Vehicle Fuel Consumption Prediction Based on Least Squares Support Vector Machine Optimized by Improved Particle Swarm Optimization Algorithm

摘要

(1) This paper summarized the vehicle operation and design parameters, which are directly related to the fuel consumption, and SU is used to select the sensitive parameters of fuel consumption. The results show that the average speed, engine displacement, vehicle weight, maximum power, and transmission types are the critical parameter that affects the vehicle fuel consumption; the vehicle body structure, idle ratio, and cylinder arrangement also have some influence on the fuel consumption; (2) The advanced nonlinear prediction methods LSSVM is introduced to predict the vehicle fuel consumption, and an IPSO algorithm is used to optimized the kernel function parameter of LSSVM; (3) The proposed fuel consumption prediction model can accurately predict the actual fuel consumption and can effectively make up for the gap between the actual fuel consumption and fuel consumption of type approve test.