Data driven tuning of state space controllers with observes

Abstract

Iterative Feedback Tuning is a purely data driven tuning algorithm for optimizing control parameters based on closed loop data. The algorithm is designed to produce an unbiased estimate of a performance cost function gradient and the control parameters are iteratively improved in order to achieve optimal loop performance. This tuning method has been derived for and is widely applied on systems using a transfer function representation. In this paper equivalent forms are found for a control system in a state space representation, with state observer and proportional feedback, and in a transfer function representation. It is shown how the parameters in the transfer function, describing the feedback control of a state space system, can be tuned by Iterative Feedback Tuning. A simulation example illustrates that the tuning converges to known analytical solutions for the feedback control gain and the Kalman gain in the state observer.