Comparison of different approaches to short-term load forecasting

This paper presents three self learning techniques for short term load forecasting. These techniques are based on Adaptive Kalman Filter (AKF), Artificial Neural Networks (ANNs), and Knowledge Based Expert Systems (KBES). Actual load data obtained from a local utility is used to show the principles of these techniques. A comparison of the results of the three methods is presented along with the typical estimation error associated with each method in the case of short term load forecasting, i.e. both hourly (one hour ahead) and daily (twenty four hours ahead) load forecasting.