High-accuracy transient fuel consumption model based on distance correlation analysis

Abstract

With the gradual aggravation of energy shortage, automobile energy saving has gained widespread attention from scholars. However, due to the lack of a high-accuracy practical fuel consumption model, it is difficult to estimate transient fuel consumption and evaluate the actual effect of real-time fuel consumption control strategies. Therefore, it is necessary to establish a more accurate and practical model according to the transient motion characteristics of the vehicles. To ensure the accuracy of the model, an integrated structure of the steady-state base module and the transient correction module is determined as the overall structure of the model. Based on the steady-state fuel consumption data, the steady-state base module is established. Then, based on the easily obtained vehicle and engine state parameters, principal component analysis and cluster analysis are used to reasonably classify different driving conditions of the vehicles. Following that, the distance correlation analysis is applied to find the combination of state parameters with the strongest correlation with the estimation error of the steady-state module, and a transient correction module is established according to the optimal state parameter combination obtained. After that, the optimal transient correction module structure is determined based on the Bayesian criterion. Finally, the model is tested, and the results show that the mean absolute percentage error (MAPE) of the fuel consumption estimation of the new model is about 15%, while that of the classical VT-Micro model and the VT-CPFM model are about 28% and 20%, respectively. It can be seen that the new model has a higher accuracy. On the other hand, compared with structured physical fuel consumption models such as VT-CPEM model, the new model has a simpler structure, a shorter computation time, and a higher computational speed. In addition, the new model has high practicability due to its clear structure and easy access to parameters.