Research Article  Open Access
Yongchi Zhao, Shengxian Zhuang, Weiming Xiang, Lin Du, "Discretized Lyapunov Function Approach for Switched Linear Systems under Dwell Time Constraint", Abstract and Applied Analysis, vol. 2014, Article ID 905968, 10 pages, 2014. https://doi.org/10.1155/2014/905968
Discretized Lyapunov Function Approach for Switched Linear Systems under Dwell Time Constraint
Abstract
This paper is concerned with the stability and disturbance attenuation properties of switched linear system with dwell time constraint. A novel timescheduled Lyapunov function is introduced to deal with the problems studied in this paper. To numerically check the existence of such timescheduled Lyapunov function, the discretized Lyapunov function technique usually used in timedelay system is developed in the context of switched system in continuoustime cases. Based on discretized Lyapunov function, sufficient conditions ensuring dwelltime constrained switched system global uniformly asymptotically stable are established, then the disturbance attenuation properties in the sense of gain are studied. The main advantage of discretized Lyapunov function approach is that the derived sufficient conditions are convex in subsystem matrices, which makes the analysis results easily used and generalized. Thus, the control synthesis problem is considered. On the basis of analysis results in hand, the control synthesis procedures including both controller and switching law design are unified into onestep method which explicitly facilitates the control synthesis process. Several numerical examples are provided to illustrate the results within our paper.
1. Introduction and Some Preliminaries
The switched systems have emerged as an important class and represent a relatively new and very active area of current research in the field of control systems [1–4]. A switched system is composed of a family of continuous or discretetime subsystems, described by differential or difference equations, respectively, along with a switching rule governing the switching between the subsystems. The motivation for studying such switched systems comes from the fact that switched system can be efficiently used to model many practical systems that are inherently multimodel in the sense that several dynamical subsystem models are required to describe their behavior, such as several realworld, and industrial processes exhibit switched and hybrid nature intrinsically [5–8] and from the fact that switching control strategies also arise in many engineering applications to improve control performance [9–11].
In this paper, we consider the switched linear system described by where is the system state and is the energybounded disturbance input. is a piecewise constant function of time, called switching law or switching signal, which takes value in a finite index set . is the number of subsystems. System matrices , , , and are constant with appropriate dimensions. Let the discontinuity points of be denoted by , and let stand for the initial time by convention; the switching sequence can be described as . Calling the set of all switching policies with dwell time , that is the set of all for which the time interval between successive discontinuities of satisfies , .
Generally, the stability and disturbance attenuation properties are the main concerns in the field of switched system such as [3, 4, 12–23], which are also the main issues considered in this paper. At first, we introduce some relevant conceptions, problems which we concentrate on, and review of several existing results on stability and disturbance attenuation properties.
Definition 1. The equilibrium of system (1) with is globally uniformly asymptotically stable (GUAS) under certain switching signal if, for initial condition , there exist a class function such that the solution of the system satisfies , . (A function is a function if it is strictly increasing and , and also a function is a function if for each fixed the function is a function with respect to , and for each fixed the function is decreasing with respect to and as .)
One of the basic problems for switched system (1) with dwelltime constrained switching signal is to identify or determine the minimal value of dwell time and the set of switching law set such that system (1) with is GUAS, that is, the computation on minimal admissible dwell time guaranteeing GUAS of system (1).
Problem 2. Given switched system (1), find the minimal value of dwell time such that switched system (1) is GUAS for any .
A famous result for above stability analysis problem is given incorporated with multiple Lyapunov function (MLF) formulated as , , where , , which is an efficient stability analysis tool for switched systems. Based on MLF, the following lemma provides sufficient conditions ensuring system (1) GUAS for continuoustime case [24–26].
Lemma 3. Consider the switched linear system (1) with and let , be given constants. If there exist a set of positive definite symmetric matrices , , such that , , and , , , then, system (1) is GUAS for any switching law , where dwell time satisfies .
Lemma 3 provides us a method to compute the admissible dwell time of switched linear system (1); however, the parameters and often need to be chosen manually in advance, and moreover, as the main disadvantage of Lemma 3, the estimation of admissible dwell time by is nonconvex in nature; thus, the obtained minimal admissible dwell time is usually significantly conservative by some inappropriate choices of parameters and in the framework of MLF approach. Hereby, to improve Lemma 3, we have to establish new sufficient conditions that are in convex form essentially.
