基于改进的GABP神经网络光伏发电短期出力预测

分析了影响光伏出力的主要因素,选取了太阳辐射量,以及隐含前一日综合气象信息的历史出力数据为关键影响因素,建立了改进的GA-BP神经网络的短期光伏发电功率预测模型.对样本空间进行了合理降维和去噪,并利用遗传算法逐步迭代出优化的初始权值,将得到的最优权值(阈值)赋值给预测网络各层进行学习和预测.仿真结果表明,改进的GA-BP神经网络模型能够剔除冗余的样本数据和优化初始权值,既具备了较快的收敛速度又不易陷入到局部极值中,具有较强的泛化能力,预测精确度大幅提高.

PV Short-term Output Forecasting Based on Improved GA-BP Neural Network

The main factors influencing the pv output are analyzed,the short-term pv model based on improved GA-BP neural network is established by the historical output data of the solar radiation and hidden comprehensive meteorological information of the previous day.The algorithm reduces the dimension and noise of the sample space,and the iterative initial weights are oprimized by using genetic algorithm,the optimal weights (threshold) are assigned to each layer of the network.The simulation results show that the improved GA-BP neural network model can eliminate the redundant data and optimize the initial weights,which not only has a faster convergence speed and is not easy to fall into local extremum,but also has strong generalization ability,and the prediction accuracy is greatly improved.