We emulated the learning property of biological control systems by developing adaptive control strategies for poorly understood nonlinear processes. Our initial work focused on indirect adaptive control where linearly parameterized radial basis functions are used to approximate unknown functions in the open-loop model. We also developed a direct adaptive control strategy in which the unknown controller functions are constructed directly. The only information required about the nonlinear process is measurements of the state variables, the relative degree and the sign of the high frequency gain. The controller parameters are adapted on-line such that the process output tracks the output of a linear reference model. The technique was extended such that fixed linear controller gains can be embedded within in the adaptive nonlinear controller. Principal component analysis was utilized to enhance the computational efficiency of the approach. The technique was applied to nonlinear chemical, biochemical and polymerization reactor models.
Funding: National Science Foundation (CTS-9501368) and DuPont
Student: Richard B. McLain (Ph.D.)
Collaborator: Dr. Martin Pottmann (DuPont)
Publications: