基于滚动的BP神经网络的光伏发电功率预报

鉴于准确预测光伏发电功率可减少大规模光伏并网发电对电网造成的冲击,以河南省某并网的光伏发电站7~11月共5个月的数据为例,通过建立非滚动的和滚动的BP神经网络模型,分别进行数值预报辐射订正和发电功率预报,并对预报的72h结果分为第1、2、3d分别进行检验。结果表明,滚动的BP神经网络对辐射订正和功率预报均具有较好的泛化能力,方法简便、实用,能够有效降低光伏发电功率的预报误差。

Photovoltaic Power Forecast Based on Rolling BP Neural Network

Accurate prediction of photovoltaic power generation can reduce the impact of large-scale grid connected. This paper studies on the photovoltaic power generation grid in Henan Province from July to November data through the establishment of non-rolling and rolling of BP neural network model respectively for WRF prediction of radiation correction and power forecasting. The results of the three days of prediction are divided into first days, second days and third days respectively. The application proves that the method is simple and practical, rolling BP neural network has better generalization ability for radiation correction and power prediction, which can effectively reduce the prediction error of photovoltaic power.