Reachable set bounding for homogeneous nonlinear systems with delay and disturbance is studied. By the usage of a new method for stability analysis of positive systems, an explicit necessary and sufficient condition is first derived to guarantee that all the states of positive homogeneous time-delay systems with degree p>1 converge asymptotically within a specific ball. Furthermore, the main result is extended to a class of nonlinear time variant systems. A numerical example is given to demonstrate the effectiveness of the obtained results.
National Natural Science Foundation of China618731101. Introduction
Recent years have witnessed a rapid development of reachable set bounding for linear systems in [1–11], to name a few. In most of existing references, the traditional Lyapunov-Krasovskii function method is most commonly used. However, such a method is usually difficult to derive explicit conditions for reachable set estimation of nonlinear systems with delay and disturbance.
Due to the ubiquitous existence of time delay in practical engineering and its adverse effect on stability [12–15] and oscillation [16–19], it has attracted wide attention in recent years. So far, less attention has been paid to reachable set bounding for nonlinear time-delay systems. Such a problem was discussed in [20, 21] for certain nonlinear perturbed systems with delay, where the involved nonlinear terms satisfy a linear growth condition. Reachable set bounding for continuous-time and discrete-time homogeneous time-delay positive systems of degree one was studied in [22]. The decay rates of homogeneous positive systems of any degree with time-varying delays were given in [23]. Recently, the same problem was considered in [24] for homogeneous positive systems of degree p>1, while time delay was not taken into consideration. The problem of reachable set estimation of switched positive systems with discrete and distributed delays subject to bounded disturbances was investigated in [25].
Positive systems are dynamical systems whose states remain nonnegative whenever the initial states are nonnegative ([26, 27]). In view of the special structure of positive systems, a special method was commonly used for stability analysis of positive systems in [28–33], which is different from the traditional Lyapunov-Krasovskii function method.
Motivated by the work in [23, 24], we study in this paper reachable set bounding for homogeneous nonlinear time-delay systems with bounded disturbance. By developing the methods used in [23, 24], we first establish a necessary and sufficient condition such that all the solutions of positive homogeneous time-delay systems with degree p>1 converge asymptotically within a specific ball, which contains those results in [23, 24] in special cases. The main result is also applied to certain nonlinear time variant systems with delay and disturbance.
Throughout this paper, Rn is the set of n-dimensional real vectors. Denote by xi the ith coordinate of x∈Rn for i∈〈n〉={1,2,…,n}. Given x,y∈Rn, say x≻y (or y≺x) if xi>yi, x⪰y (or y⪯x) if xi≥yi, i∈〈n〉. Denote R+n={x∈Rn:x⪰0}. For x=(xi)∈Rn, denote x=(xi)∈R+n and x∞=maxi∈〈n〉|xi|. Let B(ε)={x∈Rn|x∞≤ε}, where ε>0 is a constant. For given r>0, denote BFr([0,∞],Rn)={ω:[0,∞]→Rn|ω(t)∞≤r,∀t≥0}. An n×n-dimensional matrix A is called Metzler if all its off-diagonal entries are nonnegative.
2. Preliminaries
In this paper, nonlinear time-delay systems of the form(1)x˙t=fxt+gxt-τt+ωt,t≥0xt=φt,t∈-h,0,are investigated, where x(t)∈Rn is the state vector, f,g:Rn→Rn are continuous vector functions satisfying f(0)=g(0)=0, τ(t) is a time delay satisfying 0≤τ(t)≤h, h>0 is a constant, ω(t)∈BFr([0,∞],Rn) is the disturbance, and the initial state φ(t):[-h,0]→Rn is continuous. Note that when τ(t)≡0, system (1) takes the form of the system considered in [24].
The following definitions and lemma in [34] will be required.
Definition 1.
Assume that f:Rn→Rn is continuous on Rn and continuously differentiable on Rn∖{0}. The vector function f is called cooperative if the Jacobian matrix (∂f/∂x)(x), x∈Rn∖{0}, is Metzler.
Definition 2.
A vector function f:Rn→Rn is called homogeneous of degree p>0 if f(λx)=λpf(x), x∈Rn, λ>0.
Definition 3.
A vector function g:Rn→Rn is called order-preserving on R+n provided that g(x)⪰g(y), where x,y∈R+n, x⪰y.
