数学物理学报  2017, Vol. 37 Issue (4): 684-697   PDF    
扩展功能
加入收藏夹
复制引文信息
加入引用管理器
Email Alert
RSS
本文作者相关文章
刘世芳
马巧珍
具有历史记忆的阻尼吊桥方程强全局吸引子的存在性
刘世芳, 马巧珍     
西北师范大学数学与统计学院 兰州 730070
摘要:该文研究具有历史记忆的阻尼吊桥方程解的长时间行为,利用收缩函数的方法获得了解在强拓扑空间中全局吸引子的存在性.
关键词吊桥方程    全局吸引子    有界吸收集    历史记忆    
Existence of Strong Global Attractors for Damped Suspension Bridge Equations with History Memory
Liu Shifang, Ma Qiaozhen     
College of Mathematics and Statistics, Northwest Normal University, Lanzhou 730070
Abstract: In this paper, we study the long-time dynamical behavior of the solution for damped suspension bridge equations with history memory. We prove the existence of the global attractors in a strong space by using the contraction function method.
Key words: Suspension bridge equations     Global attractors     A bounded absorbing set     History memory    
1 引言

本文考虑如下带有线性记忆的阻尼吊桥方程

$ \left\{\begin{array}{ll} u_{tt}+\nu u_{t}+\alpha\triangle^{2}u-\int^{\infty}_{0}\mu(s)\Delta^{2}u(t-s){\rm d}s+ku^{+}+f(u)=g(x), ~(x, t)\in\Omega\times{\Bbb R}^{+}, \\[8pt] u=0, \triangle u=0, ~x\in\Gamma, ~t\in{\Bbb R}, \\ u(x, \tau)=u_{0}(x, \tau), u_{t}(x, \tau)=\partial_{t}u_{0}(x, \tau)=u_{1}(x, \tau), ~(x, \tau)\in\Omega\times(-\infty, 0] \end{array} \right. $ (1.1)

解的渐近性态, 其中$\Omega$${\Bbb R}^{2}$中具有光滑边界$\Gamma$的有界区域. $u(x, t)$是未知函数, 代表了桥面在竖直方向的振动; $\alpha$是正常数; $ku^{+}$表示恢复力, $k>0$是弹性系数, $u^{+}=\max\{u, 0\}$$u$的正部, $u_{0}:\Omega\times(-\infty, 0]\rightarrow{\Bbb R}$表示$u$的过去历史; $\mu$是记忆核; $\nu u_{t}$表示粘弹性阻尼, $\nu>0$是一个给定的正常数; $g(x)\in L^{2}(\Omega)$是外力项.

吊桥方程是工程中的一个重要数学模型, 1990年, 由Lazer和McKenna在文献[1]中作为非线性分析中的一个新问题提出.最近, 相似的模型被很多研究者考虑和探讨, 他们主要致力于研究解的存在性和衰退估计, 参见文献[2-3]和相关文献.作为先驱性工作, 钟承奎等人在文献[4]中首次研究了如下自治吊桥方程

$ u_{tt}+au_{t}+\triangle^{2}u+ku^{+}+f(u)=g(x) $

强解和强全局吸引子的存在性.马巧珍等人系统地探讨了单个和耦合吊桥方程解的长期行为, 包括自治和非自治的情况, 并且非线性函数满足的条件比文献[4-5]更弱, 参见文献[6-7]和相关文献.带线性阻尼的非自治吊桥方程拉回吸引子的存在性在文献[8-10]中被研究, 文献[11]中作者得到了带非线性阻尼的自治吊桥方程的全局吸引子.最近, Kang在文献[12]中通过定义合适的Lyapunov函数, 证明了带线性记忆而不带阻尼项的吊桥方程在弱拓扑空间中全局吸引子的存在性.然而, 方程中的阻尼项在实际问题中是重要且有意义的, 因此, 本文考虑具有历史记忆的的阻尼吊桥方程在强拓扑空间中解的长时间行为.关于带过去历史的相关问题的解的渐近行为的研究也可参见文献[13-16, 20]等.

为了得到本文的主要结果, 首先需要将方程$(1.1)$转化为一个确定的自治动力系统.为此, 借助文献[12, 15]的思想, 引入表示历史位移的变量, 即

$ \eta=\eta^{t}(x, s)=u(x, t)-u(x, t-s), ~(x, s)\in\Omega\times{\Bbb R}^{+}, ~t\geqslant0. $ (1.2)

$ \eta^{t}_{t}(x, s)=-\eta^{t}_{s}(x, s)+u_{t}(x, t), ~(x, s)\in\Omega\times{\Bbb R}^{+}, ~t\geqslant0. $ (1.3)

$\alpha=1+\int^{\infty}_{0}\mu(s){\rm d}s$$\mu\in L^{1}({\Bbb R}^{+})$, 方程$(1.1)$可转化为等价的自治系统

$ \left\{\begin{array}{ll} u_{tt}+\varepsilon u_{t}+\triangle^{2}u+\int^{\infty}_{0}\mu(s)\Delta^{2}\eta^{t}(s){\rm d}s+ku^{+}+f(u)=g(x), ~(x, t)\in\Omega\times{\Bbb R}^{+}, \\[8pt] \eta_{t}=-\eta_{s}+u_{t}, ~(x, t, s)\in\Omega\times{\Bbb R}^{+}\times{\Bbb R}^{+}. \end{array} \right. $ (1.4)
$ u=\Delta u=0, ~(x, t)\in\Gamma\times{\Bbb R}^{+}, ~\eta=\triangle\eta=0, ~(x, t, s)\in\Gamma\times{\Bbb R}^{+}\times{\Bbb R}^{+}; $ (1.5)
$ u(x, 0)=u_{0}(x), ~u_{t}(x, 0)=u_{1}(x), ~\eta^{t}(x, 0)=0, ~\eta^{0}(x, s)=\eta_{0}(x, s), $ (1.6)

其中

$ \left\{\begin{array}{ll} u_{0}(x)=u_{0}(x, 0), &~x\in\Omega, \\ u_{1}(x)=\partial_{t}u_{0}(x, t)\mid_{t=0}, &~x\in\Omega, \\ \eta_{0}(x, s)=u_{0}(x, 0)-u_{0}(x, -s), &~(x, s)\in\Omega\times{\Bbb R}^{+}. \end{array} \right. $
2 准备工作

不失一般性, 用$(\cdot, \cdot)$$\|\cdot\|$分别表示$L^{2}(\Omega)$中的内积和范数.特别地

$ H=V_{0}=L^{2}(\Omega), ~V=V_{1}=H^{2}(\Omega)\cap H^{1}_{0}(\Omega), $

其相应的内积和范数分别为

$ (u, v)_{V}=(\triangle u, \triangle v), ~\|u\|_{V}=\|\triangle u\|. $

定义

$ D(A)=\{u\in H^{4}(\Omega):~u|_{\partial\Omega}=\triangle u|_{\partial\Omega}=0\}, $

其中$Au=\triangle^{2}u$.内积和范数分别用$(Au, Av)$$\|Au\|^{2}=(Au, Au)$表示.

显然, $D(A)\subset V\subset H=H^{*}\subset V^{*}$, 并且嵌入是稠密的, 这里$H^{*}, V^{*}$分别为$H, V$的对偶空间.

定义

$ L_{\mu}^{2}({\Bbb R}^{+};V_{i})= \bigg\{\eta:{\Bbb R}^{+}\rightarrow V_{i}|\int_{0}^{\infty}\mu(s)\|\eta(s)\|_{V_{i}}^{2}{\rm d}s<\infty \bigg\}, $

它是一个Hilbert空间, 相应的内积和范数分别为

$ (u, v)_{\mu, V_{i}}=\int^{\infty}_{0}\mu(r)(u(r), v(r))_{V_{i}}{\rm d}r, $
$ \|u\|_{\mu, V_{i}}^{2}=(u, u)_{\mu, V_{i}}=\int^{\infty}_{0}\mu(r)\|u(r)\|_{V_{i}}^{2}{\rm d}r, ~i=0, 1, 2, $

其中$V_{2}=D(A)$.最后, 介绍如下Hilbert空间

$ {\cal H}_{0}=V\times H\times L^{2}_{\mu}({\Bbb R}^{+};V), \qquad {\cal H}_{1}=D(A)\times V\times L^{2}_{\mu}({\Bbb R}^{+};D(A)). $

利用Poincaré不等式可得

$ \lambda_{1}\|v\|^{2}\leqslant\|\triangle v\|^{2}, ~\forall v\in V, $

其中$\lambda_{1}$$\triangle^{2}$的第一个特征值.

