基于样本扩张灰色关联分析的光伏出力预测

PV OUTPUT FORECAST BASED ON GREY CORRELATION ANALYSIS WITH EXPANDED SAMPLE

光伏出力准确预测是光伏并网安全运行的重要基础,样本容量增大、计及多影响因素能有效提高光伏出力预测精度。以小时段为单位建立一种基于样本扩张灰色关联分析的光伏发电短期出力预测模型,扩张有限的样本容量,能分析多因素影响。首先分析影响光伏出力的多种因素,通过灰色关联度分析的方法对样本进行分析,得到扩张最优相似小时段样本;通过遗传算法对BP神经网络的权值和阈值进行优化,并对神经网络进行训练;最后进行光伏出力预测。该文所建立的预测模型有效扩张了样本容量,提高了突变天气时预测准确度,有一定应用价值。

It is important to forecast PV short-term output accurately for the safety of grid operation with PV. Theincreases of sample number and multiple factors considered can improve the accuracy of prediction of PV outputeffectively. A PV short-term output forecast based on grey correlation analysis with expanded sample is proposed in formof hours,which expands the limited sample size and considers multiple factors. The best samples are chosen by analyzingsamples from the period of time to be predicted and samples to be trained through grey correlation analysis. A BP neuralnetwork,whose weight and threshold are optimized by genetic algorithm,is trained by using best samples. At last theoutput is predicted and compared with traditional forecast method. The results showed that the method proposed in thispaper not only expands the sample number,but also improves the forecast accuracy when climatic jump happens,whichhas some value for application.