Direct unfalsified controller design-solution via convex optimization

Presents an algorithm for designing optimal unfalsified discrete linear time invariant (LTI) output feedback controllers directly from measured data. The approach uses a performance specification that corresponds to minimizing the time domain error between the desired and unfalsified closed loop transfer functions. No assumptions about the plant are explicitly required the only information needed is a plant input-output time history of length n. In particular, no assumptions are made about the linearity of the system. The identified controller is unfalsified with respect to the performance specification by all observed data. With some minor assumptions, the design problem can be written as a linear program that can be solved very efficiently to find the global optimum. The approach is demonstrated with a laboratory experiment.

[1]  Michael G. Safonov,et al.  The unfalsified control concept: A direct path from experiment to controller , 1995 .

[2]  Robert L. Kosut.  Uncertainty model unfalsification: a system identification paradigm compatible with robust control design , 1995, Proceedings of 1995 34th IEEE Conference on Decision and Control.

[3]  Robert L. Kosut.  Iterative Adaptive Robust Control via Uncertainty Model Unfalsification , 1996 .

[4]  Michael G. Safonov,et al.  The unfalsified control concept and learning , 1997 .

[5]  Brian D. O. Anderson,et al.  Uncertainty model unfalsification , 1997, Proceedings of the 36th IEEE Conference on Decision and Control.

[6]  Michael G. Safonov,et al.  Fitting controllers to data , 1998, Proceedings of the 1998 American Control Conference. ACC (IEEE Cat. No.98CH36207).

[7]  Jonathan P. How,et al.  Uncertainty model unfalsification with simulation , 1998, Proceedings of the 1998 American Control Conference. ACC (IEEE Cat. No.98CH36207).