基于贝叶斯概率的光伏出力组合预测方法

PHOTOVOLTAIC OUTPUT COMBINATION FORECASTING METHOD BASED ON BAYESIAN PROBABILITY

提出基于贝叶斯概率的光伏出力组合预测(CPBB)方法。根据光伏出力历史数据得到直接拟合模型(DFM)基础上,利用贝叶斯概率原理采用预测偏移率(POR)、预测调整率(PAR)以及预测误差的伪方差(PVPE)这3个指标对DFM模型修正,进而利用PSO优化算法加权寻优得到最终的光伏出力组合预测模型。最后,通过利用CPBB对合肥某光伏电站在不同的预测尺度下进行仿真验证,证明所提出的CPBB方法具有较好的普适性和较高的预测精度。

A combined prediction method of photovoltaic power output based on the Bayesian probability(CPBB)is proposed in this paper.According to the direct fitting model(DFM)coming from the historical data of photovoltaic power is modified according to the predictive offset rate(POR),the predictive adjustment rate(PAR)and the pseudo variance of prediction error(PVPE)based on the Bayesian probability,respectively.And the weighting coefficients of the CPBB model is obtained by the PSO algorithm.Under the three different time scales,the proposed CPBB model is verified with a photovoltaic power plant in Hefei.Through comparing with the traditional methods,the proposed CPBB has the better universality and higher prediction accuracy.