Then, another basic problem of switched system (1) with dwelltime constrained switching signal is how the norm of changes for any . Here the norm is defined as . The gain performance for switched system (1) is defined as follows.
Definition 4. For a scalar , system (1) is said to be GUAS and has an gain under certain switching signal if, under zero initial condition, system (1) is GUAS when , and the inequality holds.
Our objective is to characterize the relationship between gain and dwell time. Particularly, the minimal value of ensures the gain inequality satisfied is called the induced gain . The problem of computation has to be pointed out on the induced gain versus dwell time which has been viewed as an open problem [27] and not completely solved so far.
Problem 5. Given switched system (1) with a dwell time , determine the induced gain of switched system (1).
A natural solution for above problem is to extend the stability results in Lemma 3 to gain analysis problem by some standard techniques. Unfortunately, Lemma 3 cannot be extended straightforwardly to solve the open problem on the gain performance with dwell time. Under the framework of MLF, it has been proved that the straightforward extension of Lemma 3 only guarantees the undesirable weighted gain [28–30] described as , where . Obviously the weighted gain is not an anticipated performance in both mathematical analysis and practical use. The constant (nonweighted) gain analysis problem still remains unsolved, though some other important advances for constant gain performance have been reached recently. In [19, 20], the authors showed that gain stability can be characterized by the existence of a convex homogeneous (of degree two) Lyapunov function, though the construction of such a storage function (by solving the HamiltonJacobi inequalities) still remains a theoretical challenge. And in [31–33], the corresponding conditions are nonconvex in system matrices, such as terms , , or , , , , which are involved in linear matrix inequities (LMIs) conditions, which significantly restricts their further application such as controller synthesis problem.
Therefore, the key point but remaining open is to find convex sufficient conditions to establish switched system (1) GUAS associated with dwell time and generalize to gain performance to solve the open problem in [27], which is also our main objective in this paper.
The basic structure of our main results for switched system (1) is as follows. At the first step we propose a novel timescheduled Lyapunov function and provide a computable method, called discretized Lyapunov function technique, to determine the timescheduled Lyapunov function in continuoustime case. The second step involves applying discretized Lyapunov function approach into stability analysis and gain performance issues for switched system (1). The main feature of our approach is that the derived sufficient conditions are affine in system matrices, which meets the main objective in our paper. Finally, with our analysis results in hand, we turn to control synthesis and derive explicit controller formulas which ensure stability and performance the switched system in closed loop. In the context of control synthesis, it naturally contains feedback controller and switching law design. Unlike most of previous results in which controller and switching law design are split in two separate steps, a onestep design methodology is presented on the basis of discretized Lyapunov function approach; as far as we know, this is possible to be the first result that unifies the controller and switching law into one step in the synthesis process. Along with this paper, several academic examples are provided to illustrate the advantages of our approach.
Some Notations. The superscript stands for matrix transposition, denotes the dimensional Euclidean space, stands for the set of nonnegative numbers, and represents the set of nonnegative integers. The notations and refer to the Euclidean norm and norm, respectively. In addition, in symmetric block matrices, we use as an ellipsis for the terms that are introduced by symmetry and stands for a blockdiagonal matrix. The notation () means is real symmetric and positive definite (semipositive definite). is the shorthand notation for . stands for identity matrix.
2. Discretized Lyapunov Function
The main idea within this paper is to construct a new type of timescheduled Lyapunov function in the form of where , , is a timescheduled positive definite matrix. Explicitly, the timescheduled Lyapunov function with matrix function , , can provide more choices to construct Lyapunov function and yield less conservative results than MLF, since MLF is only a particular case of , . However, in practice, to numerically check the existence of such matrix function , , especially in continuoustime case, is not an easy task. Thus, in this paper we resort to discretized Lyapunov function technique which has been widely used in timedelay system analysis problems, for example [34, Section 5.7]. The basic idea of discretized Lyapunov function technique is to divide the domain of definition of matrix function , , into finite discrete points or smaller regions, thus reducing the choice of timescheduled Lyapunov function into choosing a finite number of parameters.
In order to establish a computable method to determine the continuous timescheduled Lyapunov function , , we attempt to divide the dwell time interval into a sequence of small segments and employ the discretized Lyapunov function technique commonly used in timedelay system [34, Section 5.7]. The discretized Lyapunov function technique for switched system (1) is given in detail as follows.
