Skip to Main Content
The proper tuning of multivariable PID controllers is not a straightforward task. This is specially true when data driven models and controller design methodologies are adopted to increase the industry acceptance of controller implementation and commissioning. On the other hand there are well established model based design methodologies which provide controllers of higher complexity. In this paper we propose a tuning methodology based on a PID controller matching to a figure of merit. Thus the PID tuning parameters are selected to minimize certain error metric when compared with an optimal controller. Moreover the index to be minimized is a residual that can be obtained from data. As a result it is possible to validate the proximity of the PID controller to the desired figure of merit in the frequency domain from data.