此外, 假设非线性函数$f:{\Bbb R}\rightarrow{\Bbb R}$满足如下条件

(F$_{1})~~ \liminf\limits_{|s|\rightarrow\infty}\frac{f(s)}{s}\geqslant-\rho=-\rho(\lambda_{1}), ~\forall ~s\in{\Bbb R}$, 其中$\rho>0$;

(F$_{2})~~ |f(s)|\leqslant k_{0}(1+|s|^{p})$, 其中$k_{0}>0$$p\geqslant1$.

记忆核函数$\mu(\cdot)$满足如下条件

(H$_{1})~~\mu\in C^{1}({\Bbb R}^{+})\cap L^{1}({\Bbb R}^{+}), ~\mu'(s)\leqslant0\leqslant \mu(s), ~\forall s\in{\Bbb R}^{+}$;

(H$_{2})~~\int^{\infty}_{0}\mu(s){\rm d}s=\mu_{0}>0, ~\forall s\in{\Bbb R}^{+}$;

(H$_{3})~~ \mu'(s)+\delta\mu(s)\leqslant0, ~\forall s\in{\Bbb R}^{+}, ~\delta>0$.

由文献[12, 定理2.1]可知, 问题(1.4)-(1.6) 在弱拓扑空间中的解是适定的, 由此可定义映射$S(t):{\cal H}_{0}\rightarrow{\cal H}_{0}$, 即

$ S(t)(u_{0}, u_{1}, \eta_{0})=(u(t), u_{t}(t), \eta^{t}), ~t\geqslant0, $

其中$(u(t), u_{t}(t), \eta^{t})$是系统(1.4)-(1.6) 唯一的弱解, 算子$S(t)$满足半群的性质且可定义一个在${\cal H}_{0}$上局部Lipschitz连续的非线性$C_{0}$ -半群.

下面的抽象结果对证明我们的定理是必要的.

定义2.1[17]  设$X$为Banach空间, $\{S(t)\}_{t\geqslant0}$$X$上的算子族.我们说$\{S(t)\}_{t\geqslant0}$$X$上的强弱连续半群, 如果$\{S(t)\}_{t\geqslant0}$满足

(ⅰ) $S(0)=Id$ (恒等式);

(ⅱ) $S(t)S(s)=S(t+s), ~\forall t, s\geqslant0$;

(ⅲ) 当$t_{n}\rightarrow t, ~x_{n}\rightarrow x$时, $S(t_{n})x_{n}\rightharpoonup S(t)x$.

定理2.1[17]  设$X$$Y$是使得$X\hookrightarrow Y$的两个Banach空间, $\{S(t)\}_{t\geqslant0}$$Y$中的连续或弱连续半群.如果$\{S(t)\}_{t\geqslant0}$$X\times{\Bbb R}^{+}$上的紧子集映为$X$中的有界集, 则$\{S(t)\}_{t\geqslant0}$$X$上的强弱连续半群.

定理2.2[5]  设$X$$Y$是两个Banach空间, $X\subset Y$是连续嵌入.如果函数$\varphi\in L^{\infty}(0, T;X)$且在$Y$中弱连续, 则$\varphi$$X $中弱连续.

因此, 结合上述抽象结果和在${\cal H}_{1}$上的先验估计3.2, 我们有如下结果.

定理2.3  设条件(F$_{1}$)-(F$_{2}$), (H$_{1}$)-(H$_{3})$成立且$g\in L^{2}(\Omega)$.则对任意给定的$T>0$, 具有初值$(u_{0}, u_{1}, \eta_{0})\in{\cal H}_{1}$的初边值问题(1.4)-(1.6) 有唯一的解$(u, u_{t}, \eta)$, 满足

$ u\in L^{\infty}(0, T;D(A)), ~u_{t}\in L^{\infty}(0, T;V), ~\eta\in L^{\infty}(0, T;L^{2}_{\mu}({\Bbb R}^{+}, D(A)). $

此外, $(u, u_{t}, \eta)$是从$[0, T]$${\cal H}_{1}$的弱连续函数, 且(1.4)-(1.6) 生成的半群$\{S(t)\}_{t\geqslant0}$${\cal H}_{1}$中是强弱连续的.

定义2.2[18]  设$X$为Banach空间, $B$$X$中的有界集, 定义于$X\times X$上的函数$\phi(\cdot, \cdot)$称为$B\times B$上的收缩函数, 如果对于任意的序列$\{x_{n}\}^{\infty}_{n=1}\subset B$, 存在子列$\{x_{n_{k}}\}^{\infty}_{k=1}\subset \{x_{n}\}^{\infty}_{n=1}$, 使得

$ \lim\limits_{k\rightarrow\infty}\lim\limits_{l\rightarrow\infty}\phi(x_{n_{k}}, x_{n_{l}})=0. $ (2.1)

${\mathfrak C}$表示定义于$B\times B$上的收缩函数的集合.

定理2.4[18]  设$\{S(t)\}_{t\geqslant0}$为Banach空间$(X, \|\cdot\|)$上的半群, 并存在有界吸收集$B_{0}$.进一步, 对任意的$\epsilon>0$, 存在$T=T(B_{0}, \epsilon)$以及$\phi_{T}(\cdot, \cdot)\in{\mathfrak C}(B_{0})$, 使得

$ \|S(T)x-S(T)y\|\leqslant\epsilon+\phi_{T}(x, y), ~\forall(x, y)\in B_{0}, $

其中$\phi_{T}$依赖于$T$.则$\{S(t)\}_{t\geqslant0}$$X$中渐近紧, 即对于任意的有界序列$\{y_{n}\}^{\infty}_{n=1}\subset X$$\{t_{n}\}$, 当$t_{n}\rightarrow\infty$时, $\{S(t_{n})y_{n}\}_{n=1}^{\infty}$$X$中准紧.

定理2.5[20]  一个耗散的动力系统$({\cal H}, S(t))$有一个紧的全局吸引子当且仅当它是渐近紧的.

本文的主要结果如下.

定理2.6  设条件(F$_{1}$)-(F$_{2})$, (H$_{1}$)-(H$_{3})$成立且$g\in L^{2}(\Omega)$, 则由问题(1.4)-(1.6) 生成的动力系统有一个紧的强全局吸引子${\cal A}_{\nu}\subset{\cal H}_{1}, ~\nu\geqslant0$, 且以范数$\|\cdot\|_{{\cal H}_{1}}$吸引${\cal H}_{1}$中的任意有界集.

