Practical application of object oriented techniques to designing neural networks for short-term electric load forecasting

This paper illustrates the use of object oriented programming (OOP) techniques for the design of neural networks (NNs) for short-term load forecasting. A load forecasting model has been developed using a multilayer perceptron NN with an appropriately modified backpropagation learning algorithm. The model produces a simultaneous forecast of the load in the 24 hours of the forecast day concerned. The technique has been tested on data provided by the Italian Power Company ENEL and the promising results obtained through the application of OOPNN-based approach show the effectiveness of this new approach.