Enhancing model accuracy for control: Two case studies

The modeling and control of complex systems is difficult and expensive, especially when the building and testing of prototypes is included. This dissertation considers modeling issues related to control for two high order nonlinear systems: a flexible two-link manipulator and an astronaut space suit thermal system, which represent two typical dynamic systems. For each system, both modeling and control issues are investigated using real world experimental setups.; The manipulator system is modeled by using the Euler-Lagrange equations of motion together with a modal approach. After obtaining a closed-form model, the mode shape parameters were modified based on the fact that the generalized mass matrix M has to be positive definite and that the predicted natural frequencies of the linearized model have to agree with the experimental results.; A transient model of the Advanced Space Suit design is developed and implemented using MATLAB/Simulink, to help with sizing, with design evaluation, and with the development of an automatic thermal comfort control strategy. The model is described and the thermal characteristics of the Advanced Space Suit are investigated including various parametric design studies. The steady state performance envelope for the Advanced Space Suit is defined in terms of the thermal environment and human metabolic rate and the transient response of the human-suit-MPLSS system is analyzed.; A robust adaptive control design for a flexible two-link system considering joint angle control in the presence of system uncertainties is proposed. The system uncertainties are treated as unknown bounded continuous functions of the states. A Gaussian network is used to map the uncertainties in the MIMO system. A convergent stable learning algorithm is developed for the control design. Simulation and experimental results show the effectiveness of the proposed learning control strategy.; The regulation of the thermal comfort of the astronauts relies on the control of the inlet temperature of the liquid cooling garment (LCG). A control design for EVA thermal comfort based on the Generalized Predictive Control (GPC) technique is proposed. Multiple models and a switching scheme are employed. Experimental results show the effectiveness of the proposed method.