Comparison of a Regime-Dependent Artificial Neural Network Method to a Regression Tree Method for Short-Range Solar Power Forecasting

Energy grid systems operators and utilities require accurate solar power forecasts to effectively balance the supply and demand of electricity due to the inherent variability in solar power production. The objective of a forecasting system is to model the actual relationships between the predictors and the predictand, and in the case of solar power forecasting, the relationship between the predictors and the predictand is frequently non-linear. Thus, two non-linear artificial intelligence prediction techniques are tested in this study. One method is a model tree that uses a “separate-and-conquer” algorithm to search for a rule that explains part of the training instances, separates these instances, builds a regression model on these instances, and continues this process until no instances remain. The prediction for a given instance is a weighted average of the multivariate linear regression equations at each node in the tree down to the final node where the instance was assigned. This method, which has inherent “regime classification,” is compared to an artificial intelligence method that uses an unsupervised approach to specifically classify regimes (k-means clustering) and then trains a supervised technique (artificial neural network) separately on each regime. The goal of this study is to compare the two methodologies for predicting short-range (15-min to 180-min) solar power, which has inherent weather regimes, or cloud types, that have varying levels of predictability and non-linear relationships between predictors and predictand.