CONTROLLER ORDER REDUCTION BY IDENTIFICATION IN CLOSED LOOP

This module allows to perform identification of a reduced order model of a controller from data which have been acquired in closed loop or data generated by simulation. We suppose that the plant model is perfectly known.

The key idea of this method with respect to classical identification methods is to find an adjustable predictor (the reduced order model of the controller) which will be tuned in order to minimize the error between the controller output (operating in closed loop) and a model of the loop made of the adjustable predictor and of the plant model.

Three different recursive algorithms based on the minimization of the closed loop output error are developed:

1. CONID : Controller identification by the minimization of closed loop output error .

2. CONIDF : Controller identification by the minimization of filtered closed loop output error .

3. CONIDAF : Controller identification by the minimization of adaptive filtered closed loop output error .

Controller Order Reduction Function

A controller order reduction function may be used to decrement the order of a nominal controller from the initial order to zero. In each step some mesures of robust stability and performance of the reduced order controller like Vinnicombe gap metric and stability margin are displayed. There are two possibilities for controller order reduction :

- Controller order reduction by simulated data.

- Controller order reduction by real time acquired data.

The comparison of the reduced order controllers may be carried out on detail with the comparison of the frequency responses of the controllers and the magnitude of the closed loop sensitivity functions.

Help topics.

(file cicl.htm)