基于多气象类型加权和改进高斯混合模型的光伏出力超短期概率预测

ULTRA-SHORT-TERM PROBABILITY PREDICTION OF PV OUTPUT BASED ON MULTI-WEATHER-TYPE-WEIGHTED AND IMPROVED GAUSSIAN MIXTURE MODEL

提出一种多气象类型加权和改进高斯混合模型,可实现对光伏出力提前15 min的超短期概率预测。首先,依据气象特征将历史数据划分为若干气象类型,然后,构建改进高斯混合模型获得每个气象类型出力数据的概率分布,其次,构建隶属度函数量化待预测时刻气象特征对于各气象类型的相似程度,最后,根据隶属度对各气象类型的概率分布加权。以实际光伏电站数据进行算例分析,结果表明相较于单一气象类型,多气象类型加权模型的MAPE平均减少16.87%,ACD平均提升10.45%,AW平均下降2.49%。

This paper proposes a PV output ultra-short term probability prediction method based on multi-weather-type-weighted and improved gaussian mixture model,which can realize ultra-short-term probability prediction of photovoltaic output 15 minutes ahead.Firstly,the historical data is divided into several meteorological types according to the meteorological characteristics.And then,the improved Gaussian mixture model is constructed to obtain the probability distribution of each meteorological type.Thirdly,the membership function is constructed to quantify the similarity degree of the meteorological characteristics to each meteorological type with the time to be predicted.Finally,the probability distribution of each meteorological type is weighted according to the membership.Taking the actual PV power station data as an example,the results show that compared with methods with a single meteorological type,the weighted model of multiple meteorological types has an average reduction of 16.87%in MAPE,an average increase of 10.45%in ACD,and an average decrease of 2.49%in AW.