We divide the dwelltime interval into segments described as , of equal length , and then , . The continuous matrix function , are chosen to be linear within each segments , . Letting , then since the matrix function is piecewise linear in dwelltime interval ; it can be expressed in terms of the values at dividing points using a linear interpolation formula, that is, for , where . Then the continuous matrix function is completely determined by , , in the dwelltime interval . Afterwards, in the interval , is fixed as a constant matrix . Hence the discretized matrix function , , for continuoustime case is described as and the corresponding discretized Lyapunov function is
Obviously, the number of division segments has to be when the discretized Lyapunov function approach is used. If it is enforced that , by (5), the discretized Lyapunov function , , is reduced to MLF form. Thus, the discretized Lyapunov function is also a generalized form of MLF in continuoustime case.
The discretized Lyapunov function technique provides us a computable method to determine the timescheduled Lyapunov function in continuoustime case, and what is more important, in contrast to the MLF , , in which the matrices , , are completely independent of dwell time , the discretized Lyapunov function is constructed with the aid of information of dwell time , and this unique feature can make the resulting sufficient conditions for GUAS and gain affine in system matrices, which is a convenient tool to deal with the stability and disturbance attenuation problems and will be shown in next sections.
3. Stability Analysis
In this section, our analysis results employ a family of discretized Lyapunov functions in the form of (5) for each subsystem. Motivated by the idea that the value of Lyapunov function is nonincreasing at switching instant [17], the following theorem can be obtained with the help of a family of discretized Lyapunov function in the form of (5).
Theorem 6. Consider the switched linear system (1) with . If there exist a set of matrices , , , such that where , then system (1) is GUAS under switching signal with dwell time .
Proof. Choosing the discretized Lyapunov function for th subsystem , , where matrix function , , is discretized according to (5). Thus, we see . In each discretized segment , the following equation can be derived:
Due to , we have , where . Hence becomes
Moreover, one has
By the linear interpolation relationship as (3), note that for , we see
Thus, from (6) and (7), we get
On the other hand, since , when , it yields , . By (8), , . Thus, we can conclude that
For overall switched system (1), we choose Lyapunov function described by
where and . Then, by the structure of discretized matrix function , , and (9), one has
where and , if system switches from to at switching instant . Combining (15) and (17), is decreasing along with the time and converges to zero as . The GUAS of system (1) under switching signal with dwell time can be established.
Remark 7. Here it has to be noted that the number of discretized matrices , is which has to be prescribed in advance, and different choices of could lead to different analysis results. Roughly speaking, the larger is chosen; a denser division of dwelltime interval can be produced and, intuitively, a less conservative result can be obtained, which will be demonstrated by numerical example later. However, these less conservative results obtained by larger will directly lead to increase of computational complexity.
Remark 8. The following property can hold if Theorem 6 holds for some , then it holds for any . Since Theorem 6 holds for some , we can still choose interval to be divided into same segments , ; then, (6) and (7) can guarantee the . And (8) has , . Thus, , , can be also obtained to establish the stability for any , and the minimal admissible dwell time guaranteeing system (1) GUAS can be estimated by
When we choose indicating , along with (9), Theorem 6 is reduced to common Lyapunov function approach.
Since the discretized Lyapunov function contains the information of dwell time in its own structure, the estimation of dwell time is automatically executed by solving a set of LMIs without any additional nonconvex relations such as as in Lemma 3, where and have to be chosen manually. This feature makes our approach easily used with least conservativeness in the framework of discretized Lyapunov function approach.
Example 9. A continuoustime switched linear system is given as , , where
Hereby, we are focusing on the important parameter concerned with division of dwell time interval in the context of continuoustime case and to show how affects the result obtained by Theorem 6.
In Table 1, we see that the minimal admissible dwell time tends to smaller value as is increased, which implies, as what Remark 7 states, the larger concerned with a denser division of dwell time interval can lead to a less conservative result.

4. Disturbance Attenuation Performance Analysis
As what is discussed in Section 1, the disturbance attenuation property analysis in the sense of gain performance related to dwell time has been viewed as an open problem. The general property relationship between the induced gain and dwell time can be described as a function that maps each dwell time with induced gain of switched system (1), for the dwelltime switching signal . For , at least the following two obvious observations are known from [27]:(i) is monotonically decreasing;(ii) is bounded below by , where is the induced gain ( induced gain) of th subsystem.
The first property is a trivial consequence of the fact that, given two dwell times , we have that . Then the second property results from the fact that, for every switching set , contains all the constant switching signals , .
