Abstract: The development of hybrid electric control strategies typically use American or Japanese driving cycles. However, there is great difference between these driving cycles and the domestic actual ones, which causes the developed control strategy unable to make the hybrid electric bus attain the optimal fuel economy. In view of this problem, a driving cycle self-learning system is set up upon consideration of the traits of the bus, such as fixed commuting route and strong periodicity. Experimental results from the driving cycle self-learning show that the constructed driving cycle can adequately represent the route. Based on the constructed driving cycle, the control strategy is optimized by using dynamic programming. The target vehicle runs on the collected driving cycles according to the power distributed by dynamic programming. The simulation result shows that the fuel consumption is reduced by 10.2 percent in comparison to the actual vehicle running on the route with the power follower control strategy. Through this method, “one route one strategy” for each hybrid electric bus is realized and the adaptability thereof is improved.

Key words: Driving cycle, Dynamic optimization, Hybrid electric, Self-learning