Prediction of Vehicle Fuel Consumption Based on Combination of Improved Whale Optimization and XGBoost

The highway transportation industry is the most oil-intensive industry in the world. Therefore, an accurate prediction of fuel consumption not only reveals the relationship between fuel consumption and driving characteristics, but also promotes green development of the automobile industry. A fuel consumption prediction model based on Extreme Gradient Boosting (XGBoost) is built in this research, where an improved Whale Optimization Algorithm (IWOA) is developed to optimize the main parameters in the prediction model. In the IWOA, Sobol sequence initialization is introduced to improve the diversity of the population and expand the search range. The nonlinear convergence coefficient and Levy Flight are also introduced in the algorithm to balance the global exploration ability and the local exploitation ability. The experimental results show that the IWOA-XGBoost model performs better than some existing popular models in both accuracy and stability. This study provides fundamental theoretical, methodological, and technical support for automobile fuel saving and emission reduction.