Research On The Photovoltaic Prediction Model Based On VMD-LAO-LSTM

With the rapid popularization and development of renewable energy,solar photovoltaic power generation has become an important energy choice.In order to better utilize solar energy,it is crucial to accurately predict photovoltaic power generation.On the basis of exploring a photovoltaic power prediction model based on optimizing convolutional neural network(AO-CNN),this paper proposes a photovoltaic power prediction model for optimizing LSTM network(LAO-LSTM)based on Cauchy variation factor improving the eagle algorithm(LAO).The main research contents of the paper include:First of all,this paper proposes a convolutional neural network prediction model(AO-CNN)based on the AO for the long intensive training system of the neural network model(Convolutional Neural Networks,CNN).In addition,in order to speed up the prediction of the neural network,the improved Skyhawk optimization algorithm AO will optimize the neural network structure,encode the structural parameters of the neural network as the individual in the algorithm,select the optimal solution as the neural network structure,and then make the prediction.The test is tested and verified in the test set,and the comparative analysis of the experimental results of the optimization algorithm proposed in recent years shows that the proposed AO-CNN algorithm has better solution performance.Secondly,a kind of LSTM network prediction model(LAO-LSTM)prediction model based on the improvement of the sky eagle optimization algorithm is proposed.And separately with the conventional LSTM neural network and the LSTM neural network optimized after a modified Skyhawk algorithm.Experimental results show that LAO-LSTM neural networks have higher accuracy,smaller error and faster speed than conventional LSTM neural networks.Finally,the LSTM network prediction model is proposed for the distortion of photovoltaic power parameter data in the early stage of the algorithm.Through wavelet transform processing,the abnormal data of the problem model is processed,and the problem of the abnormal data is solved.Through a large number of experiments,the improved neural network model has predictable accuracy requirements and higher neural network model,which provides new possibilities for optimizing the basic structure of deep neural network,and can generally act on many foreseeable problems.In conclusion,the proposed method to optimize the neural network structure can automatically determine the best neural network structure,and can better improve the performance prediction accuracy of the neural network.This method has widely applications in photovoltaic power prediction applications and also provides a new idea for applications in other fields.