However, the above observations on function are trivial for solving disturbance attenuation problem, where the quantitative relationship is required. In this section, we are going to present a method to compute quantitatively, which solves disturbance attenuation problem.
Theorem 10. Consider the switched linear system (1). If there exist a set of matrices , , , such that where . Then switched system (1) is GUAS when and has an gain with a dwell time .
Proof. Let and define same as (16) on the basis of discretized Lyapunov function in continuoustime case (5). From (23), it is easy to see that , , where and are defined same as in (17). Then following result can be derived with assumption ; the following derivation can be obtained:
where . For , we seewhere , , , and
Then, by Schur complement formula, (20) and (21) guarantee and for and , respectively. Thus,
And when , (22) also ensures
Thus, combining (27) and (28) implies . Therefore, the prescribed gain performance is guaranteed, and GUAS with is established by , , and , by (23).
Remark 11. The nonincreasing and bounded properties with relationship between induced gain and dwell time hold for continuoustime case through considering optimization problem
Furthermore, similar to the stability analysis case, the related to the division of dwell time interval has to be prespecified, and, obviously, if a larger is chosen which implies a denser division, a more precise characterization can be attained.
Similar to the discussion in Remark 8 for stability, if is a dwell time such that the optimization (29) is feasible, so is any . Furthermore, since for any , we infer that the minimal value of for solving (29) is nonincreasing as dwell time increases, and the minimal admissible dwell time guaranteeing optimization (29) feasible is
In addition, by the wellknown results for a single system, the induced gain of th subsystem can be simply figured out as . (22) for each subsystem. Compared with optimization (29), it is obvious that we have , , which means the gain performance is worsened by the occurrence of switching, and hence the induced gain is bounded below by .
Noting the sufficient conditions in Theorems 6 and 10, the LMIs are all affine in system matrices. As a result, many other existing standard techniques can be easily employed into Theorems 6 and 10; for example, when the system matrices contain uncertainties, the standard techniques such as [35] dealing additive uncertainties or [17] for polytopic uncertainties can be straightforwardly used in Theorems 6 and 10. Thus, the discretized Lyapunov function approach not only obtains results for stability and disturbance attenuation issues, but also provides a general convex tool for dwelltime constrained switched systems.
5. Application in Control Synthesis
Consider the control synthesis problem, the following switched linear system with control input is given as where is the control input, , , and are defined the same as in (1). As for switched system (31), the control synthesis problem basically involves controller and switching law design. By the previous results for dwelltime constrained switched system, such as Lemma 3, due to its nonconvex nature, most of the control synthesis procedures only have to be executed undesirably by two separate steps: (i)design controllers for subsystems which are obtained at first step;(ii)design the switching law; that is, determine the admissible dwell time, for the closedloop system in the second step.
The twostep design method abruptly splits the controller and switching law design into two independently parts, which ought to be related to each other, and thus it often leads to conservative results for control problems under the dwell time constraint.
In this section, we are going to establish a onestep design methodology, by which the controller can be determined with respect to each specific dwell time, or in other words, the controller and dwell time can be determined simultaneously. We note that the information of dwell time is involved within the LMIbased convex sufficient conditions proposed in previous sections; this important feature could inspire us to use the discretized Lyapunov function techniques to build the onestep design framework.
Hereby, we are going to formulate the controller of discretized timescheduled structure; then, the controller and switching law can be obtained by one unified step which explicitly facilitates the control synthesis process. Assuming the switching between system and the corresponding controller is synchronous, the modedependent timescheduled state feedback controller which we are interested in is given in the form of where , , are timescheduled gains that have to be determined. After substituting controller (32) into system (31), the closedloop system becomes
As far as we know, due to the unsolved open problem on gain, few results about relevant control problem under dwell time constraint have been reported so far, though there exist numerous papers which are concerned with the weighted performance based on Lemma 3. Some previous results [26, 36], claimed that the control performance can be obtained by a straightforward extension from Lemma 3. Unfortunately, these results suffer from a mistake that the desired performance should be replaced by weighted performance; readers can refer to the recent comment paper [30]; however, the nonweighted control problem performance related to dwell time still remains unsolved.
In this section, inspired by the discretized Lyapunov function approach presented in previous section, the sufficient conditions for the existence of controller of switched system (31).
