Study On Shift Schedule Under Multi-Performance Constraints For Stepped Automatic Transmissions

Shift schedule determines the time when a stepped automatic transmission should be shifted from one gear position to an adjacent gear position,which plays a crucial role in the automatic transmission control system that affects the dynamic performance,fuel economy and emission of vehicle,directly.In this paper,based on neural network modeling of the performances of engine that include torque,fuel consumption and emission,a comprehensive evaluation function taking into account the dynamic performance,fuel economy and emission of vehicle together,has been established.And then Genetic Algorithm(GA)has been used to optimize the shift schedule under multi-performance constraints for a stepped automatic transmission with the comprehensive evaluation function as target.Furthermore,a simulation-based evaluation system of shift schedule has been built to evaluate the proposed shift schedules under multi-performance constraints.The main contents of the paper are as follows.(1)The types and working principle of stepped automatic transmissions have been introduced briefly.And the current research status and development trends of shift schedules have been summarized.(2)BP network,radical basis function network and generalized regression neural network have been introduced in detail.Based on that,the neural network modeling method for engine performance has been described.And three different types of neural networks including BP network,radial basis function network and generalized regression neural network,have been used to identify the model of engine’s output torque,fuel consumption,CO emission,HC emission and NOx emission respectively.Furthermore,the performances of neural networks have been compared to find the best one suitable for modeling each performance of engine.(3)Based on the analysis of the dynamic performance,fuel economy and emission of vehicle,evaluation function of each performance has been established.With the non-dimensional treatment and driver’s intention weighting for performance indices,the comprehensive evaluation function of shift schedule has been proposed.And two different shift schedules under multi-performance constraints for a stepped automatic transmission,one is dynamic performance dominant and another is fuel economy dominant,are calculated using GA optimization algorithm.(4)A simulation-based evaluation system for shift schedule,which consists of the neural network models of engine performances,the simulation model of AMT(automated manual transmission),the simulation model of vehicle longitudinal dynamics,the shift logic simulation model and the simulation model of ECE+EUDC test cycle,have been built using MATLAB/Simulink/Stateflow platform.And the dynamic performance dominant and fueleconomy dominant shift schedules of AMT have been evaluated through simulation experiments,respectively.The simulation results show that: the shift schedule under multi-performance constraints can achieve good dynamic performance(regarding dynamic performance dominant shift schedule)or fuel economy(regarding fuel economy dominant shift schedule)of vehicle,meanwhile keeps the fuel economy or dynamic performance and emission at acceptable levels,depending on the weights that reflects the driver’s intentions.