基于相似日和WNN的光伏发电功率超短期预测模型

A very short-term prediction model for photovoltaic power based on similar days and wavelet neural network

光伏发电功率预测对提高并网后电网的稳定性及安全性具有重要意义。文章提出一种基于相似日和小波神经网络(WNN)的光伏功率超短期预测方法。首先利用光伏发电系统的历史气象信息建立气象特征向量,通过计算灰色关联度寻找到合适的相似历史日。再根据自相关性分析法找出与预测时刻功率相关性最大的几个历史时刻功率,结合历史时刻的温度,辐照度,风速等光伏出力的主要天气影响因素科学合理的确定模型输入因子。最后使用小波神经网络(WNN)创建预测模型,通过相似历史日数据作为训练样本训练小波网络,而后对预测日的出力情况进行逐时刻预测。实例分析表明,该方法具有较高的预测精度,为解决光伏发电系统超短期功率预测提供了一种可行路径。

Photovohaic （PV） generation power prediction has great significance for the stability and security of power grid after the PV grid-connection. In this paper, we propose a very short-term photovoltaic power forecasting method- which is based on similar days and wavelet neural networks （WNN）. Firstly, the historical weather information from the PV power generation system is utilized to establish meteorological feature vectors, and similar days are found based on computation grey correlation degree. Secondly, the autocorrelation analysis method isused to discover historical out- put power which has great relation with predicted output power. The historical meteorological data, such as tempera- ture, irradiance and wind speed, are utilized to determine the input factor of this model. Finally, the wavelet neural network （WNN） is utilized to create a forecast model, which is to predict forecasting daily output one by one moment- though the similar historical day data as training sample of WNN. The instance analysis shows that this model has high accuracy, and can provide an effective and feasible way to forecast the very short-term power output of the PV system.