基于VMD-SE和优化支持向量机的光伏预测方法

Photovoltaic Prediction Method Based on VMD-SE and Optimized Support Vector Machine

针对光伏电站短期功率预测准确性的问题,提出了一种基于VMD-SE与改进的灰狼优化(Grey Wolf Optimization,GWO)算法优化支持向量机回归(Support Vector Regression,SVR)的组合预测方法。由于不同天气类型的光伏功率输出相差较大,因而利用相似日选取相同天气类型下的数据进行预测;考虑到光伏功率输出随机波动性较强,采用变分模态分解对原始光伏功率序列进行分解,以减少数据的非平稳性;为了克服支持向量回归参数盲目选取的弊端,利用改进的灰狼优化算法对其参数进行优化,以进一步提高数据的预测精度;最后,将分解后的子序列经样本熵重组后相加求和得到最终预测结果。算例结果表明,该组合预测方法整体上预测误差最小,有效提高了光伏输出功率预测的准确性,可以更好地保障电力系统的可靠运行。

Aiming at improving the short-term power prediction accuracy for photovoltaic station, a combination predic- tion method based on the VMD-SE and improved grey-wolf optimization algorithm to optimize the support vector regres- sion is proposed in this paper. Due to the great difference of photovoltaic power output in different weathers, the data of the same weather are selected by selecting the similar days. Taking into account the strong random fluctuation of photo- voltaic power output, the variational mode decomposition is used to decompose the original photovoltaic power sequence in order to reduce the non-stationary of the data. In order to overcome the disadvantage of blind selection of support vector regression parameters, the improved grey wolf optimization algorithm is applied to optimize the parameters to im- prove the prediction accuracy of the data. Finally, using sample entropy composed the subsequences and then summing them up, the final results are obtained. The results of a case shows that the combination forecasting method has the least overall prediction error, and it effectively improves the prediction accuracy for the PV output power, which can better guarantee the reliable operation of the power system.