3 ${\cal H}_{1}$中的吸收集
3.1 ${\cal H}_{0}$上的先验估计

首先, 由(F$_{1})$可知, 存在常数$K_{1}, K_{2}>0, \eta=\eta(\rho)>0$, 使得

$ \begin{equation} f(s)s+\eta s^{2}+K_{1}\geqslant0, ~\forall s\in{\Bbb R}; \end{equation} $ (3.1)
$ \begin{equation} F(s)+\eta s^{2}+K_{2}\geqslant0, ~\forall s\in{\Bbb R}. \end{equation} $ (3.2)

$v=u_{t}+\sigma u$与方程$(1.4)_{1}$$H$中作内积后得

$ \begin{eqnarray} &&\frac{1}{2}\frac{\rm d}{{\rm d}t}(\|\triangle u\|^{2}+\|v\|^{2}+k\|u^{+}\|^{2})+\sigma\|\triangle u\|^{2}+(\nu-\sigma)(u_{t}, v)+(\eta^{t}, u_{t})_{\mu, V} \\ &&+\sigma(\eta^{t}, u)_{\mu, V}+\sigma k\|u^{+}\|^{2}+ (f(u), v)=(g, v). \end{eqnarray} $ (3.3)

结合$(1.4)$式, (H$_{2}$)-(H$_{3})$以及Hölder不等式, 由$(3.3)$式可得

$ \begin{eqnarray} &&\frac{1}{2}\frac{\rm d}{{\rm d}t}(\|\triangle u\|^{2}+\|v\|^{2} +\|\eta^{t}\|^{2}_{\mu, V}+k\|u^{+}\|^{2})+\sigma\left(1-\frac{\mu_{0}\sigma} {\delta}\right)\|\triangle u\|^{2}+(\nu-\sigma)\|v\|^{2}\\ &&-\sigma(\nu-\sigma)(u, v)+\frac{\delta}{4}\|\eta^{t}\|^{2}_{\mu, V}+\sigma k\|u^{+}\|^{2}+ (f(u), v)\leqslant(g, v). \end{eqnarray} $ (3.4)

取充分小的$\sigma$, 使得

$ \begin{equation} 1-\frac{\mu_{0}\sigma}{\delta}-\frac{\nu\sigma}{2\lambda_{1}}\geqslant1-\sigma, ~~\frac{\nu}{2}-\sigma\geqslant\frac{\nu}{4}. \end{equation} $ (3.5)

结合Hölder, Young和Poincaré不等式有

$ \begin{eqnarray} && \sigma \left(1-\frac{\mu_{0}\sigma}{\delta}\right)\|\triangle u\|^{2}+(\nu-\sigma)\|v\|^{2}-\sigma(\nu-\sigma)(u, v)\\ &\geqslant &\sigma\left(1-\frac{\mu_{0}\sigma}{\delta}\right)\|\triangle u\|^{2}+(\nu-\sigma)\|v\|^{2}-\frac{\nu\sigma}{\sqrt{\lambda_{1}}}\|\triangle u\|_{2}\|v\|_{2} \\ &=&\sigma\left(1-\frac{\mu_{0}\sigma}{\delta}-\frac{\nu\sigma}{2\lambda_{1}}\right)\|\triangle u\|^{2}+\left(\frac{\nu}{2}-\sigma\right)\|v\|^{2}\\ &\geqslant &\sigma(1-\sigma)\|\triangle u\|^{2}+\frac{\nu}{4}\|v\|^{2}. \end{eqnarray} $ (3.6)

$(3.6)$式代入$(3.4)$式, 整理可得

$ \begin{eqnarray} &&\frac{1}{2}\frac{\rm d}{{\rm d}t} \left(\|\triangle u\|^{2}+\|v\|^{2}+\|\eta^{t}\|^{2}_{\mu, V}+k\|u^{+}\|^{2}+2\int_{\Omega}F(u){\rm d}x-2\int_{\Omega}gu{\rm d}x\right)+ \sigma(1-\sigma)\|\triangle u\|^{2}\\ &&+\frac{\nu}{4}\|v\|^{2}+\frac{\delta}{4}\|\eta^{t}\|^{2}_{\mu, V}+\sigma k\|u^{+}\|^{2}+\sigma\int_{\Omega}f(u)u{\rm d}x-\sigma\int_{\Omega}gu{\rm d}x\leqslant0. \end{eqnarray} $ (3.7)

$\sigma_{0}=\min\left\{2\sigma(1-\sigma), \frac{\nu}{2}, \frac{\delta}{2}, \sigma\right\}$, 令

$ \begin{equation} E(t)=\|\triangle u\|^{2}+\|v\|^{2}+\|\eta^{t}\|^{2}_{\mu, V}+ k\|u^{+}\|^{2}+2\int_{\Omega}F(u){\rm d}x-2\int_{\Omega}gu{\rm d}x, \end{equation} $ (3.8)
$ \begin{equation} I(t)=\|\triangle u\|^{2}+\|v\|^{2}+\|\eta^{t}\|^{2}_{\mu, V}+k\|u^{+}\|^{2}+2\int_{\Omega}f(u)u{\rm d}x-2\int_{\Omega}gu{\rm d}x. \end{equation} $ (3.9)

则有

$ \frac{\rm d}{{\rm d}t}E(t)+\sigma_{0}I(t)\leqslant0, $

$ \begin{equation} E(t)\leqslant-\sigma_{0}\int_{0}^{t}I(\tau){\rm d}\tau+E(0), \end{equation} $ (3.10)

其中

$ \begin{equation} E(0)=\|\triangle u_{0}\|^{2}+\|u_{1}+\sigma u_{0}\|^{2}+\|\eta_{0}\|^{2}_{\mu, V}+k\|u^{+}_{0}\|^{2}+2\int_{\Omega}F(u_{0}){\rm d}x-2\int_{\Omega}gu_{0}{\rm d}x. \end{equation} $ (3.11)

根据(3.1)-(3.2) 式和(3.8)-(3.9) 式, 利用Sobolev紧嵌入定理可得

$ \begin{eqnarray} E(t)&\geqslant &\|\triangle u\|^{2}+\|v\|^{2}+\|\eta^{t}\|^{2}_{\mu, V}+k\|u^{+}\|^{2}-2\int_{\Omega}(\eta u^{2}+K_{2}){\rm d}x-\sigma_{0}\|u\|^{2}-\frac{1}{\sigma_{0}}\|g\|^{2} \\ &\geqslant & \left(1-\frac{k+2\eta+\sigma_{0}}{\lambda_{1}}\right)\|\triangle u\|^{2}+\|v\|^{2}+\|\eta^{t}\|^{2}_{\mu, V}-M_{1}, \end{eqnarray} $ (3.12)

其中$M_{1}=2K_{2}|\Omega|+\frac{\|g\|^{2}}{\sigma_{0}}$.类似地

$ \begin{eqnarray} I(t)&=&\|\triangle u\|^{2}+\|v\|^{2}+\|\eta^{t}\|^{2}_{\mu, V}+k\|u^{+}\|^{2}+2\int_{\Omega}f(u)u{\rm d}x-2\int_{\Omega}gu{\rm d}x \\ &\geqslant & \left(1-\frac{k+2\eta+\sigma_{0}}{\lambda_{1}}\right)\|\triangle u\|^{2}+\|v\|^{2}+\|\eta^{t}\|^{2}_{\mu, V}-M_{2}, \end{eqnarray} $ (3.13)

其中$M_{2}=2K_{1}|\Omega|+\frac{\|g\|^{2}}{\sigma_{0}}$.令$\frac{k+2\eta}{\lambda_{1}}<1$, 且$0<\sigma_{0}<\lambda_{1}-k-2\eta$, 则有

$ \begin{equation} 1-\frac{k+2\eta+\sigma_{0}}{\lambda_{1}}>0. \end{equation} $ (3.14)

结合(3.12)-(3.14) 式, 存在一个常数$C_{1}$, 使得

$ \begin{equation} E(t)\geqslant C_{1}(\|\triangle u\|^{2}+\|v\|^{2}+\|\eta^{t}\|^{2}_{\mu, V})-M_{1}, \end{equation} $ (3.15)
$ \begin{equation} I(t)\geqslant C_{1}(\|\triangle u\|^{2}+\|v\|^{2}+\|\eta^{t}\|^{2}_{\mu, V})-M_{2}. \end{equation} $ (3.16)

故由(3.15)-(3.16) 式和$(3.10)$式可知

$ \begin{eqnarray} &&C_{1} (\|\triangle u\|^{2}+\|v\|^{2}+\|\eta^{t}\|^{2}_{\mu, V})-M_{1} \\ &\leqslant & -\sigma_{0}\int^{t}_{0}[C_{1}(\|\triangle u\|^{2}+\|v\|^{2}+\|\eta^{t}\|^{2}_{\mu, V})-M_{2}]{\rm d}t+E(0). \end{eqnarray} $ (3.17)

因此, 对$\forall K>\frac{M_{2}}{C_{1}}$, 存在$t_{0}=t_{0}(B)$使得

$ \begin{equation} \|\triangle u(t_{0})\|^{2}+\|v(t_{0})\|^{2}+\|\eta^{t_{0}}\|^{2}_{\mu, V}\leqslant K. \end{equation} $ (3.18)

所以, 如果$u$是系统(1.4)-(1.6) 的解, 令

$ B_{0}=\bigcup\limits_{t\geqslant0}S(t)B_{1}, $

其中

$ B_{1}=\{(u_{0}, u_{1}, \eta_{0})^{T}\in{\cal H}_{0}:\|\triangle u_{0}\|^{2}+\|v_{0}\|^{2}+\|\eta_{0}\|^{2}_{\mu, V}\leqslant K\}, $

$B_{0}$是半群$\{S(t)\}_{t\geqslant0}$的一个有界吸收集.

