Iterative Data-Driven Controller Tuning with Actuator Constraints and Reduced Sensitivity

Abstract

This paper proposes a novel iterative data-driven algorithm for the data-driven tuning of controllers for nonlinear systems. The iterative data-driven algorithm uses an experiment-based solving of the optimization problems for nonlinear processes, with linear controllers accounting for actuator constraints in terms of a quadratic penalty function approach. A neural network-based identification provides the gradient information used in the search algorithm for controller tuning and ensures a reduced sensitivity with respect to the controller parameters. A case study dealing with the data-driven controller tuning for the angular position control of a nonlinear aerodynamic system is included to validate the new iterative data-driven algorithm.