| This research focuses on the development, analysis, and implementation of a multivariable nonlinear control strategy and its application to complex mechanical systems such as an electro-hydraulic powertrain.; First of all, local plant dynamics are modeled as a Multi-Input-Multi-Output linear system. The models presented are a combination of first principles and experimentally identified system dynamics. Secondly, a robust gain scheduling strategy is developed. Local controllers are designed using algorithms and then gain-scheduled by multiple scheduling variables to cover the operating envelope. The major challenge in gain scheduling is the dynamics of the scheduling variable in the LPV framework for the global controller, which complicates the LTI local control design. It is suggested here to model these scheduling variable dynamics as a scheduling uncertainty, which can be used to design a gain-scheduled global controller with guaranteed robust stability and performance. Finally, a Hardware-In-the-Loop testbed, namely an Earthmoving Vehicle Powertrain Simulator (EVPS), is designed to emulate target machine dynamics and to test control strategies. Based on a fundamental performance limitation analysis, a dynamic emulation method is developed to emulate power sources or loads using low-bandwidth actuators. Controllers are comprehensively evaluated by EVPS experiments including reference tracking, disturbance rejection, and loading cycles.; The benefits of the above methodology to a MIMO powertrain system include a simplified mechanical structure, model-based control coordination, economical prototyping, as well as improved performance, efficiency, and ease of operation. An immediate application of these results to the earthmoving industry can reduce the cost of the control design cycle by formalizing the procedure. Simultaneously, a superior overall system will result by taking advantage of the multivariable structure. |