另一方面, 由上述讨论可知, 存在一个常数$\mu_{1}$使得

$ \begin{equation} \|\triangle u(t)\|^{2}+\|v(t)\|^{2}+\|\eta^{t}\|^{2}_{\mu, V}\leqslant \mu_{1}^{2}, ~\forall t\geqslant t_{0}. \end{equation} $ (3.19)
3.2 ${\cal H}_{1}$上的先验估计

$Av=Au_{t}+\sigma Au$与方程$(1.4)_{1}$$H$中作内积后得

$ \begin{eqnarray} &&\frac{1}{2}\frac{\rm d}{{\rm d}t} (\|Au\|^{2}+\|\triangle v\|^{2})+\sigma\|A u\|^{2}+(\nu-\sigma)(u_{t}, Av)+(\eta^{t}, u_{t})_{\mu, D(A)} \\ &&+\sigma(\eta^{t}, u)_{\mu, D(A)}+(ku^{+}, Av)+ (f(u), Av)=(g, Av).\end{eqnarray} $ (3.20)

类似前面的讨论, 我们有

$ \begin{eqnarray} &&\frac{1}{2}\frac{\rm d}{{\rm d}t} (\|Au\|^{2}+\|\triangle v\|^{2}+\|\eta^{t}\|^{2}_{\mu, D(A)}) +\sigma\left(1-\frac{\mu_{0}\sigma}{\delta}\right)\|Au\|^{2}+(\nu-\sigma)\|\triangle v\|^{2} \\ &&-\sigma(\nu-\sigma)(Au, v)+\frac{\delta}{4}\|\eta^{t}\|^{2}_{\mu, D(A)}+(ku^{+}, Av)+ (f(u), Av)\leqslant(g, Av).\end{eqnarray} $ (3.21)

进一步, 类似$(3.6)$式的估计有

$ \begin{equation} \sigma\left(1-\frac{\mu_{0}\sigma}{\delta}\right)\|Au\|^{2}+(\nu-\sigma)\|\triangle v\|^{2}-\sigma(\nu-\sigma)(Au, v)\geqslant \sigma(1-\sigma)\|Au\|^{2}+\frac{\nu}{4}\|\triangle v\|^{2}. \end{equation} $ (3.22)

此外, 由条件$(F_{2})$可知, $f(u), f'(u)$$L^{\infty}$中一致有界, 即存在一个常数$M>0$, 使得

$ \begin{equation}\|f(u)\|_{L^{\infty}}\leqslant M, ~~\|f'(u)\|_{L^{\infty}}\leqslant M. \end{equation} $ (3.23)

因此, 结合Hölder, Young, Cauchy不等式及$(3.19)$式, 可得

$ \begin{eqnarray} (ku^{+}, Av)&\geqslant & \frac{\rm d}{{\rm d}t}(ku^{+}, Au)+\sigma(ku^{+}, Au)-k\|(u^{+})_{t}\|\|Au\| \\ &\geqslant & \frac{\rm d}{{\rm d}t}(ku^{+}, Au)+\sigma(ku^{+}, Au)-k\|u_{t}\|\|Au\|\\ &\geqslant & \frac{\rm d}{{\rm d}t}(ku^{+}, Au)+\sigma(ku^{+}, Au)-\frac{\sigma}{2}\|Au\|^{2}-\frac{k^{2}\mu_{1}^{2}}{2\sigma}, ~~t\geqslant t_{0}, \end{eqnarray} $ (3.24)
$ \begin{eqnarray} (f(u), Av)&\geqslant & \frac{\rm d}{{\rm d}t}(f(u), Au)+\sigma(f(u), Au)-\int_{\Omega}|f'(u)||u_{t}||Au|{\rm d}x \\ &\geqslant & \frac{\rm d}{{\rm d}t}(f(u), Au)+\sigma(f(u), Au)-\frac{\sigma}{4}\|Au\|^{2}-\frac{M^{2}\mu_{1}^{2}}{\sigma}, ~~t\geqslant t_{0}, \end{eqnarray} $ (3.25)

$(3.22)$式和(3.24)-(3.25) 式代入$(3.21)$式, 整理可得

$ \begin{eqnarray} &&\frac{\rm d}{{\rm d}t}(\|Au\|^{2}+\|\triangle v\|^{2}+\|\eta^{t}\|^{2}_{\mu, D(A)}+2(ku^{+}, Au)+2(f(u), Au) -2(g, Au))\\ &&+ \sigma\left(\frac{1}{2}-2\sigma\right)\|Au\|^{2} +\frac{\nu}{2}\|\triangle v\|^{2}+\frac{\delta}{2}\|\eta^{t}\|^{2}_{\mu, D(A)} \\ &&+2\sigma(ku^{+}, Au)+ 2\sigma(f(u), Au)-2\sigma(g, Au)\leqslant C, \end{eqnarray} $ (3.26)

其中$C=\frac{\mu_{1}^{2}(2M^{2}+k^{2})}{\sigma}$, 令

$ \sigma_{0}=\min\left(\frac{\sigma}{2}-2\sigma^{2}, \frac{\nu}{2}, \frac{\delta}{2}\right), $

结合$(3.26)$式可得

$ \begin{eqnarray} &&\frac{\rm d}{{\rm d}t}(\|Au\|^{2}+\|\triangle v\|^{2}+\|\eta^{t}\|^{2}_{\mu, D(A)}+2(ku^{+}, Au)+2(f(u), Au) -2(g, Au)) \\ &&+\sigma_{0}(\|Au\|^{2} +\|\triangle v\|^{2}+\|\eta^{t}\|^{2}_{\mu, D(A)}+2(ku^{+}, Au)+ 2(f(u), Au)-2(g, Au))\leqslant C. \end{eqnarray} $ (3.27)

此外, 利用Hölder不等式, Sobolev嵌入定理, 并结合$(3.19)$式和$(3.23)$式, 我们有

$ \begin{eqnarray} \frac{\rm d}{{\rm d}t}\left(\frac{1}{2}\|Au\|^{2}+2(ku^{+}, Au)\right) &= & \frac{\rm d}{{\rm d}t}\|\sqrt{\frac{1}{2}}Au+\sqrt{2}ku^{+}\|^{2}-4k^{2}\int_{\Omega}|u^{+}||u^{+}_{t}|{\rm d}x \\ &\geqslant & \frac{\rm d}{{\rm d}t}\|\sqrt{\frac{1}{2}}Au+\sqrt{2}ku^{+}\|^{2}-4k^{2}\mu_{1}^{2} \end{eqnarray} $ (2.28)

