Integrated plant and control design for vehicle -- environment interaction.

摘要

Vehicle-environment interaction (VEI) systems are a specific class of systems whereby a vehicle, or some component of the vehicle, is required to interact with its environment above and beyond simple locomotion tasks. This dissertation examines the automation of the VEI system where the positioning of a particular vehicle subsystem or end effector with respect to the environment is accomplished in a controlled manner. The interaction between the vehicle and its environment or terrain acts as a disturbance to the controller attempting to position the subsystem. To obtain accurate and highly productive operation, the VEI system under control is required to have a sufficiently high bandwidth so as to achieve the desired task. Unfortunately, this is often not achievable due to fundamental limitations arising from the physical system design. This dissertation studies the fundamental limitations that stem from these physical systems and develops a systematic solution to alleviate the limitations. The results developed are made specific to a VEI system of interest: a combine harvester with automated header height control (HHC).;By modeling and system analysis of the system, the under-actuation and non-collocation sensor/actuator properties of the mechanical system are presented. These properties cause lightly damped low frequency zeros and poles in the open loop transfer functions of interest, thereby preventing effective feedback control beyond a given frequency. In addition, the actuation system, which is electro-hydraulic in the system of interest in this work, introduces a considerable delay and places another bandwidth limitation. To mediate the limitations caused by the mechanical system dynamics, an integrated design and control method is proposed to synergistically optimize parameters in both the physical system and the controller to minimize a novel type of control-oriented objective function. As a result, the redesigned open loop plant obtains more desirable zeros and poles compared to the original product thereby leading to higher performance in the closed loop behavior. To compensate for the delay cause by the hydraulic actuator, this dissertation proposes a two DOF control, adding a feedforward channel to the optimally designed feedback control loop. Both model uncertainty and signal uncertainties are considered in the two DOF controller design to guarantee robust stability and performance. The methods mentioned above are applied and validated with both simulation and experimental results for the combine harvester system of interest. The generality of the approaches presented indicate the research can be applied on other VEI systems with possibly slight modifications due to the nature of the specific system.