基于INFO优化CNN-BiLSTM混合网络模型的光伏发电站功率预测研究OA

The accurate prediction of photovoltaic power generation is of great significance to optimize energy management and grid planning.Aiming at the problems of low prediction accuracy of previous photovoltaic power forecasting methods,insufficient mining of input characteristic information,uncertainty of parameter selection and slow convergence of CNN-BILSTM hybrid network model,based on historical meteorological data and photovoltaic power generation data,This paper proposes a new algorithm combining vector weighted mode decomposition INFO and CNN and BiLSTM neural network for photovoltaic power generation prediction.First of all,a large number of measured data are collected,including solar radiation,temperature,humidity,wind speed,air pressure and other meteorological parameters,as well as the corresponding photovoltaic power generation.Then,a prediction model based on CNN-BiLSTM hybrid network is constructed,and the INFO algorithm is used to optimize the number of hidden layer nodes,initial learning rate and L2 regularization coefficient.By adaptively adjusting these parameters,the INFO algorithm shorens the time of manual parameter modulation and improves the accuracy and efficiency of hyperparameter setting.The experimental results show that the CNN-BiLSTM hybrid network optimized by INFO algorithm has higher prediction accuracy and better generalization ability than the traditional CNN-BiLSTM hybrid network.