Control strategy optimization of the hybrid electric bus based on remote self-learning driving cycles

In order to obtain the globally optimal solutions, dynamic programming ( DP ) technique have been investigated. But one major difficulty lies in the need of global /preview information about the trip to be made and the difficulty in obtaining the specific driving cycle for each trip. Another difficulty is the computational load for global optimization algorithms in the micro-processor inside the vehicle. In this paper a new control scheme called control strategy optimization based on the remote self- learning driving cycles is presented. The remote driving cycle self-learning system adopts an in-vehicle device to acquire driving cycle data of electric vehicle through CAN bus and communicate wirelessly with data server by GPRS network and INTERNET. The vehicle driving cycle performance analyses is performed using a database server and then the optimum data is transformed to update the control parameter of hybrid electric vehicle control unit(VCU). As hybrid electric bus is driven in fix-up driving course, the actual driving cycle may be different from that produced by the trip model because of the stochastic nature of the traffic flow. The control strategy optimized by dynamic programming (DP) should be modified according to the vehicle state , battery state. When the vehicle start to run , the optimized control strategy will be transmitted to the vehicle control unit(VCU) and In-vehicle Device will transmit the vehicle message to the central server. While the vehicle state and battery state are different from the pre- optimized, a new optimized computation with dynamic programming (DP) will be executed on the central server and result transmitted to VCU in time.