Design And Implementation Of Power Load Energy Consumption Forecasting System Based On EEMD-WOA-LSTM

Electricity is seen as vital for the country’s industrial development.Predicting the use of electricity can better help power grids and enterprises understand the operation of factories,so the balance of supply and demand between power grids and enterprises is directly affected by the accuracy of power load forecasting.The disadvantages of low efficiency,heavy workload,large error and poor economy will be faced when the power load data is predicted only by manual work,so it is very important that the intelligent algorithm is introduced in the field of power load forecasting under the condition of huge power use today.Among them,the selection and optimization of the algorithm can not be substituted to improve the prediction accuracy and working efficiency of power load.In this paper,the power load forecasting algorithm is studied,and the optimization of long-term and short-term memory neural network is regarded as the main research object,and the problem of determining the parameters of the neural network and the nonlinear stationary problem of the power load data are solved,the effect of accurate prediction of nonlinear and stable power load data is achieved.The main contents studied by this article are as follows:(1)Firstly,the traditional algorithm is introduced in this article,and the advantages and disadvantages of Recurrent Neural Network(RNN)and short-term and short-term memory neural network(LSTM)are analyzed.Because LSTM have strong adaptive learning ability and good memory function,the learning advantage of long-term memory for long-term data is reflected,and the results are predicted more accurately.As a result,LSTM neural network is used as the prediction algorithm in this article.In order to solve the problem that the parameters of LSTM neural network are difficult to be determined,Whale Optimization Algorithm(WOA)is introduced for parameter optimization.Based on experience,the process of setting parameters artificially is changed to the process of automatic iterative search of optimal parameters by Whale Optimization Algorithm.The real factory data were used as the original data for verification.Compared with the traditional LSTM model Mean Absolute Error(MAE),Root Mean Square Error(RMSE)and Mean Absolute Percent Error(MAPE),the WOA-LSTM model optimized by Whale Optimization Algorithm decreased by 35.9%,38.7% and 35.9% respectively.Results the model optimized by the Whale Optimization Algorithm is proved to have higher prediction accuracy than the model without optimization.(2)Then,for the nonlinear stationary problem of power load data,the model EEMD-WOA-LSTM of Ensemble Empirical Mode Decomposition(EEMD)algorithm combined with WOA-LSTM is proposed.The historical data are decomposed by the EEMD algorithm to obtain a number of feature components IMF;each feature component is IMF trained by the neural network,and the prediction results of each component are obtained,and then the final can be obtained by linear superposition.Through the comparison the EEMD-WOA-LSTM algorithm and the WOA-LSTM algorithm shows that the MAE is reduced by 30.6%,the RMSE by35.1%,and the MAPE by 29.6%.The results show that the EEMD-WOA-LSTM model has obvious effect in improving the prediction accuracy.(3)Finally,a complete intelligent energy consumption management system including data acquisition,display and prediction is designed and implemented in this paper.From the long-term operation results of the system in the actual production of the factory,compared with the traditional neural network algorithm,the EEMD-WOA-LSTM model proposed in this paper has higher generalization ability,and the prediction accuracy is obviously improved.It is of positive significance to improve the power load forecasting level.