Short-term load forecasting: Similar day-based wavelet neural networks

In deregulated electricity markets, short term load forecasting is important for reliable power system operations, and significantly affects market participants. It is difficult and challenging in view of the complicated effects on load by a variety of factors. To appropriately capture the complex features of load, this paper presents a novel similar day-based wavelet neural network method. The key idea is to use a similar day technique to select good input load, use wavelet to decompose the load into low and high frequency components, and then use separate neural networks to predict the different frequency components. Factors affecting these frequency components are identified. Numerical testing shows that our method significantly improves prediction accuracy.