A Hybrid Forecasting Approach for Short-Term Photovoltaic Power Generation Based on WOACO and LSSVM
Tianyu Chen *
School of Electric Power Engineering, School of Shenguorong, Nanjing Institute of Technology, Nanjing 211167, China.
*Author to whom correspondence should be addressed.
Abstract
Due to significant random volatility of photovoltaic power generation, a short-term prediction method for photovoltaic power generation based on Chaos Whale Algorithm (WOACO) optimized Least Squares Support Vector Machine (LSSVM) was designed to further improve prediction accuracy of photovoltaic power generation. Firstly, meteorological factors that play a crucial role in photovoltaic power generation are extracted through Pearson correlation coefficient; Secondly, using WOACO to optimize parameters of LSSVM prediction model; Finally, simulate and analyze actual photovoltaic power generation in a certain region in northwest China. Whether in summer or winter, the prediction accuracy of the prediction model has significantly improved under three different evaluation indicators. Results demonstrate that WOACO-LSSVM combination model has higher accuracy in predicting photovoltaic power generation.
Keywords: Photovoltaic power forecasting, least squares support vector machine, whale optimization algorithm, Pearson correlation coefficient, combination model