Research and application of a combined model based on frequent pattern growth algorithm and multi-objective optimization for solar radiation forecasting

Solar radiation forecasting plays a significant role in precisely designing solar energy systems and in the efficient management of solar energy plants. Most research only focuses on accuracy improvements; however, for an effective forecasting model, considering only accuracy or stability is inadequate. To solve this problem, a combined model based on nondominated sorting-based multiobjective bat algorithm (NSMOBA) is developed for the optimization of weight coefficients of each model to achieve high accuracy and stability results simultaneously. In addition, a statistical method and data mining-based approach are used to determine the input variables for constructing the combined model. Monthly average solar radiation and meteorological variables from six datasets in the U.S. collected for case studies were used to assess the comprehensive performance (both in accuracy and stability) of the proposed combined model. The simulation in four experiments demonstrated the following: (a) the proposed combined model is suitable for providing accurate and stable solar radiation forecasting; (b) the combined model exhibits a more competitive forecasting performance than the individual models by using the advantage of each model; (c) the NSMOBA is an efficient algorithm for providing accurate forecasting results and improving the stability where the single bat algorithm is insufficient.