基于BP-ANN和出力波动特性的光伏系统短期功率预测模型

摘要:

目前，光伏系统出力预测在精度方面还远不能满足电力系统调度的要求，已成为阻碍光伏发展的主要瓶颈问 题之一。考虑到光伏系统的高度非线性特性，难以用传统的数学模型表征其输出功率与外界条件之间的映射关系。 首先以传统反传播神经网络作为建模基础，建立光伏系统出力初步预测模型，再基于由光伏系统历史出力数据统计 分析得到的波动量统计规律对初步预测结果进行修正，建立了具有较高精度的光伏系统出力预测模型。进一步进行 算例仿真验证，结果表明所建立的光伏系统出力预测模型能够较好地反应现场实际情况。

Abstract:

Currently，the output prediction of photovoltaic system could not meet the requirements of power system dispatching in the aspect of precision，which has become one of the major bottlenecks to hinder the development of photovoltaic． Taking the highly non － linear characteristics of photovoltaic system into account，it is difficult to use the conventional mathematical model to represent the mapping relationship between the output power and the external conditions． Firstly，an output preliminary prediction mode of photovoltaic system is established using the traditional back － propagation neural network as the foundation of modeling． Secondly，the initial forecast results are corrected based on the statistical regularities of fluctuation quantity in historical output data． And a higher precision model for output prediction of photovoltaic system is established． The simulation and the error analysis are verified，whose results show that the proposed output prediction model of photovoltaic system can reflect the actual situation perfectly．