基于改进LSSVM的太阳辐射量预测

为准确预测太阳辐射量,提出一种基于变分模态分解和粒子群优化算法的最小二乘支持向量机组合预测模型。针对太阳辐射量序列具有不稳定性的特点,首先利用变分模态分解将历史太阳辐射量数据分解成一系列相对稳定的分量序列,再应用粒子群优化最小二乘支持向量机参数,以预测各分量序列,将各分量太阳辐射量预测值集成,从而得到最终太阳辐射量预测值。实例分析和对比研究表明,该模型预测太阳辐射量有效可行,具有较高的预测精度。研究成果可为太阳辐射量预测提供参考。

Prediction of Solar Irradiation Based on Improved LSSVM Model

This paper proposes a combined forecasting model based variational mode decomposition (VMD) and least squares support vector machine (LSSVM) to predict the solar irradiation. According to the instability of solar radiation series, the historical solar irradiation data was first decomposed into a series of relatively stable components by the VMD. And then the parameters of the LSSVM prediction model were optimized by PSO, and it was further adopted to predict each component of solar irradiation series. Finally, the results of each component forecasting were superimposed to obtain the final forecasting values of solar irradiation. The case study and comparative analysis show that the proposed model is effective for prediction of solar irradiation and it has higher prediction accuracy. The research provides reference for solar irradiation forecasting.