Stability and Performance for Saturated Systems via Quadratic and Nonquadratic Lyapunov Functions

• Published in: IEEE transactions on automatic control Vol. 51; no. 11; pp. 1770 - 1786
• Main Authors: Tingshu Hu, Teel, A.R., Zaccarian, L.
• Format: Journal Article
• Language: English
• Published: New York, NY IEEE 01.11.2006; Institute of Electrical and Electronics Engineers; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Actuators; Algebra; Applied sciences; Bismaleimides; Computer science; control theory; systems; Control systems; Control theory. Systems; Deadzone; Design optimization; domain of attraction; Exact sciences and technology; Feedback; Inclusions; Inspection; Level set; Linear matrix inequalities; Lyapunov functions; Lyapunov method; Mathematical models; Nondestructive testing; nonlinear {\cal L}_{2} gain; Performance analysis; reachable set; Regional; saturation; Sensor systems; Stability; Stability analysis; Studies; Polytope; Differential form; reachable set; Regional analysis; Saturation; Feedback regulation; Dead zone; Quadratic system; Reachability analysis; Matrix function; Local effect; Convex hull; Lyapunov functions; Deadzone; Linear matrix inequality; Convex function; Performance analysis; Domain of attraction; Differential inclusion; Non destructive test; nonlinear C2 gain; Lyapunov function; Quadratic function
• ISSN: 0018-9286; 1558-2523
• DOI: 10.1109/TAC.2006.884942

In this paper, we develop a systematic Lyapunov approach to the regional stability and performance analysis of saturated systems in a general feedback configuration. The only assumptions we make about the system are well-posedness of the algebraic loop and local stability. Problems to be considered include the estimation of the domain of attraction, the reachable set under a class of bounded energy disturbances and the nonlinear L 2 gain. The regional analysis is established through an effective treatment of the algebraic loop and the saturation/deadzone function. This treatment yields two forms of differential inclusions, a polytopic differential inclusion (PDI) and a norm-bounded differential inclusion (NDI) that contain the original system. Adjustable parameters are incorporated into the differential inclusions to reflect the regional property. The main idea behind the regional analysis is to ensure that the state remain inside the level set of a certain Lyapunov function where the PDI or the NDI is valid. With quadratic Lyapunov functions, conditions for stability and performances are derived as linear matrix inequalities (LMIs). To obtain less conservative conditions, we use a pair of conjugate non-quadratic Lyapunov functions, the convex hull quadratic function and the max quadratic function. These functions yield bilinear matrix inequalities (BMIs) as conditions for stability and guaranteed performance level. The BMI conditions cover the corresponding LMI conditions as special cases, hence the BMI results are guaranteed to be as good as the LMI results. In most examples, the BMI results are significantly better than the LMI results