Motivated by the discretized Lyapunov function technique in continuoustime case, we define the following two timescheduled matrices and , , which is discretized as follows: where , , , and , , , are defined the same as in (3). Then, with the aid of above discretized matrices, the following theorem can be obtained.
Theorem 12. Consider switched system (1) with dwell time . If there exist a set of matrices , , , , such that where , , , and . Then the closedloop system (33) with controller (32) is GUAS when and has an gain , and the controller gain is given as where .
Proof. From the guidelines in Theorem 10, the key point is to prove the following inequality:
where and . We can choose , where , , is defined by (34). And using the fact , (41) is equivalent to
Multiplying both sides of the previous LMI by , and due to the controller structure (40), we have , where is given as (35); substituting it into (42), it becomes
Moreover, by the similar guidelines in Theorem 6, we observe that . Thus, due to the structure of discretized and , it is guaranteed that (43) can be established by (36)–(38).
Afterwards, by and and according to Schur complement formula, it is noted that
Then, together with , , , we can conclude that the closedloop system (33) is GUAS and has an gain with the controller gain (40).
Remark 13. The philosophy of onestep method is also applicable in continuoustime case with Theorem 12. Once the parameter is chosen, the controller gains are determined with respect to a specific dwell time by one step via solving LMIs (36)–(39). And the controller with optimal control performance can be computed by with respect to a specific dwell time and prescribed . Similar to the discussion for disturbance attenuation analysis, a larger will yield a less conservative design result.
Example 14. The switched system considered is composed of two subsystems. The system matrices are given as follows: , , are same as the example in Section 3, and , , , , and , are given as The control problem will be considered in the following example involved. The computation on optimal control performance versus dwell time with difference is shown in Figure 1.
The relationship between dwell time and induced gain is characterized in Figure 1. From the numerical results in Figure 1, we see following facts:(i)the same monotonically decreasing property holds with any choices of ;(ii)better results such as the smaller admissible dwell time and smaller disturbance attenuation level can be achieved by a larger , which implies a denser division of dwell time interval leads to a less conservative analysis result.
Taking a specific dwell time with , the timescheduled gains , with optimal performance are determined by where , and and , and , can be easily obtained by solving optimization problem .(36)–(39), which are omitted here.
6. Conclusions
The stability and disturbance attenuation properties for switched system with dwell time constraint are investigated in this paper. By introducing the discretized Lyapunov function technique, sufficient convex conditions ensure that the switched system GUAS are derived for continuoustime case. Then, the idea of discretized Lyapunov function is further extended to disturbance attenuation performance analysis in the sense of gain, which has been viewed as an open problem not completely solved so far. Due to the convex feature of analysis results on the basis of discretized Lyapunov function, the control synthesis problem is taken into consideration, and a onestep design method is proposed, which has obvious advantages over conventional twostep design method. By onestep design method, the controllers for subsystems and the switching law can be determined simultaneously, which facilitates the design process significantly. As what has been discussed in this paper, the discretized Lyapunov function technique can not only derive results for stability and disturbance attenuation issues, but also provide a general convex tool for dwelltime constrained switched systems, which can be used to studied other issues for switched systems in our future work.
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
Acknowledgments
This work is supported by Sichuan Provincial Education Department (no. 14ZB0266) and the School of Education Reform Project mnuJY1408.
