基于遗传算法优化神经网络的光伏电站短期功率预测

Short-term Power Prediction of PV Based on Combined BP-GA Neural Network

针对现有光伏发电预测的不足,基于遗传算法(GA)和神经网络(BP)算法构建光伏电站功率预测模型,并使用组合权重法遴选相似日对模型进行修正。采用新疆某光伏电站运行实例验证模型的有效性,并对比BP-GA模型与单一BP模型的预测误差。结果表明,BP-GA模型克服了传统单一BP模型的不足,具有较高的预测精度,可为光伏发电预测工程实践提供参考。

Taking into account the insufficient of existed PV power generation forecasting,this paper established a new PV system power forecasting model based on genetic algorithm and neural network.And then the model was corrected by using selection of the similar days with combined weight.Furthermore,running instance of a certain photovoltaic power plant in Xinjiang was adopted to verify the effectiveness of the proposed model.Compared with the prediction errors of single BP model,the results show that the proposed prediction model overcame the defect of the traditional single model and had higher accuracy.Thus,it can provide decision support for PV power generation projects.