基于支持向量回归的PV/T组件温度实时预测

Real-time prediction of PV/T component temperature based on support vector regression

在光伏光热系统中,光伏板的发电效率与PV/T组件温度密切相关。实时、精确地预测PV/T组件温度,对优化控制决策、提高光伏板发电效率具有重要意义。文章利用支持向量回归(SVR)算法建立PV/T组件温度预测模型。为了提高该模型预测结果的精确度,采用网格搜索与交叉验证相结合的方法对SVR核函数参数g和惩罚因子c进行寻优;然后,结合实验平台的测量数据,划分训练集和预测集,并对原始数据进行归一化处理;最后,文章将基于SVR算法温度预测模型的预测结果与BP神经网络的预测结果进行对比。分析结果表明:基于SVR算法温度预测模型的预测值与实测值基本一致,该模型的预测精度和泛化性能均优于BP神经网络的预测结果。

In photovoltaic and photothermal system,the efficiency of photovoltaic panel power generation is closely related to the temperature of PV/T module.Accurate and real-time prediction of PV/T module temperature is of great significance for optimizing control decisions and improving power generation efficiency.In this paper,support vector regression(SVR)algorithm is used to model PV/T module temperature.In order to improve the accuracy of model prediction results,the parameter of SVR kernel function g and the penalty factors c are optimized by combining grid search with cross validation.Combined with the data collected from the experimental platform,the training set and prediction set are divided,and then the original data are normalized,a PV/T module temperature prediction model based on support vector regression is established.The analysis results show that the temperature prediction value of the model is basically consistent with the measured value,and its prediction accuracy and generalization performance are better than those of BP neural network.