Moving Horizon Optimal Control For Automotive Driveline Systems

In the process of starting, shifting, urgently accelerating and decelerating, mechan-ical resonances of vehicle drivetrain may occur due to the elasticity of the drivetrainparts, such as clutch spring, propeller shaft and drive axle shaft. How to avoid or re-duce the oscillations of drivetrain is an important problem, especially for heavy dutyvehicles which have relatively large drivetrain torsion. In order to improve the controlperformance, this thesis uses the newly advanced control theory and methods, includingmodel predictive control theory, subspace identification and data-driven control methods,to address some typical control issues of automotive drivetrain. The following issues areinvestigated: clutch speed control of start-up process, anti-jerk control, gearshift controland clutch pressure estimation. By analyzing the diferent control problems and controlrequirements, this thesis uses diferent system models to predict the future behavior ofthe system, optimizes the various control requirements to coordinate dynamic and com-fort in the process of starting, shifting, urgently accelerating and decelerating. Moreover,the saturation of actuators, the limitation of frequency response and the threshold valueof some variables are considered. On the premise of well meeting the constraints of thesystem, the proposed control methods improve the vehicle’s longitudinal speed controlperformance.The objective of this thesis is to present systematic moving horizon control schemefor automotive drivetrain system. Firstly，the controller is proposed for vehicle start-upprocess by using Model Predictive Control (MPC) strategy, which considers the state andinput variables (engine speed, clutch friction torque and engine torque) constraints in theproblem formulation. The contradictive requirements of less friction loss and less drivelineshock are included in the objective function. The proposed control strategy is tested ona powertrain simulation model. Simulation results obtained are presented and discussed.Secondly, a data-driven predictive controller is designed for the start-up process ofvehicles with automated manual transmissions (AMTs). It is obtained directly fromthe input–output data of a driveline simulation model constructed by the commercialsoftware AMESim. In order to obtain ofset-free control for the reference input, the pre- dictor equation is gained with incremental inputs and outputs. Because of the physicalcharacteristics, the input and output constraints are considered explicitly in the problemformulation. The contradictory requirements of less friction losses and less driveline shockare included in the objective function. The designed controller is tested under nominalconditions and changed conditions. The simulation results show that, during the start-upprocess, the AMT clutch with the proposed controller works very well, and the processmeets the control objectives: fast clutch lockup time, small friction losses, and the preser-vation of driver comfort, i.e., smooth acceleration of the vehicle. At the same time, theclosed-loop system has the ability to reject uncertainties, such as the vehicle mass androad grade.Thirdly, a model predictive controller is proposed for anti-jerk during the tip-in/outprocess of the vehicles. The contradictive control objectives of smooth acceleration/decel-eration and satisfying the driver’s torque request are included in the objective function,which aims to improve driving comfort and guarantee fast dynamic response. Due to thephysical characteristics, the maximum frequency response amplitude of engine as actua-tor is considered explicitly in the problem formulation. The designed constrained modelpredictive controller is tested in the nominal conditions and the changed conditions. Thesimulations illustrate the efciency of the proposed controller. At the same time, theclosed-loop system has ability to reject uncertainties.Finally, during the gear shift management process, the shift time and shift shockafect the shift quality (smoothness and efciency) greatly. In order to improve the shiftquality of vehicle with Automatic Transmissions (ATs), a model predictive controlleris designed for the clutch slip control. In this paper, the designed Model PredictiveControl (MPC) strategy is based on a clutch cylinder pressure observer. The contradictivecontrol requirements of short shift time and small shift shock are taken into account in theoptimal objective function by tracking a selected proper reference trajectory smoothly andlimiting the change rate of the actuator. Moreover, due to the physical characteristics,the constraints of actuator are considered explicitly in the problem formulation. At last,the efectiveness of the proposed model predictive controller is tested on an AMESimpowertrain simulation model in the nominal conditions and the changed conditions. It isverified that the system is robust to the variations of driving conditions.The design processes and the analysis of mentioned drivetrain systems control algo-rithms are presented in detail in this thesis. Moreover, in order to validate the efcienciesand cost of the proposed approaches, we give simulation results for each approach in dif-ferent driving conditions. The simulation results indicate that the control performancewith the proposed approaches are improved and their own features are explicit. Deeper research work needs to be done since some problems are still remain to besolved. For the constrained Model Predictive Control methods, the optimal problemneeds to be solved repeatedly at each sample time, it will take more time for calculation.In order to apply the designed controllers into the practice, some new strategy shouldbe studied to speed up the online calculation of optimal problem. At the same time, theproblem that how to take advantage of the online/ofine data to predict the future outputof the system also needs to be solved. Because in-vehicle test has not been carried outuntil now, the major future work should be the implementations of Hardware-In-the-LoopSimulation (HILS) and in-vehicle experiments.