Applying genetic search techniques to drivetrain modeling

Work carried out to identify a nonlinear model of a vehicle engine and drivetrain is discussed. A hybrid approach that combines both physical modeling and parameter optimization using genetic algorithm (GA) search techniques is used. The resulting models, which cover a range of operating conditions, have allowed the sensitivity to variation of key parameters to be assessed and have been used to help optimize the overall response of the vehicle drivetrain. A comparison of the GA search and a gradient based method, which highlights the intelligent nature of the former approach, is presented.<
>