References
 D. Liberzon, Switching in Cystems and Control, Birkhäuser, Boston, Mass, USA, 2003. View at: Publisher Site  MathSciNet
 Z. Sun and S. Ge, Switched Linear SystemsControl and Design, Springer, London, UK, 2005.
 R. A. Decarlo, M. S. Branicky, S. Pettersson, and B. Lennartson, “Perspectives and results on the stability and stabilizability of hybrid systems,” Proceedings of the IEEE, vol. 88, no. 7, pp. 1069–1082, 2000. View at: Publisher Site  Google Scholar
 H. Lin and P. J. Antsaklis, “Stability and stabilizability of switched linear systems: a survey of recent results,” IEEE Transactions on Automatic Control, vol. 54, no. 2, pp. 308–322, 2009. View at: Publisher Site  Google Scholar  MathSciNet
 A. Balluchi, M. D. Benedetto, C. Pinello, C. Ross, and A. SangiovanniVincentelli, “Cutoff in engine control: a hybrid system approach,” in Proceedings of the 36th IEEE Conference on Decision and Control, pp. 4720–4725, San Diego, Calif, USA, December 1997. View at: Google Scholar
 B. E. Bishop and M. W. Spong, “Control of redundant manipulators using logicbased switching,” in Proceedings of the 37th IEEE Conference on Decision and Control (CDC '98), pp. 1488–1493, Tampa, Fla, USA, December 1998. View at: Google Scholar
 W. Zhang, M. S. Branicky, and S. M. Phillips, “Stability of networked control systems,” IEEE Control Systems Magazine, vol. 21, no. 1, pp. 84–99, 2001. View at: Publisher Site  Google Scholar
 I. Kolmanovsky and S. Jing, “A multimode switchingbased command tracking in network controlled systems with pointwiseintime constraints and disturbance inputs,” in Proceedings of the 6th World Congress on Intelligent Control and Automation (WCICA '06), pp. 199–204, Dalian, China, June 2006. View at: Publisher Site  Google Scholar
 K. S. Narendra, O. A. Driollet, M. Feiler, and K. George, “Adaptive control using multiple models, switching and tuning,” International Journal of Adaptive Control and Signal Processing, vol. 17, no. 2, pp. 87–102, 2003. View at: Publisher Site  Google Scholar  Zentralblatt MATH
 B. CastilloToledo, S. Di Gennaro, A. G. Loukianov, and J. Rivera, “Hybrid control of induction motors via sampled closed representations,” IEEE Transactions on Industrial Electronics, vol. 55, no. 10, pp. 3758–3771, 2008. View at: Publisher Site  Google Scholar
 C. Sreekumar and V. Agarwal, “A hybrid control algorithm for voltage regulation in DCDC boost converter,” IEEE Transactions on Industrial Electronics, vol. 55, no. 6, pp. 2530–2538, 2008. View at: Publisher Site  Google Scholar
 R. Shorten, F. Wirth, O. Mason, K. Wulff, and C. King, “Stability criteria for switched and hybrid systems,” SIAM Review, vol. 49, no. 4, pp. 545–592, 2007. View at: Publisher Site  Google Scholar  MathSciNet
 M. S. Branicky, “Multiple Lyapunov functions and other analysis tools for switched and hybrid systems,” IEEE Transactions on Automatic Control, vol. 43, no. 4, pp. 475–482, 1998. View at: Publisher Site  Google Scholar  MathSciNet
 K. S. Narendra and J. A. Balakrishnan, “A common Lyapunov function for stable LTI systems with commuting Amatrices,” IEEE Transactions on Automatic Control, vol. 39, no. 12, pp. 2469–2471, 1994. View at: Publisher Site  Google Scholar  MathSciNet
 J. Daafouz, P. Riedinger, and C. Iung, “Stability analysis and control synthesis for switched systems: a switched Lyapunov function approach,” IEEE Transactions on Automatic Control, vol. 47, no. 11, pp. 1883–1887, 2002. View at: Publisher Site  Google Scholar  MathSciNet
 W. Xiang and J. Xiao, “Stabilization of switched continuoustime systems with all modes unstable via dwell time switching,” Automatica, vol. 50, no. 3, pp. 940–945, 2014. View at: Publisher Site  Google Scholar  MathSciNet
 L. I. Allerhand and U. Shaked, “Robust stability and stabilization of linear switched systems with dwell time,” IEEE Transactions on Automatic Control, vol. 56, no. 2, pp. 381–386, 2011. View at: Publisher Site  Google Scholar  MathSciNet
 J. C. Geromel and P. Colaneri, “Stability and stabilization of discrete time switched systems,” International Journal of Control, vol. 79, no. 7, pp. 719–728, 2006. View at: Publisher Site  Google Scholar  Zentralblatt MATH  MathSciNet
 K. Hirata and J. P. Hespanha, “ℒ2induced gain analysis for a class of switched systems,” in Proceeding of thr 48th IEEE Conference on Decision and Control held jointly with the 28th Chinese Control Conference (CDC/CCC '09), pp. 2138–2143, Shanghai, China, December 2009. View at: Publisher Site  Google Scholar
