SVM-based Capacity Prediction of Photovoltaic Generations in Regional Grids
Conventional methods of predicting PV generation capacity generally rely on historical data,and different ex-tents of data deviations from various data sources induced by differences in data collection and quality can lead to inaccurate prediction results.Therefore this study proposed a SVM-based prediction method for PV generation capacity in regional grids.First the PV generation data were preprocessed and divided via chain smoothing,and PV units were normalized to detect and extract abnormal power station data.The processed and extracted data were then input into SVM predictive model after gap filling,and finally the PV generation capacity prediction was achieved through the optimized model.The proposed predictive method achieved in the experiment a prediction accuracy of 92.37%,superior to those of methods based on modified firefly algorithm(85.43%)and grey correlation and sparrow optimization algorithm(74.66%),indica-ting a relatively accurate predictive performance and practical applicability.
SVMregional power gridphotovoltaic power generationgeneration capacity prediction
万方数据
维普