Lemma 4.
A cooperative vector function f satisfies fi(u)≥fi(v), where u,v∈Rn∖{0}, u⪰v, ui=vi, i∈〈n〉.
In this paper, we need the following assumptions:
f and g are continuously differentiable on Rn∖{0} and homogeneous of degree p>1;
f is cooperative and g is order-preserving on R+n;
ω(t)⪰0 for t≥0.
Following the proof given in [22], we can easily obtain the following lemma.
Lemma 5.
System (1) is positive under assumptions (H2) and (H3).
3. Main ResultsTheorem 6.
Suppose that (H1)-(H3) are valid. Then, we have the following equivalent statements:
There is an n-dimensional vector v≻0 satisfying f(v)+g(v)≺0.
The solution x(t) of system (1) satisfies(2)xt∞≤α+β+γt-1/p-1
for any t≥0, any initial state φ(t)∈C([-h,0],R+n), any disturbance ω(t)∈BFr([0,∞],R+n), and any bounded delay τ(t), where α, β, and γ are appropriate nonnegative constants dependent on r, h, and the initial state φ, and α=0 if r=0.
In addition, if condition (i) holds, α, β, and γ can be chosen as follows:(3)α=θρ,β=Kρ1-p,γ=p-1ηρ1-p,where ρ=maxi∈〈n〉vi,(4)θ=r-maxi∈nfiv+giv1/p,K=0,φv≤θ,φvp-θp1/p,φv>θ,φv=maxi∈n,t∈-h,0|φi(t)|/vi,η satisfies 0<η<mini∈〈n〉ηi, and ηi satisfies the following equation:(5)fivvi+givvi1+p-1Kp-1ηihp/p-1+ηi=0,i∈n.
Proof.
(i)⇒(ii) Given the initial state φ∈C([-h,0],R+n), from Lemma 5 we have x(t)⪰0, t≥0. Based on definitions of K and φv, we have(6)xitvi≤θp+Kp1/p,t∈-h,0,i∈n.Set(7)zit=xitvii-θp+K1-p+p-1ηt-p/p-11/p,t≥0,i∈n,xitvi-θp+Kp1/p,t∈-h,0,i∈n.Then (6) and (7) yield zi(t)≤0, t∈[-h,0], i∈〈n〉. Next, we show that zi(t)≤0 for i∈〈n〉 and t≥0. If it is not true, there is a constant t∗≥0 and an index k∈〈n〉 guaranteeing zi(t)≤0 for i∈〈n〉, t∈[0,t∗], and zk(t∗)=0. Therefore,(8)z˙kt∗≥0,(9)xitvi≤θp+K1-p+p-1ηt-p/p-11/p,t∈0,t∗,i∈n.(10)xkt∗vk=θp+K1-p+p-1ηt∗-p/p-11/p.Using Lemma 4 and the homogeneity of f, we get from (9) and (10) that(11)fkxt∗≤fkθp+K1-p+p-1ηt∗-p/p-11/pv=θp+K1-p+p-1ηt∗-p/p-1fkv=θpfkv+K1-p+p-1ηt∗-p/p-1fkv.For the case when τ(t∗)≤t∗, it holds that(12)xit∗-τt∗vi≤θp+K1-p+p-1ηt∗-τt∗-p/p-11/p,i∈n.Considering g is homogeneous and order-preserving, we conclude(13)gkxt∗-τt∗≤gkθp+K1-p+p-1ηt∗-τt∗-p/p-11/pv=θp+K1-p+p-1ηt∗-τt∗-p/p-1gkv=θpgkv+K1-p+p-1ηt∗-τt∗-p/p-1gkv.Note that(14)K1-p+p-1ηt∗-τt∗-p/p-1=K1-p+p-1ηt∗-p/p-1×ηp-1τt∗K1-p+ηp-1t∗-τt∗+1p/p-1≤K1-p+p-1ηt∗-p/p-1×1+p-1Kp-1ηhp/p-1.We further get from (13) and (14) that(15)gkxt∗-τt∗≤θpgkv+K1-p+ηp-1t∗-p/p-11+p-1Kp-1ηhp/p-1gkv.For the case when τ(t∗)>t∗, it holds that zi(t∗-τ(t∗))≤0; i.e.,(16)xit∗-τt∗vi≤θp+Kp1/p,i∈n.It thus follows that(17)gkxt∗-τt∗≤θp+Kpgkv=θpgkv+K1-p+p-1ηt∗-p/p-1KpK1-p+p-1ηt∗p/p-1gkv≤θpgkv+K1-p+p-1ηt∗-p/p-11+p-1Kp-1ηhp/p-1gkv.Next, we can conclude from (1) and (7) that(18)z˙kt∗=fkxt∗+gkxt∗-τt∗+wkt∗vk+ηθp+K1-p+p-1ηt∗-p/p-11-p/pK1-p+p-1ηt∗1-2p/p-1≤fkxt∗+gkxt∗-τt∗+wkt∗vk+ηK1-p+p-1ηt∗-p/p-1.Consequently, (11), (15), (17), and (18) imply that(19)z˙kt∗≤θpfkv+gkv+rvk+K1-p+p-1ηt∗-p/p-1×fkvvk+gkvvk1+p-1Kp-1ηhp/p-1+η.On the other hand, the definitions of θ and η yield that (20)θpfkv+gkv+r≤θpmaxi∈nfiv+giv+r=0,and (21)fkvvk+gkvvk1+p-1Kp-1ηhp/p-1+η<0.Combining this with (19), we have z˙k(t∗)<0, which contradicts (8). Therefore, zi(t)≤0, t≥0, i∈〈n〉; i.e., (22)xitvi≤θp+K1-p+p-1ηt-p/p-11/p,t≥0,i∈n.From the well-known inequality (a+b)q≤aq+bq for a,b≥0 and 0<q<1, we further get (23)xitvi≤θ+K1-p+p-1ηt-1/p-1,t≥0,i∈n.It implies (2).
(ii)⇒(i) For the particular case when r=0 and h=0, system (1) reduces to(24)x˙t=fxt+gxt,t≥0.Given the initial condition x(0)⪰0, each solution of system (24) satisfies (25)xt∞≤β+γt-1/p-1.That is, system (24) is asymptotically stable. Based on Proposition 4.1 in [35], there is a vector v≻0 such that f(v)+g(v)≺0. The proof is complete.
Remark 7.
It can be seen from Theorem 6 that the bound of the reachable set is determined by the bound of disturbances, the choice of v, and the value of p. When the bound of disturbances and the value of p are given, an appropriate vector v can be chosen to guarantee a minimal bound of the reachable set by solving the following nonlinear optimization problem: minv≻0θ subject to f(v)+g(v)≺0, where θ is defined as in Theorem 6.
Remark 8.
If ω(t)=0 for t≥0, then Theorem 6 reduces to the main result given in [23]. If g(x)=0 for x∈Rn, then Theorem 6 reduces to the main result given in [24].
Finally, consider the following nonlinear time-varying system(26)x˙t=f~t,xt+g~t,xt-τt+ωt,t≥0xt=φt,t∈-h,0,where x(t), τ(t), ω(t), and φ(t) are the same as in (1), and f~,g~:[0,∞)×Rn→Rn are vector functions satisfying f~(t,0)=g~(t,0)=0.
Suppose that f~ and g~ satisfy the following assumption:
(H4) f~ and g~ are continuous on [0,∞)×Rn, continuously differentiable with respect to x on Rn∖{0}, and there are vector functions f and g satisfying (H1) and (H2), and for xi≠0,(27)f~it,xsignxi≤fix,g~it,x≤gix,t≥0,i∈n.
Without the restriction on the disturbance that ω(t)⪰0 for t≥0, we can get the following reachable set bounding criterion for system (26).
Theorem 9.
Suppose that (H4) is valid. If there is an n-dimensional vector v≻0 such that f(v)+g(v)≺0, the solution of system (26) satisfies (2), where constants α, β, and γ are defined by (3).
Proof.
Set(28)yit=xitvi-θp+K1-p+p-1ηt-p/p-11/p,t≥0,i∈n,xitvi-θp+Kp1/p,t∈-h,0,i∈n.Based on definitions of K and φv, it holds that yi(t)≤0, t∈[-h,0], i∈〈n〉. For the case when xi(t)≠0, t≥0, notice that(29)y˙it=D-xit=x˙itsignxit≤fixt+gixt-τt+wit,i∈n.Here D- denotes the left derivative. Similar to the analysis in Theorem 6, it is not difficult to conclude that yi(t)≤0, t≥0, i∈〈n〉. Consequently, (2) holds. The proof is complete.
