基于数据挖掘的神经网络光伏发电预测研究

Research of neural network power forecasting model of photovoltaic system based on data mining method

在建立神经网络光伏发电预测模型时,可选择的输入变量较多,其中部分是重复或者与发电量密切度不高,引入这类变量会降低预测精度,因此需要选择合适的输入变量。为了提高神经网络光伏发电预测的精度,该研究首先应用数据挖掘分析可输入变量与预测日发电量的相关系数,进而建立神经网络模型,在分析比较的基础上最终确定合适的输入变量。研究结果表明合理选择输入变量,可有效提高神经网络光伏发电系统的预测性能,有利于减少光伏发电随机性对电力系统的影响。

There are many input variables to choose while building the neural network power forecasting model of the photovoltaic power system.Since some of these input variables are repeated or do not have close relation with the power generation,this may reduce the precision of the prediction results.To improve the accuracy of the neural network forecasting model,this study utilizes data mining to analyze the correlation coefficient between input variables and power generation,to build several models and to find the best input variables by analyzing and comparing the results.The results show that reasonable input variables can not only improve the performance of the neural network power forecasting system,but also reduce the influence of photovoltaic randomness on the power system.