提升光伏发电预测的准确性有助于电网的调度管理和经济运行，其关键在于建立高效适用的预测模型。首先采用皮尔逊（Person）相关系数法对影响光伏出力的主要因素进行筛选，确立预测模型的输入特征向量；提出了一种融合Person相关系数法与遗传算法（GA）优化的ELM混合预测模型，并对预测模型中随机生成的参数进行了优化；以某一光伏电站的历史数据为研究对象，采用GA-ELM预测模型对预处理后的数据进行训练和测试，基于模型开展了四个季节典型日的光伏发电功率预测。结果表明：混合预测模型比单一的ELM预测模型和Person相关系数与ELM混合的预测模型的预估偏差率分别降低了19.2%和4.3%，验证了本文模型具有更高的准确性和稳定性。

Improving the accuracy of photovoltaic power generation forecast is helpful to the dispatching management and economic operation of power grid,and the key is to establish an efficient and applicable forecast model. Firstly,Pearson correlation coefficient method was used to choose the main factors affecting the photovoltaic output,so as to establish the input feature vector of the prediction model.Then,a hybrid ELM prediction model optimized by Pearson correlation coefficient method and genetic algorithm (GA) was proposed to improve the optimization of randomly generated parameters in the prediction model. Taking the historical data of a photovoltaic power station as the research object,GAELM prediction model was used to train and test the preprocessed data. Finally,based on the prediction model,the PV power of typical days in four seasons was predicted. The results show that the prediction deviation rate of the hybrid prediction model proposed in this paper is 19.2% and 4.3% lower than those of the single ELM prediction model and the prediction model with a mixture of Pearson correlation coefficient and ELM,respectively,using the actual output curve as the reference,thus verifying that the model in this paper has higher accuracy and stability.

TK519

南京工业大学 能源科学与工程学院，江苏 南京 211816