4. Numerical Example
Consider system (1) with(30)fx1,x2=-362-2x13/2x23/2-x13+x2331,gx1,x2=0.2x23/20.4x13/2,ωt=0.05sint0.04cost,τt=5+sint,t≥0.It is easy to verify that assumptions (H1)-(H2) hold. Let v=(1,1)T. Then f(v)+g(v)≺0. By a direct calculation, it yields h=6, r=0.05, α≈0.1345, and γ≈0.11.
We conclude from Theorem 6 that there is a ball B(0.1345) such that all the states of system (1) converge asymptotically within it. Given the initial state φ(t)=(1,1)T, t∈[-6,0], noting that φv=1 and β≈1.0749, solution (1) satisfies (31)xt∞≤0.1345+1.0749+0.11t-2,t≥0.Figure 1 presents the simulation.
The states of system (1).
5. Conclusion
This paper has been concerned with reachable set bounding for homogeneous nonlinear time-delay systems with disturbance. We not only derive explicit reachable set bounding criterion independent of delay, but also estimate the decay rate. It will be interesting to extend our work to the case of unbounded delays and discrete-time systems.
Data Availability
The data used to support the findings of this study are included within the article.
Conflicts of Interest
We declare that there are no conflicts of interest regarding this paper.
Acknowledgments
This work is supported by the National Natural Science Foundation of China under grant 61873110.
FridmanE.ShakedU.On reachable sets for linear systems with delay and bounded peak inputs200339112005201010.1016/s0005-1098(03)00204-8MR2142837Zbl1037.930422-s2.0-0141754135KimJ.-H.Improved ellipsoidal bound of reachable sets for time-delayed linear systems with disturbances20084411294029432-s2.0-5504911174010.1016/j.automatica.2008.03.015Zbl1152.93309ZuoZ. Q.HoD. W. C.WangY. J.Reachable set bounding for delayed systems with polytopic uncertainties: the maximal Lyapunov-Krasovskii functional approach201046594995210.1016/j.automatica.2010.02.022MR2877171Zbl1191.931272-s2.0-77950626595KwonO. M.LeeS. M.ParkJ. H.On the reachable set bounding of uncertain dynamic systems with time-varying delays and disturbances2011181173735374810.1016/j.ins.2011.04.045MR2806097Zbl1243.930152-s2.0-79957584735NamP. T.PathiranaP. N.Further result on reachable set bounding for linear uncertain polytopic systems with interval time-varying delays20114781838184110.1016/j.automatica.2011.05.003MR2886793Zbl1226.930242-s2.0-79960906598NamP. T.PathiranaP. N.TrinhH.Exponential convergence of time-delay systems in the presence of bounded disturbances2013157384385210.1007/s10957-012-0240-1MR3047033Zbl1271.930672-s2.0-84876861907ThatN. D.NamP. T.HaQ. P.Reachable set bounding for linear discrete-time systems with delays and bounded disturbances201315719610710.1007/s10957-012-0179-2MR3034045Zbl1264.930202-s2.0-84875393245ZuoZ.ChenY.WangY.HoD. W.ChenM. Z.LiH.A note on reachable set bounding for delayed systems with polytopic uncertainties201335071827183510.1016/j.jfranklin.2013.04.025MR30730932-s2.0-84893671024Zbl1392.93004LamJ.ZhangB.ChenY.XuS.Reachable set estimation for discrete-time linear systems with time delays201310.1002/rnc.3086ZhangB.LamJ.XuS.Reachable set estimation and controller design for distributed delay systems with bounded disturbances201435163068308810.1016/j.jfranklin.2014.02.007MR32010202-s2.0-84900031113Zbl1290.93020HienL. V.TrinhH. M.A new approach to state bounding for linear time-varying systems with delay and bounded disturbances20145061735173810.1016/j.automatica.2014.04.025MR3214923Zbl1296.930462-s2.0-84899778983QiJ.SunY.Global exponential stability of certain switched systems with time-varying