$ \begin{eqnarray} &&\frac{\rm d}{{\rm d}t} \left(\frac{1}{2}\|Au\|^{2}+2(f(u), Au)-2(g, Au)\right) \\ &\geqslant & \frac{\rm d}{{\rm d}t}\|\sqrt{\frac{1}{2}}Au+\sqrt{2}f(u)-\sqrt{2}g\|^{2}-4\int_{\Omega}|f(u)||f'(u)||u_{t}|{\rm d}x+4\int_{\Omega}f'(u)u_{t}g{\rm d}x \\ &\geqslant & \frac{\rm d}{{\rm d}t}\|\sqrt{\frac{1}{2}}Au+\sqrt{2}f(u)-\sqrt{2}g\|^{2}-4M^{2}\mu_{1}-4M\mu_{1}\|g\|, ~~t\geqslant t_{0}. \end{eqnarray} $ (3.29)

将(3.28)-(3.29) 式代入$(3.27)$式可得

$ \begin{eqnarray} &&\frac{\rm d}{{\rm d}t}\left(\|\sqrt{\frac{1}{2}}Au+\sqrt{2}ku^{+}\|^{2}+\|\sqrt{\frac{1}{2}}Au+\sqrt{2}f(u)-\sqrt{2}g\|^{2}+\|\triangle v\|^{2}+\|\eta^{t}\|^{2}_{\mu, D(A)} \right)\\ &&+\sigma_{0}\left(\|\sqrt{\frac{1}{2}}Au+\sqrt{2}ku^{+}\|^{2}+\|\sqrt{\frac{1}{2}}Au+\sqrt{2}f(u)-\sqrt{2}g\|^{2}+\|\triangle v\|^{2}+\|\eta^{t}\|^{2}_{\mu, D(A)}\right) \leqslant\tilde{C}, \\ \end{eqnarray} $ (3.30)

其中$t\geqslant t_{0}$, $\tilde{C}=C+4k^{2}\mu_{1}^{2}(1+\sigma_{0})+ 2M^{2}(\sigma_{0}+2\mu_{1})+4M\|g\|_{2}(\sigma_{0}+ \mu_{1})+2\sigma_{0}\|g\|^{2}$.

定义泛函

$ W(t)=\|\sqrt{\frac{1}{2}}Au+\sqrt{2}ku^{+}\|^{2}+\|\sqrt{\frac{1}{2}}Au+\sqrt{2}f(u)-\sqrt{2}g\|^{2}+\|\triangle v\|^{2}+\|\eta^{t}\|^{2}_{\mu, D(A)}, $

则有

$ \frac{\rm d}{{\rm d}t}W(t)+\sigma_{0}W(t)\leqslant\tilde{C}, ~t\geqslant t_{0}(B). $

根据Gronwall引理可得

$ W(t)\leqslant W(t_{0}){\rm e}^{-\sigma_{0}(t-t_{0})}+\frac{\tilde{C}}{\sigma_{0}}, ~t\geqslant t_{0}. $ (3.31)

因此, 若${\cal H}_{1}$上有一个以$P_{0}$为中心, 以$\rho$为半径的球$B_{{\cal H}_{1}}(P_{0}, \rho)$, 则根据$(3.31)$式, 存在一个常数$R_{1}>0$, 使得

$ \sup\limits_{(u(t_{0}), u_{t}(t_{0}), \eta^{t_{0}})\in B}W(t_{0}) \leqslant R_{1}^{2}. $

$t_{1}-t_{0}\geqslant\frac{1}{\sigma_{0}}\log R_{1}^{2}$, 则$W(t)\leqslant\mu_{2}^{2}, ~t\geqslant t_{1}$, 其中$\mu_{2}^{2}=1+\frac{\tilde{C}}{\sigma_{0}}$.

这样, 我们得到${\cal H}_{1}$上有界吸收集的存在性, 即如下结果.

引理3.1  设条件(F$_{1}$)-(F$_{2}$), (H$_{1}$)-(H$_{3})$成立且$g\in L^{2}(\Omega)$, 则空间${\cal H}_{1}$中以0为中心, $\mu_{2}$为半径的球$B_{1}=B_{{\cal H}_{1}}(0, \mu_{2})$是问题(1.4)-(1.6) 生成的解半群$\{S(t)\}_{t\geqslant0}$${\cal H}_{1}$中的有界吸收集, 即对${\cal H}_{1}$中的任意有界集$B$, 存在$t=t_{0}(B)$, 使得当$t\geqslant t_{0}(B)$时, 有$S(t)B\subset B_{1}$.

有界吸收集的存在性意味着对于依赖于有界集$B\subset{\cal H}_{1}$的初值, 问题(1.4)-(1.6) 的解全局有界, 即若$(u, u_{t}, \eta)$是问题(1.4)-(1.6) 在有界集$B$上对应于初值$(u_{0}, u_{1}, \eta_{0})$的解, 则

$ \|(u(t), u_{t}(t), \eta^{t})\|_{{\cal H}_{1}}\leqslant C_{B}, ~\forall t\geqslant0, $ (3.32)

其中$C_{B}>0$是依赖于$B$的常数.

4 ${\cal H}_{1}$上全局吸引子的存在性

本节中, 借助文献[11, 19, 21]的思想, 首先给出一个能量不等式的先验估计.然后用定理4.1得到${\cal H}_{1}$上的渐近紧性.方便起见, 令$B_{1}$为引理3.1中得到的有界吸收集, 且定义如下泛函

$ E_{w}(t)=\frac{1}{2}\|Aw\|^{2}+\frac{1}{2}\|\triangle w_{t}\|^{2}+\frac{1}{2}\|\xi^{t}\|^{2}_{\mu, D(A)}. $
4.1 先验估计

为了得到渐近紧性, 首先给出一个先验估计.

$(u_{i}(t), u_{i_{t}}(t), ~\eta_{i}^{t}(s))\ (i=1, 2)$是对应于$(u_{0}^{i}, v_{0}^{i}, \eta_{0}^{i})\in B_{1}$的解, $w(t)=u_{1}(t)-u_{2}(t)$, $\xi^{t}(s)=\eta_{1}^{t}(s)-\eta_{2}^{t}(s)$.则$w(t)$$\xi^{t}(s)$满足

$ \begin{eqnarray} \left\{\begin{array}{ll} w_{tt}+\nu w_{t}+\triangle^{2}w+\int^{\infty}_{0}\mu(s)\Delta^{2}\xi^{t}(s){\rm d}s+ku_{1}^{+}-ku_{2}^{+}+f(u_{1})-f(u_{2})=0, \\ \xi_{t}=-\xi_{s}+w_{t}. \end{array} \right. \end{eqnarray} $ (4.1)

相应的初值条件为$(w(0), w_{t}(0), \xi^{0}(s))=(u_{0}^{1}, v_{0}^{1}, \eta_{0}^{1})-(u_{0}^{2}, v_{0}^{2}, \eta_{0}^{2})$.

首先, 用$Aw$与方程$(4.1)_{1}$作内积且在$[0, T]\times\Omega$上积分得

$ \begin{eqnarray} \int^{T}_{0}\|Aw\|^{2}{\rm d}s &=&\int_{\Omega}\triangle w_{t}(0)\triangle w(0){\rm d}x -\int_{\Omega}\triangle w_{t}(T)\triangle w(T){\rm d}x \\ && +\int^{T}_{0}\|\triangle w_{t}\|^{2}{\rm d}s-\frac{\nu}{2}\|\triangle w(T)\|^{2} +\frac{\nu}{2}\|\triangle w(0)\|^{2} \\ &&-\int^{T}_{0}(\xi, w)_{\mu, D(A)}{\rm d}s -k\int^{T}_{0}\int_{\Omega}(u^{+}_{1}-u^{+}_{2})Aw{\rm d}x{\rm d}s \\ &&-\int^{T}_{0}\int_{\Omega}(f(u_{1}(s))-f(u_{2}(s)))Aw(s){\rm d}x{\rm d}s. \end{eqnarray} $ (4.2)

