HETEROGENEOUS ARTIFICIAL NEURAL-NETWORK FOR SHORT-TERM ELECTRICAL LOAD FORECASTING

Short term electrical load forecasting is a topic of major interest fo r the planning of energy production and distribution. The use of artif icial neural networks has been demonstrated as a valid alternative to classical statistical methods in term of accuracy of results. However a common architecture able to forecast the load in different geographi cal regions, showing different load shape and climate characteristics, is still missing. In this paper we discuss a heterogeneous neural net work architecture composed of an unsupervised part, namely a neural ga s, which is used to analyze the process in sub models finding local fe atures in the data and suggesting regression variables, and a supervis ed one, a multilayer perceptron, which performs the approximation of t he underlying function. The resulting outputs are then summed by a wei ghted fuzzy average, allowing a smooth transition between sub models. The effectiveness of the proposed architecture is demonstrated by two days ahead load forecasting of EOS power system sub areas, correspondi ng to five different geographical regions, and of its total electrical load.