A prediction optimization scheme based on improved GRU neural network is proposed to meet the accuracy require-ments of energy consumption prediction in the current comprehensive energy control system.Firstly,considering that the rate of selec-ting hyperparameters in the GRU neural network prediction model directly affects the accuracy of the prediction model,a whale opti-mization algorithm is proposed to optimize the hyperparameters;Then,the hyperparameters obtained by the WO A algorithm are set on the GRU neural network,and the optimized hyperparameters are used to predict the comprehensive energy load;Finally,this algo-rithm is compared with traditional GRU prediction models and BP neural network prediction models through evaluation indicators MAE,MPAE,and RMSE.The results show that the average absolute error percentage of this optimization scheme is 1.79％,while the average absolute error percentages of the traditional GRU prediction model and BP prediction model are 3.06％and 4.45％.From this,it can be concluded that the improvement of the GRU neural network using the whale optimization algorithm makes the GRU pre-diction model more accurate and stable.
关键词
GRU神经网络/综合能源管控系统/超参数/鲸鱼优化算法/系统优化
Key words
GRU neural network/comprehensive energy management and control system/hyperparameters/whale optimization algorithm/system optimization
维普
万方数据