新组合模型在光电功率预测中的应用

Application of New Combined Model in Photoelectric Power Prediction

为提高光伏预测要求的精准性,文章提出一种新算法将神经网络和ARMA算法改进组合,构成NEW ARMA-BP模型算法.以某30兆瓦的光伏电站采集的输出功率为输入样本,基于ARMA和BP神经网络算法在Matlab环境下依次搭建了相应的预测模型,预估光伏短期输出量.采用"误差正态检验图"判断基于两种不同算法的误差水平,依据两种单模型预测误差,运用所提出的新方法计算权值并获得新的预测值.基于Matlab的仿真结论验证了组合预测在光伏输出预测领域的适用性.

His paper proposes a new algorithm that combines the neural network and ARMA algorithm to form a NEW ARMA-BP model algorithm.Based on the output power collected by a 30 MW photovoltaic power plant as an input sample and based on the ARMA algorithm and BP neural network algorithm,a corresponding prediction model was built in order to predict the short-term PV output under the Matlab environment.The"error normality test chart"is used to judge the error level based on two different algorithms.Then based on the two single model prediction errors,the proposed new method is used to calculate the weights and obtain new prediction values.Simulation results based on Matlab validate the applicability of combined forecasting in photovoltaic output prediction。