利用Hölder和Young不等式有

$ \begin{eqnarray} -k\int^{T}_{0}\int_{\Omega}(u_{1}^{+}-u_{2}^{+})Aw{\rm d}x{\rm d}s &\leqslant&k\int^{T}_{0}\int_{\Omega}|u^{+}_{1}-u^{+}_{2} |\cdot|Aw|{\rm d}x{\rm d}s \\ &\leqslant& kL\int^{T}_{0}\int_{\Omega}|w|\cdot|Aw|{\rm d}x{\rm d}s \\ &\leqslant& \frac{1}{2}\int^{T}_{0}\|Aw\|^{2}{\rm d}s+\frac{k^{2}L^{2}}{2} \int^{T}_{0}\|w\|^{2}{\rm d}s, \end{eqnarray} $ (4.3)

上面的估计我们用到: $|u_{1}^{+}-u_{2}^{+}|\leqslant L|u_{1}-u_{2}|=L|w|$, 其中$L>0$是一个合适的正常数.则由(4.2)-(4.3) 式可得

$ \begin{eqnarray} \int^{T}_{0}\|Aw\|^{2}{\rm d}s &=&\int_{\Omega}\triangle w_{t}(0)\triangle w(0){\rm d}x -\int_{\Omega}\triangle w_{t}(T)\triangle w(T){\rm d}x+ \int^{T}_{0}\|\triangle w_{t}\|^{2}{\rm d}s\\ &&-\frac{\nu}{2}\|\triangle w(T)\|^{2}+\frac{\nu}{2} \|\triangle w(0)\|^{2}-\int^{T}_{0}(\xi, w)_{\mu, D(A)}{\rm d}s +\frac{1}{2}\int^{T}_{0}\|Aw\|^{2}{\rm d}s \\ &&+\frac{k^{2}L^{2}}{2}\int^{T}_{0}\|w\|^{2}{\rm d}s- \int^{T}_{0}\int_{\Omega}(f(u_{1}(s))-f(u_{2}(s)))Aw(s){\rm d}x{\rm d}s. \end{eqnarray} $ (4.4)

利用Young不等式和条件(H$_{2})$, 有

$ \begin{eqnarray} (\xi^{t}, w)_{\mu, D(A)}\geqslant-\frac{\mu_{0}}{2}\|\xi^{t}\|^{2}_{\mu, D(A)}-\frac{1}{2}\|Aw\|^{2}. \end{eqnarray} $ (4.5)

其次, 用$Aw_{t}$与方程$(4.1)_{1}$作内积且在$[s, T]\times\Omega$上积分得

$ \begin{eqnarray} &&E_{w} (T)+\nu\int^{T}_{s}\|\triangle w_{t}\|^{2}{\rm d}\tau +\frac{\delta}{2}\int^{T}_{s}\|\xi^{\tau}\|_{\mu, D(A)}^{2}{\rm d}\tau \\ &\leqslant& E_{w}(s)-k\int^{T}_{s}\int_{\Omega}(u_{1}^{+}-u_{2}^{+}) Aw_{t}{\rm d}x{\rm d}\tau- \int^{T}_{s}\int_{\Omega}(f(u_{1}(\tau))-f(u_{2}(\tau)))Aw_{t}(\tau) {\rm d}x{\rm d}\tau. \end{eqnarray} $ (4.6)

其中$0\leqslant s\leqslant T$.利用Young不等式和(3.19) 式可知

$ \begin{eqnarray} -k\int^{T}_{s}\int_{\Omega}(u_{1}^{+}-u _{2}^{+})Aw_{t}{\rm d}x{\rm d}\tau &= & -k\int^{T}_{s}\left[\frac{\rm d}{{\rm d}t}(u^{+}_{1}-u^{+}_{2}, Aw)-((u^{+}_{1}-u^{+}_{2})_{t}, Aw)\right]{\rm d}\tau \\ &\leqslant& -k(u^{+}_{1}(T)-u^{+}_{2}(T), Aw(T))+k(u^{+}_{1}(s)-u^{+}_{2}(s), Aw(s)) \\ &&+k\int^{T}_{s}[\|(u^{+}_{1})_{t}\|+\|(u^{+}_{2})_{t}\|]\cdot\|Aw\|{\rm d}\tau \\ &\leqslant& -k(u^{+}_{1}(T)-u^{+}_{2}(T), Aw(T))+k(u^{+}_{1}(s)-u^{+}_{2}(s), Aw(s))\\ &&+2k\mu_{1}\int^{T}_{s}\|Aw\|{\rm d}\tau. \end{eqnarray} $ (4.7)

从而由(4.6)-(4.7) 式可得

$ \begin{eqnarray} &&E_{w} (T)+\nu\int^{T}_{s}\|\triangle w_{t}\|^{2}{\rm d}\tau +\frac{\delta}{2}\int^{T}_{s}\|\xi^{\tau}\|_{\mu, D(A)}^{2}{\rm d}\tau \\ &\leqslant& E_{w}(s)-k(u^{+}_{1}(T)-u^{+}_{2}(T), Aw(T))+k(u^{+}_{1}(s )-u^{+}_{2}(s), Aw(s)) \\ &&+2k\mu_{1}\int^{T}_{s}\|Aw\|{\rm d}\tau- \int^{T}_{s}\int_{\Omega}(f(u_{1}(\tau))-f(u_{2}(\tau)))Aw_{t} (\tau){\rm d}x{\rm d}\tau. \end{eqnarray} $ (4.8)

$[0, T]$上对(4.8) 式积分, 结合Young不等式和(3.19) 式, 我们有

$ \begin{eqnarray} TE_{w}(T)&\leqslant&\int^{T}_{0}E_{w}(s){\rm d}s-k\int^{T}_{0}(u^{+}_{1}(T) -u^{+}_{2}(T), Aw(T)){\rm d}s \\ &&+k\int^{T}_{0}(u^{+}_{1}(s)-u^{+}_{2}(s), Aw(s)){\rm d}s +2k\mu_{1}\int^{T}_{0}\int^{T}_{s}\|Aw\|{\rm d}\tau {\rm d}x \\ &&-\int^{T}_{0}\int^{T}_{s}\int_{\Omega}(f(u_{1}(\tau))-f(u_{2}(\tau)))Aw_{t}(\tau){\rm d}x{\rm d}\tau{\rm d}s \\ &\leqslant& \int^{T}_{0}E_{w}(s){\rm d}s-k\int^{T}_{0}(u^{+}_{1}(T)-u^{+}_{2}(T), Aw(T)) {\rm d}s \\ &&+2k\mu_{1}\int^{T}_{0}\|Aw(s)\|{\rm d}s +2k\mu_{1}\int^{T}_{0}\int^{T}_{s}\|Aw\|{\rm d}\tau {\rm d}x \\ &&-\int^{T}_{0}\int^{T}_{s}\int_{\Omega}(f(u_{1}(\tau))-f(u_{2}(\tau))) Aw_{t}(\tau){\rm d}x{\rm d}\tau{\rm d}s. \end{eqnarray} $ (4.9)

此外, 将(4.8) 式中的$s$用0代替可得

$ \begin{eqnarray} &&\int^{T}_{0} \|\triangle w_{t}(\tau)\|^{2}{\rm d}\tau+\int^{T}_{0} \|\xi^{\tau}\|_{\mu, D(A)}^{2}{\rm d}\tau\\ &\leqslant&C_{2}E_{w}(0)-C_{2}k(u^{+}_{1}(T)-u^{+}_{2}(T), Aw(T)) +C_{2}k(u^{+}_{1}(0)-u^{+}_{2}(0), Aw(0)) \\ &&+2C_{2}k\mu_{1}\int^{T}_{0}\|Aw\|{\rm d}\tau- C_{2}\int^{T}_{0}\int_{\Omega}(f(u_{1}(\tau))-f(u_{2}(\tau)))Aw_{t} (\tau){\rm d}x{\rm d}\tau, \end{eqnarray} $ (4.10)

