Stealth identification strategy for closed loop system structure

This paper was not presented at any IFAC meeting.

Abstract

Many identification strategies for closed loop system structure assume that an open loop system is closed by a feedback mechanism, which contains a known and linear controller. This assumption means these identification strategies are feasible under some prior knowledge of feedback controller. To relax this assumption for some complex systems with unknown controller or nonlinear controller, a new stealth identification strategy is proposed to tackle the identification problem for closed loop system with unknown controller or nonlinear controller. Stealth identification modifies closed loop system, so that the new prediction error and inverse covariance matrix are all independent of the unknown controller or nonlinear controller. This independence simplifies the problem of estimating parameter vector and designing optimal input. The Robbins-Monro algorithm is applied to identify the unknown parameter vector in closed loop system with unknown controller, and the nonlinear controller is replaced by its equivalent linear time invariant controller. Some consideration about the linear approximation to nonlinear system are studied, and two linear approximation forms are constructed to approximate the nonlinear system. The effectiveness of our proposed stealth identification strategy is demonstrated through simulation example.

Keywords:

• Closed loop system
• stealth identification
• unknown controller
• nonlinear controller: linear approximation

Acknowledgments

The authors wish to thank many valuable suggestions from the anonymous reviewers, especially constructive and insightful comments on linear approximation to nonlinear system, which have significantly improved the readability and quality of the paper. This work is partially supported by the Grants from the Foundation of Jiangxi University of Science and Technology (No. jxxjb18020). The authors are grateful to Professor Eduardo F Camacho for his warm invitation in his control lab at the University of Seville, Seville, Spain. Also, the authors wish to thank Xavier Bombois for his fruitful suggestion and the reviewers'suggestion about one new added part on linear approximation to nonlinear system.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

This paper was not presented at any IFAC meeting.

Additional information

Funding

This work is partially supported by the Grants from the Foundation of Jiangxi University of Science and Technology [grant number jxxjb18020]; National Nature Science Foundation of China [grant number 6257537].

Notes on contributors

Hong Wang-jian

Hong Wang-jian received the diploma in Engineering Cybernetics from the University of Yun Nan China, in 2007. In 2011, he received the Dr.sc. degree in College of Automation Engineer from Nanjing University of Aeronautics and Astronautics, China. He is currently a professor in Tecnologico de Monterrey. From 2013 to 2015, he was a postdoctoral fellow in the 28th Research Institute of China Electronics Technology Group Corporation. From June 2016 to September 2016, he was a visiting professor in Informazione Politecnico di Milano. His current research interests include real-time and distributed control, and optimization and system identification. Hong Wang-jian can be contacted at: [email protected].

Ricardo A. Ramirez-Mendoza

Ricardo A. Ramirez-Mendoza received a Ph.D. degree from INPG, France in1997. He is now Professor of Mechatronics-Mechanics and Dean of Research, School of Engineering and Science at the Tecnologico de Monterrey. His main research interests include applications of advanced control to automotive systems. He is the (co-)author of 3 books, more than 80 papers in top journals and more than 200 international conference papers and more than 1,000 citations. He has worked as expert consulting for different industries and he is certified reviewer for regional development projects. Ricardo A. Ramirez-Mendoza can be contacted at: [email protected].