基于样条估计分位数回归的光伏功率回归模型

Photovoltaic Power Regression Model Based on Spline Estimation and Quantile Regression

光伏发电功率预测的准确与否是太阳能光伏发电是否能够有效地并入当前电网从而大大地提高太阳能利用率的关键.分位数回归是一种能够给出输出量的详细完整分布,从而便于分析与研究的回归模型.样条就是仅在节点处平滑连接的多项式函数,样条估计具有简单易行和计算速度快的优点.本文通过建立基于样条估计的分位数回归模型,在光伏面板发电功率数据的基础上,拟合光伏功率曲线,通过计算残差平方和和确定系数进行对拟合效果的评估.结果表明,该模型利用已有的光伏面板发电功率数据,可以在给出功率预测值的完整分布的同时,准确有效地分析相关因素对光伏发电功率的影响,展现不同分位点的回归拟合效果,从而有效地提高光伏系统对太阳能的利用率,避免光伏发电在接入电网时所产生的不利影响.

A quantile regression model was established based on spline estimation. Based on the output data of photovoltaic panel,the curve of power was fitted,which was compared with ordinary least squares regression model and simple quantile regression model. The results show that the quantile regression model can give much more accurate and effective analysis on the influence of relative factors on the output power of photovoltaic panels, improve the efficiency of solar energy, provide the complete distribution of output power,and reduce the negative impact of photovoltaic panel when connected to the grid.