High-precision transient fuel consumption model based on support vector regression

Due to the increasingly severe environmental issues caused by transportation, energy conservation and emission reduction have become one of the most important research directions in the field of transportation. As the key tool for estimating transient fuel consumption, fuel consumption models are crucial for quantifying the energy consumption of automobiles and achieving energy conservation and emission reduction. However, most existing fuel consumption models suffer from problems such as a complex structure, difficult-to-determine model coefficients, and difficult-to-measure input parameters. Therefore, it is necessary to propose a high-precision fuel consumption model with good applicability and easy implementation. The model proposed in this paper adopted a structure of steady-state estimation and transient correction and used support vector regression to predict fuel consumption. First, the engine torque was fitted based on the steady-state data, and the steady-state module was established. Then, based on data analysis using the Gaussian mixture model, the transient motion was divided into three driving conditions, and the transient module was obtained by transient correction for different driving conditions. Ultimately, the performance of the model was validated. The results show that the mean absolute percentage error of the new model is less than 14%, and the mean absolute error is less than 0.16 cc/s, which is a significant improvement compared with other classic models, indicating that the new model has a high prediction accuracy. In addition, the new model takes the most common engine speed, vehicle speed, and acceleration as inputs, making it easy to use and has good applicability and popularization value.