Long-Term Load Forecasting Using System Type Neural Network Architecture

Abstract

This paper presents a methodology for long-term electric power demands using a semigroup based system-type neural network architecture. The assumption is that given enough data, the next year's loads can be predicted using only components from the previous few years. This methodology is applied to recent load data, and the next year's load data is satisfactorily forecasted. This method also provides a more in depth forecasted time interval than other methods that just predict the average or peak power demand in the interval.