forecasts based on historical data indicate that the generated forecasts is quite similar in accuracy to more complicated methods and a mean absolute percentage error less than 3% is reported.