Energy Consumption Analysis and Prediction of New Energy Buses

Under the concept of protecting the environment and green travel, the popularity of new energy buses is gradually increasing in China. As the main public transportation for urban travel, new energy buses have become the choice of the market. How to manage the new energy buses economically and effectively requires quantitative analysis through a large number of energy consumption statistics. In this paper, we establish an attention-based LSTM model with more accurate prediction results according to the 11 influencing factors of new energy vehicle energy consumption, including the classification results of driving behavior, and then conduct comparison experiments and ten-fold cross-validation of the model. The results show that the accuracy of the model is high enough to meet the engineering requirements. To improve the generalization of the prediction model, a Stacking-based integrated model with stronger generalization is built and trained, and the training and preservation of the integrated model are optimized by incremental algorithms so that the model can improve its learning ability and prediction accuracy on old and new data while maintaining the generalization ability of different distributions.