 K. Hirata and J. Hespanha, “${\mathcal{L}}_{2}$induced gain analysis of switched linear systems via finitely parametrized storage functions,” in Proceedings of the 29th American Control Conference (ACC '10), pp. 4064–4069, Baltimore, Md, USA, June 2010. View at: Google Scholar
 W. Xiang and J. Xiao, “Stability analysis and control synthesis of switched impulsive systems,” International Journal of Robust and Nonlinear Control, vol. 22, no. 13, pp. 1440–1459, 2012. View at: Publisher Site  Google Scholar  MathSciNet
 W. Xiang, J. Xiao, and M. N. Iqbal, “Asymptotic stability, ${\mathcal{l}}_{2}$ gain, boundness analysis, and control synthesis for switched systems: a switching frequency approach,” International Journal of Adaptive Control and Signal Processing, vol. 26, no. 4, pp. 350–373, 2012. View at: Publisher Site  Google Scholar  MathSciNet
 M. Margaliot and J. P. Hespanha, “Rootmeansquare gains of switched linear systems: a variational approach,” Automatica, vol. 44, no. 9, pp. 2398–2402, 2008. View at: Publisher Site  Google Scholar  MathSciNet
 A. S. Morse, “Supervisory control of families of linear setpoint controllers. I. Exact matching,” IEEE Transactions on Automatic Control, vol. 41, no. 10, pp. 1413–1431, 1996. View at: Publisher Site  Google Scholar  MathSciNet
 J. P. Hespanha and A. S. Morse, “Stability of switched systems with average dwelltime,” in Proceedings of the 38th IEEE Conference on Decision and Control (CDC '99), pp. 2655–2660, Phoenix, Ariz, USA, December 1999. View at: Google Scholar
 L. Zhang and P. Shi, “Stability, l_{2}gain and asynchronous H_{∞} control of discretetime switched systems with average dwell time,” IEEE Transactions on Automatic Control, vol. 54, no. 9, pp. 2192–2199, 2009. View at: Publisher Site  Google Scholar  MathSciNet
 J. Hespanha, “${L}_{2}$induced gains of switched linear systems,” in Unsolved Problems in Mathematical Systems and Control Theory, V. D. Blondel and A. Megretski, Eds., pp. 131–133, Princeton University Press, Princeton, NJ, USA, 2003. View at: Google Scholar
 G. Zhai, B. Hu, K. Yasuda, and A. N. Michel, “Disturbance attenuation properties of timecontrolled switched systems,” Journal of the Franklin Institute, vol. 338, no. 7, pp. 765–779, 2001. View at: Publisher Site  Google Scholar  MathSciNet
 G. Zhai, B. Hu, K. Yasuda, and A. N. Michel, “Qualitative analysis of discretetime switched systems,” in Proceedings of the American Control Conference, pp. 1880–1885, May 2002. View at: Publisher Site  Google Scholar
 W. Xiang and J. Xiao, “Discussion on stability, l_{2}gain and asynchronous ${H}_{\infty}$ control of discretetime switched systems with average dwell time,” IEEE Transactions on Automatic Control, vol. 57, no. 12, pp. 3259–3261, 2012. View at: Publisher Site  Google Scholar  MathSciNet
 C. Cai, “Dwelltime approach to inputoutput stability properties for a class of discretetime dynamical systems,” Systems & Control Letters, vol. 60, no. 6, pp. 383–389, 2011. View at: Publisher Site  Google Scholar  MathSciNet
 J. C. Geromel and P. Colaneri, “${\mathcal{H}}_{\infty}$ and dwell time specifications of continuoustime switched linear systems,” IEEE Transactions on Automatic Control, vol. 55, no. 1, pp. 207–212, 2010. View at: Publisher Site  Google Scholar  MathSciNet
 P. Colaneri, P. Bolzern, and J. C. Geromel, “Root mean square gain of discretetime switched linear systems under dwell time constraints,” Automatica, vol. 47, no. 8, pp. 1677–1684, 2011. View at: Publisher Site  Google Scholar  MathSciNet
 K. Gu, V. L. Kharitonov, and J. Chen, Stability of TimeDelay Systems, Springer, Berlin , Germany, 2003. View at: MathSciNet
 W. Xiang, J. Xiao, and M. N. Iqbal, “Robust observer design for nonlinear uncertain switched systems under asynchronous switching,” Nonlinear Analysis: Hybrid Systems, vol. 6, no. 1, pp. 754–773, 2012. View at: Publisher Site  Google Scholar  MathSciNet
 B. Lu and F. Wu, “Switching LPV control designs using multiple parameterdependent Lyapunov functions,” Automatica, vol. 40, no. 11, pp. 1973–1980, 2004. View at: Publisher Site  Google Scholar  MathSciNet
Copyright
Copyright © 2014 Yongchi Zhao et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.