delays201326776076510.1016/j.aml.2013.02.010MR3046391Zbl1307.933452-s2.0-84876694903ZongG.RenH.HouL.Finite-time stability of interconnected impulsive switched systems201610664865410.1049/iet-cta.2015.0617MR34690682-s2.0-84962052407LiY.SunY.MengF.New criteria for exponential stability of switched time-varying systems with delays and nonlinear disturbances201726284291MR369534210.1016/j.nahs.2017.06.0072-s2.0-85025167018Zbl1373.93271LiY.SunY.MengF.TianY.Exponential stabilization of switched time-varying systems with delays and disturbances2018324131140MR374366210.1016/j.amc.2017.12.011LiuH.MengF.LiuP.Oscillation and asymptotic analysis on a new generalized Emden-Fowler equation201221952739274810.1016/j.amc.2012.08.106MR2988149Zbl1308.340842-s2.0-84868213011ShaoJ.ZhengZ.MengF.Oscillation criteria for fractional differential equations with mixed nonlinearities201332310.1186/1687-1847-2013-323MR3213901Zbl1391.34069ShaoJ.MengF.ZhengZ.Oscillation theorems for linear matrix Hamiltonian systems201525340240910.1016/j.amc.2014.12.101MR3312729Zbl1338.340742-s2.0-85028141575LiuH.MengF.Interval oscillation criteria for second-order nonlinear forced differential equations involving variable exponent2016Paper No. 291, 1410.1186/s13662-016-0983-3MR35725472-s2.0-84995684967Zbl06988346NamP. T.PathiranaP. N.TrinhH.Reachable set bounding for nonlinear perturbed time-delay systems: the smallest bound201543687110.1016/j.aml.2014.11.015MR3305631Zbl1310.930292-s2.0-84919742962ZhuX.SunY.XieX.-J.State bounding for nonlinear time-varying systems with delay and disturbance2018355168213822410.1016/j.jfranklin.2018.09.021MR3864043Zbl1398.93143ZhangN.SunY.ZhaoP.State bounding for homogeneous positive systems of degree one with time-varying delay and exogenous input201735472893290410.1016/j.jfranklin.2017.01.031Zbl06726127DongJ. G.On the decay rates of homogeneous positive systems of any degree with time-varying delays201560112983298810.1109/TAC.2015.2414793MR3419586Zbl1360.93494SunYuangongMengFanweiReachable Set Estimation for a Class of Nonlinear Time-Varying Systems20172017610.1155/2017/58763715876371Zbl1373.93061TianD.LiuS.SunY.XiaJ.Reachable set estimation for switched positive systems with mixed time-varying delays and bounded disturbances201812152003200910.1049/iet-cta.2018.0037MR3887528FarinaL.RinaldiS.2000New York, NY, USAWiley10.1002/9781118033029MR1784150KaczorekT.2002New York, NY, USASpringer-VerlagZbl1005.68175LiuX.YuW.WangL.Stability analysis of positive systems with bounded time-varying delays200956760060410.1109/TCSII.2009.20233052-s2.0-68249138708LiuX.YuW.WangL.Stability analysis for continuous-time positive systems with time-varying delays20105541024102810.1109/TAC.2010.2041982MR26544492-s2.0-77950791828Zbl1368.93600FeyzmahdavianH. R.CharalambousT.JohanssonM.Exponential stability of homogeneous positive systems of degree one with time-varying delays20145961594159910.1109/tac.2013.22927392-s2.0-84901405909SunY.TianY.XieX.-J.Stabilization of positive switched linear systems and its application in consensus of multiagent systems201762126608661310.1109/TAC.2017.2713951MR3743547SunYuangongWuZhaorongMengFanweiCommon Weak Linear Copositive Lyapunov Functions for Positive Switched Linear Systems20182018710.1155/2018/13659601365960ZhangN.SunY.MengF.State bounding for switched homogeneous positive nonlinear systems with exogenous input20182936337210.1016/j.nahs.2018.03.004MR3795607Zbl1388.93049SmithH. L.1995BokharaieV. S.MasonO.WirthF.Stability and positivity of equilibria for subhomogeneous cooperative systems201174176416642610.1016/j.na.2011.06.023MR2833425