考虑空间相关性的分布式光伏发电出力预测及误差评价指标研究

对于待预测的分布式光伏电站,基于已提出的大规模区域光伏分群方法,提出了筛选良好空间相关性光伏电站群的光伏发电出力预测方法。首先,对待预测电站的出力数据进行了天气类型划分;其次,选择与待预测电站具有相关关系的光伏电站作为相关性从站,并采取ARIMA模型识别待预测电站与从站之间的时间、空间关系,继而对待预测电站的出力进行预测;然后,通过多种预测误差指标对比,提出了更符合光伏预测的误差评价指标,即引用误差,以突显高功率输出的预测精度;最后,通过典型电站以及整个区域里所有分布式光伏用户的滚动预测和误差分析,证明了所提方法的普遍适用性。

Distributed Photovoltaic Power Generation Output Prediction Based on Spatial Correlation and Error Evaluation Indexes

Based on the proposed large-scale regional PV clustering method, this paper proposes a photovoltaic(PV) power generation output power prediction method screening PV power plant with spatial correlation for the distributed PV power stations to be predicted. Firstly, the output power data of the power station to be predicted is classified according to the weather types;secondly, a photovoltaic power station correlated to power stations to be predicted is selected as a slave station, and ARIMA model is employed to identify the time and spatial relationship between the station to be predicted and the slave station, and predict the output power of the station to be predicted;thirdly, by comparison of various prediction error indexes, an error evaluation index more suitable for PV prediction, namely the quoted error, is proposed to highlight the prediction precision of high-power output. Finally, the universal applicability of the method is proved by rolling forecast and error analysis of all PV consumers in typical substation and all the regions.