Short-term Electrical Load Prediction Using Evolving Neural Network

Abstract

A short-term electrical load prediction can use an artificial neural network approach. In this paper, an optimized neural network, namely Evolving Neural Network (ENN) has been developed for short term electric load prediction. ENN uses a genetic algorithm to optimize the weighting of neural networks. After the feedforward algorithm, the process continues with optimization, instead of learning process normally applied to the neural network. The proposed algorithm is implemented in MatLab. Data from April 2010 to April 2011 will be used as training data and data in May 2011 will be used as checking data. To evaluate the performance of Evolving Neural Network, Wavelet Neural Network (WNN) is also involved for comparison. The evaluation is conducted by observing the prediction results. Performance measurements are performed by observing errors that occur. The smaller the error value, the better the accuracy. The experimental result shows that the accuracy performance of ENN is better than WNN.