其中$C_{2}$是依赖于$\nu, ~\delta$的常数.因此, 由(4.4)-(4.5) 式和(4.10) 式有

$ \begin{eqnarray} \int^{T}_{0}E_{w}(s){\rm d}s&\leqslant &C_{2}C_{3}E_{w}(0) -C_{2}C_{3}k(u^{+}_{1}(T)-u^{+}_{2}(T), Aw(T)) \\ &&+C_{2}C_{3}k(u^{+}_{1}(0)-u^{+}_{2}(0), Aw(0)) +2C_{2}C_{3}k\mu_{1}\int^{T}_{0}\|Aw\|{\rm d}\tau \\ &&- C_{2}C_{3}\int^{T}_{0}\int_{\Omega}(f(u_{1}(\tau))-f(u_{2}(\tau))) Aw_{t}(\tau){\rm d}x{\rm d}\tau\\ &&+\frac{1}{2}\int_{\Omega}\triangle w_{t}(0)\triangle w(0){\rm d}x -\frac{1}{2}\int_{\Omega}\triangle w_{t}(T)\triangle w(T){\rm d}x \\ && -\frac{\nu}{4}\|\triangle w(T)\|^{2}+\frac{\nu}{4}\|\triangle w(0)\|^{2} +\frac{1}{2}\int^{T}_{0}\|Aw\|^{2}{\rm d}s +\frac{k^{2}L^{2}}{4} \int^{T}_{0}\|w\|^{2}{\rm d}s \\ &&-\frac{1}{2}\int^{T}_{0}\int_{\Omega}(f(u_{1}(s)) -f(u_{2}(s)))Aw(s){\rm d}x{\rm d}s, \end{eqnarray} $ (4.11)

其中$C_{3}=\max\{1, (\mu_{0}+2)/4\}$, 结合(4.9) 式和(4.11) 式可得

$ \begin{eqnarray} TE_{w}(T)&\leqslant &C_{2}C_{3}E_{w}(0)-C_{2}C_{3}k(u^{+}_{1}(T) -u^{+}_{2}(T), Aw(T)) \\ &&+C_{2}C_{3}k(u^{+}_{1}(0)-u^{+}_{2}(0), Aw(0)) +\frac{1}{2}\int_{\Omega}\triangle w_{t}(0)\triangle w(0){\rm d}x \\ &&-\frac{1}{2}\int_{\Omega}\triangle w_{t}(T)\triangle w(T){\rm d}x-\frac{\nu}{4}\|\triangle w(T)\|^{2}+\frac{\nu}{4}\|\triangle w(0)\|^{2}\\ &&-k\int^{T}_{0}(u^{+}_{1}(T)-u^{+}_{2}(T), Aw(T)){\rm d}s+ 2C_{2}C_{3}k\mu_{1} \int^{T}_{0}\|Aw\|{\rm d}\tau \\ && +\frac{1}{2}\int^{T}_{0}\|Aw\|^{2}{\rm d}s -C_{2}C_{3}\int^{T}_{0}\int_{\Omega}(f(u_{1}(\tau))-f(u_{2}(\tau))) Aw_{t}(\tau){\rm d}x{\rm d}\tau \\ && +\frac{k^{2}L^{2}}{4}\int^{T}_{0}\|w\|^{2}{\rm d}s -\frac{1}{2}\int^{T}_{0}\int_{\Omega}(f(u_{1}(s)) -f(u_{2}(s)))Aw(s){\rm d}x{\rm d}s \\ &&+2k\mu_{1}\int^{T}_{0}\|Aw(s)\|{\rm d}s +2k\mu_{1}\int^{T}_{0}\int^{T}_{s}\|Aw\|{\rm d}\tau {\rm d}x \\ &&-\int^{T}_{0}\int^{T}_{s} \int_{\Omega}(f(u_{1}(\tau))-f(u_{2}(\tau)))Aw_{t} (\tau){\rm d}x{\rm d}\tau{\rm d}s. \end{eqnarray} $ (4.12)

$ \begin{eqnarray} C_{B}&=&C_{2}C_{3}E_{w}(0)-C_{2}C_{3}k(u^{+}_{1}(T)-u^{+}_{2}(T), Aw(T))+ C_{2}C_{3}k(u^{+}_{1}(0)-u^{+}_{2}(0), Aw(0)) \\ &&+\frac{1}{2}\int_{\Omega}\triangle w_{t}(0)\triangle w(0){\rm d}x -\frac{1}{2}\int_{\Omega}\triangle w_{t}(T)\triangle w(T){\rm d}x -\frac{\nu}{4}\|\triangle w(T)\|^{2}+\frac{\nu}{4}\|\triangle w(0)\|^{2} \\ &&-k\int^{T}_{0}(u^{+}_{1}(T)-u^{+}_{2}(T), Aw(T)){\rm d}s, \end{eqnarray} $ (4.13)
$ \begin{eqnarray} &&\phi_{T}((u_{0}^{1}, v_{0}^{1}, \eta_{0}^{1}), (u_{0}^{2}, v_{0}^{2}, \eta_{0}^{2})) \\ &=&2C_{2}C_{3}k\mu_{1}\int^{T}_{0}\|Aw\|{\rm d}\tau +\frac{1}{2}\int^{T}_{0}\|Aw\|^{2}{\rm d}s+\frac{k^{2}L^{2}}{4}\int^{T}_{0}\|w\|^{2}{\rm d}s \\ &&- C_{2}C_{3}\int^{T}_{0}\int_{\Omega}(f(u_{1}(\tau))-f(u_{2}(\tau))) Aw_{t}(\tau){\rm d}x{\rm d}\tau+2k\mu_{1}\int^{T}_{0}\|Aw(s)\|{\rm d}s \\ &&-\frac{1}{2}\int^{T}_{0}\int_{\Omega}(f(u_{1}(s))-f(u_{2}(s)))Aw(s){\rm d}x{\rm d}s +2k\mu_{1}\int^{T}_{0}\int^{T}_{s}\|Aw\|{\rm d}\tau {\rm d}x \\ &&-\int^{T}_{0}\int^{T}_{s}\int_{\Omega}(f(u_{1}(\tau))-f(u_{2}(\tau))) Aw_{t}(\tau){\rm d}x{\rm d}\tau{\rm d}s. \end{eqnarray} $ (4.14)

则有

$ \begin{eqnarray}E_{w}(T)\leqslant\frac{C_{B}}{T}+\frac{1}{T}\phi_{T} ((u_{0}^{1}, v_{0}^{1}, \eta_{0}^{1}), (u_{0}^{2}, v_{0}^{2}, \eta_{0}^{2})). \end{eqnarray} $ (4.15)
4.2 渐近紧性

引理4.1  设$f(\cdot)$满足条件(F$_{1})$-(F$_{2})$, 且$\{u_{n}(t), u_{n_{t}}(t), \eta_{n}(t)\}$$L^{\infty}(s, T;{\cal H}_{1})$中弱*收敛.则

$ \begin{eqnarray} \lim\limits_{n\rightarrow\infty}\lim\limits_{m\rightarrow\infty}\int^{T}_{s}(f(u_{n}(t, x))-f(u_{m}(t, x)) , Au_{n_{t}}(t, x)-Au_{m_{t}}(t, x)){\rm d}t=0. \end{eqnarray} $ (4.16)

  设

$ \begin{eqnarray} \left\{ \begin{array}{ll} w_{n}\rightarrow w, &~~\texttt{在}~L^{\infty}(s, T;D(A))~\texttt{中弱收敛}, \\ w_{n_{t}}\rightarrow w_{t}, &~~\texttt{在}~L^{\infty}(s, T;V)~\texttt{中弱收敛}. \end{array} \right.\end{eqnarray} $ (4.17)

则由嵌入定理可得

$ \begin{eqnarray} \mbox{在空间 $C([s, T];V)$ 中, $u_{n}\rightarrow u$ 强收敛.} \end{eqnarray} $ (4.18)

另一方面, 根据定理2.2和(4.17) 式可知, 序列$\{u_{n}\}$$C(s, T;D(A))$中有界且对每一个$t\in[s, T]$, 序列$\{u_{n}(t)\}$$D(A)$中有界.故由(4.17) 式可得在空间$D(A)$中, $u_{n}(t)\rightarrow u(t)$弱收敛.从而有

