Font Size: a A A

Fault Detection For Nonlinear Switched Systems With Missing Measurements

Posted on:2017-06-20Degree:MasterType:Thesis
Country:ChinaCandidate:Q H LiuFull Text:PDF
GTID:2348330503487225Subject:Control Science and Engineering
Abstract/Summary:PDF Full Text Request
Switched system is a dynamic system which consists of a set of subsystems and a group of switching signals. Many physical systems can be modeled by switched systems due to their nature of multi-mode or variable structures, such as aeronautics and astronautics, electrical systems, network systems, chemical systems and so on. Network communication has become a popular tool for propagation of information, which attracts a lot of attention and applications. When a switched system is running in a networked environment, its performance and stability are seriously influenced by time-delays and data dropouts induced by a constrained network. Analysis and synthesis of switched systems under a constrained network have been brought to the forefront by researchers, and some results on linear switched systems have been published in international journals, which in advance provides fundamental theoretical support for the other complex multiple modes switched systems. However, the results in the field of fault detection for switched systems under a constrained network are not fruitful, which are concerned mainly on linear switched systems and rarely on nonlinear switched systems.Nonlinearities are ubiquitous in physical systems, sometimes in the analysis of systems modeling, nonlinearities are approximated as linearities by linearization techniques and an approximated linear system is obtained. While in some situation where a high performance is expected, nonlinear term can't be ignored in modeling analysis, thus it's of significance to explore the switched systems with nonlinearities. Motivated by this problem, this thesis investigate the fault detection method of switched systems with sector nonlinearities in a constrained network framework. Refer to the achievements existing in literatures on linear switched system and nonlinear switched system, and consider that the system operates under a network where exists data dropouts, then construct an unified framework for this problem, our aim is to study some methods of fault detection for nonlinear switched systems.Firstly, a class of discrete-time switched system model with sector nonlinearity is introduced, and the thesis models the packet dropouts phenomenon in a network system uniformly. Then the problem of designing a fault detector for a nonlinear switched system in a network environment with data dropouts is formulated. The aim is to design a fault detector based on Luenberger Observer, which ensures the augmented filtering error system asymptotically stable and a given l2 gain. This problem is investigated under the stochastic frame, a sufficient condition for the solvability of this problem has been established by using mode-dependent Lyapunov function and some linear matrix inequality techniques, and an observer-based fault detector designing method is given in the form of LMIs(Linear matrix inequalities).A numerical example is included to show the feasibility and effectiveness of the obtained result.Secondly, according to the physical environment described above,a challenge task is to design a fault detector based on Kalman Filter, which ensures the augmented filtering error system asymptotically stable and a given l2 gain. This problem is investigated under the stochastic frame, a sufficient condition for the solvability of this problem has been established by using mode-dependent Lyapunov function and some linear matrix inequality techniques, and an filter-based fault detector designing method is given in the form of LMIs(Linear matrix inequalities).A numerical example is included to show the feasibility and effectiveness of the obtained result.The feature of this thesis lies in that: a sector nonlinearity is introduced into a switched system, a desired filter/observer for fault detection is proposed, then we get a fault detection system under a network where exists data dropouts, by constructing modedependent Lyapunov function, and using average dwell-time methods as well as some LMI techniques, a new method to design the fault detection filter/observer is firstly proposed to deal with the problem in thesis, which provides a way to solve more complex nonlinear switched systems.
Keywords/Search Tags:Sector nonlinearity, nonlinear switched system, fault detection, filter design
PDF Full Text Request
Related items