PCA-based least squares support vector machines in week-ahead load forecasting

Week-Ahead load forecasting is essential in the planning activities of every electricity production and distribution company. This paper proposes the application of principal component analysis (PCA) to least squares support vector machines (LS-SVM) in a week-ahead load forecasting problem. New realistic features are added to better and more efficiently train the model. For instance, it was found that hours of daylight are influential in shaping the load profile. This is particularly important in case of cities that are situated in the northern hemisphere. To show the effectiveness, the introduced model is being trained and tested on the data of the historical load obtained from Ontario's Independent Electricity System Operator (IESO) for the Canadian metropolis, Toronto. Analysis of the experimental results proves that LS-SVM by feature extraction using PCA can achieve greater accuracy and faster speed than other models including LS-SVM without feature extraction and the popular feed forward back-propagation neural network (FFBP) model.