Prediction model of fuel consumption of heavy truck based on improved BP neural network

Abstract

In recent years, with the continuous growth of the number of heavy trucks, the total amount of fuel consumed also increased. However, the total energy reserves are fixed. So it is crucial to improve the fuel efficiency of heavy trucks. In this paper, based on the heavy truck driving behavior data collected, a Back Propagation(BP) neural network is established to predict the fuel consumption value of heavy trucks. In order to improve the accuracy, genetic algorithm (GA) and simulated annealing algorithm (SA) are proposed to improve the fuel consumption model. The improved heavy truck fuel consumption model based on genetic algorithm (GA-BP), Simulated Annealing algorithm (SA-BP), and genetic annealing algorithm (GSA-BP) is established respectively. By comparing the prediction accuracy, root mean square error (RMSE), and mean absolute error (MAE) of the three models, it is concluded that the prediction model based on GSA-BP fuel consumption of heavy trucks satisfies the criteria of this study and has the best prediction performance. The study of this method can provide a theoretical basis for energy-saving driving in the future.