Neural network-based short-term load forecasting for unit commitment scheduling.

摘要

Today's electricity power industry is experiencing many fundamental changes under the process of deregulation. The power system operation will become more competitive in the new market environment. The accuracy of load forecast is crucial due to its direct influence on generation planning. The objective of this thesis is to develop an Artificial Neural Network-based (ANN) program to perform short-term load forecast (STLF) for a utility company. The program developed in this study is designed for automatic operation with the Energy Management System (EMS). Neural network structures are carefully adjusted to work with load characteristics of Western Farmers Electric Cooperative (WFEC). The forecasting result indicates that ANN forecaster produces more accurate results compared to the conventional adaptive regression based load model and can be modified to satisfy the real time operating requirements.