Robust adaptive control of single input - single output (SISO) linearizable systems: Applications to an exothermic jacketed stirred-tank reacto

The objective of this work is develop paradigms to improve feedback linearizing control of SISO systems in the presence of uncertainties in model parameters, incomplete state measurements and singular (fast) dynamics, the methods being specifically developed for a continuous stirred tank reactor with first-order irreversible reaction. A comparative study of using recursive least squares (RLS) versus a Lyapunov-stability based method for parameter adaptation is performed on a first-order, two-state, continuous stirred tank reactor (CSTR) model. A procedure to design nonlinear adaptive observers with linear error dynamics and guaranteed parameter identifiability for a class of SISO systems is proposed. We use a combination of singular perturbation theory and Lyapunov stability analysis to study the effect of unmodeled cooling jacket dynamics on the stability of a feedback linearizing controller designed for a reduced order reactor system. Theoretical analysis and simulation results are presented to compare the effects of adaptive and nonadaptive schemes on the performance of the feedback linearization design.