$ \begin{eqnarray} &&\mbox{在空间 $L^{1}(0, T)$ 中, $(f(u_{n}), Au_{n})\rightarrow(f(u), Au)$ 强收敛.}\end{eqnarray} $ (4.19)
$ \begin{eqnarray} &&\mbox{在空间 $L^{1}(0, T)$ 中, $(f'(u_{n})u_{n_{t}}, Au_{n})\rightarrow (f'(u)u_{t}, Au)$ 强收敛. } \end{eqnarray} $ (4.20)

此外, 由于

$ \begin{eqnarray} (f(u_{n}(t)), Au_{n_{t}}(t))=\frac{\partial}{\partial t} (f(u_{n}(t)), Au_{n}(t))-(f'(u_{n}(t))u_{n_{t}}(t), Au_{n}(t)), \end{eqnarray} $ (4.21)

则有

$ \begin{eqnarray} &&\lim\limits_{n\rightarrow\infty} \lim\limits_{m\rightarrow\infty}\int^{T}_{s}(f(u_{n}(t))-f(u_{m}(t)) , Au_{n_{t}}(t, x)-Au_{m_{t}}(t, x)){\rm d}t\\ &=& \lim\limits_{n\rightarrow\infty}\int^{T}_{s}\frac{\partial}{\partial_{t}}f(u_{n}(t), Au_{n}(t)){\rm d}t -\lim\limits_{n\rightarrow\infty}\int^{T}_{s}(f'(u_{n}(t))u_{n_{t}}(t), Au_{n}(t)){\rm d}t\\ &&+ \lim\limits_{m\rightarrow\infty}\int^{T}_{s}\frac{\partial}{\partial_{t}}f(u_{m} (t), Au_{m}(t)){\rm d}t -\lim\limits_{m\rightarrow\infty}\int^{T}_{s}(f'(u_{m}(t))u_{m_{t}}(t), Au_{m}(t)){\rm d}t \\ &&- \lim\limits_{n\rightarrow\infty}\lim\limits_{m\rightarrow\infty}\int^{T}_{s}\frac{\partial}{\partial_{t}}f(u_{n}(t), Au_{m}(t)){\rm d}t+ \lim\limits_{n\rightarrow\infty}\lim\limits_{m\rightarrow\infty}\int^{T}_{s}(f'(u_{n}(t))u_{n_{t}}(t), Au_{m}(t)){\rm d}t\\ &&- \lim\limits_{n\rightarrow\infty}\lim\limits_{m\rightarrow\infty}\int^{T}_{s}\frac{\partial}{\partial_{t}}f(u_{m}(t), Au_{n}(t)){\rm d}t+ \lim\limits_{n\rightarrow\infty}\lim\limits_{m\rightarrow\infty}\int^{T}_{s}(f'(u_{m}(t))u_{m_{t}}(t), Au_{n}(t)){\rm d}t\\ &=& 2f(u(T), Au(T))-f(u(s), Au(s))-2\int^{T}_{s}(f'(u)u_{t}, Au){\rm d}t\\ &&-2f(u(T), Au(T))-f(u(s), Au(s))+2\int^{T}_{s}(f'(u)u_{t}, Au){\rm d}t=0. \end{eqnarray} $ (4.22)

证毕.

定理4.1  在定理$2.6$的假设条件下, 由问题(1.4)-(1.6) 生成的半群$\{S(t)\}_{t\geqslant0}$${\cal H}_{1}$中渐近紧.

  由于半群$\{S(t)\}_{t\geqslant0}$存在有界吸收集, 对任意的$\{S(t)\}_{t\geqslant0}$, 取$T$足够大, 使得$\frac{C_{B}}{T}\leqslant\varepsilon.$因此, 根据定理2.4, 只需证明对固定的$T$, (4.14) 式中定义的函数属于${\mathfrak C}(B_{1})$, 即$\phi_{T}(\cdot, \cdot)$满足(2.1) 式.令$(u_{n}, u_{t_{n}}, \eta_{n})$是关于初始值$(u_{0}^{n}, v_{0}^{n}, \eta_{0}^{n})\in B_{1}, n=1, 2, \cdots$的解.则由于$B_{1}$${\cal H}_{1}$中的有界正不变集, 不失一般性, 假设

$ \begin{eqnarray} &&\mbox{在空间 $L^{\infty}(0, T;D(A))$ 中, $u_{n}\rightarrow u$ 弱$^{*}$收敛.} \end{eqnarray} $ (4.23)
$ \begin{eqnarray} &&\mbox{在空间 $L^{\infty}(0, T;V)$ 中, $u_{n_{t}}\rightarrow u_{t}$ 弱$^{*}$收敛.} \end{eqnarray} $ (4.24)
$ \begin{eqnarray} &&\mbox{在空间 $L^{2}(0, T;D(A))$ 中, $u_{n}\rightarrow u$ 强收敛.} \end{eqnarray} $ (4.25)
$ \begin{eqnarray} &&\mbox{在空间 $L^{2}(0, T;V)$ 中, $u_{n_{t}}\rightarrow u_{t}$ 强收敛.} \end{eqnarray} $ (4.26)

下面, 依次处理(4.14) 式中的每一项.首先, 由(4.25) 式有

$ \lim\limits_{n\rightarrow\infty}\lim\limits_{m\rightarrow\infty} \int^{T}_{0}\int_{\Omega}(f(u_{n}(s))-f(u_{m}(s))) (Au_{n}(s)-Au_{m}(s)){\rm d}x{\rm d}s=0. $ (4.27)

进一步, 根据引理4.1, 可知

$ \lim\limits_{n\rightarrow\infty}\lim\limits_{m\rightarrow\infty} \int^{T}_{0}\int_{\Omega}(f(u_{n}(s))-f(u_{m}(s))) (Au_{n_{t}}(s)-Au_{m_{t}}(s)){\rm d}x{\rm d}\tau=0, $ (4.28)
$ \lim\limits_{n\rightarrow\infty}\lim\limits_{m\rightarrow\infty} \int^{T}_{0}\int^{T}_{s}\int_{\Omega}(f(u_{n}(\tau))-f(u_{m}(\tau))) (Au_{n_{t}}(\tau)-Au_{m_{t}}(\tau)){\rm d}x{\rm d}\tau{\rm d}s=0. $ (4.29)

因此, 结合(4.27)-(4.29) 式, 可得$\phi_{T}(\cdot, \cdot)\in {\mathfrak C}(B_{1})$.

定理2.6的证明  由引理3.1和定理4.1可知, $({\cal H}_{1}, S(t))$是耗散的动力系统且是渐近紧的.则由定理2.5可得其紧的吸引子的存在性.

参考文献
[1] Lazer A C, McKenna P J. Large-amplitude periodic oscillations in suspension bridge:some new connections with nonlinear analysis. SIAM Review, 1990, 32: 537–578. DOI:10.1137/1032120
[2] Ahmed N U, Harbi H. Mathematical analysis of dynamic models of suspension bridges. SIAM J Appl Math, 1998, 58(3): 853–874. DOI:10.1137/S0036139996308698
[3] Litcanu G. A mathematical of suspension bridge. Appl Math, 2004, 49(1): 39–55. DOI:10.1023/B:APOM.0000024519.46627.4f
[4] Zhong C K, Ma Q Z, Sun C Y. Existence of strong solutions and global attractors for the suspension bridge equations. Nonlinear Anal, 2007, 67: 442–454. DOI:10.1016/j.na.2006.05.018
[5] Temam R. Infinite Dimensional Dynamical Systems in Mechanics and Physics. New York:Springer-Verlag, 1997 http://www.springer.com/us/book/9780387948669
[6] Ma Q Z, Zhong C K. Existence of strong solutions and global attractors for the coupled suspension bridge equations. J Diff Equ, 2009, 246: 3755–3775. DOI:10.1016/j.jde.2009.02.022
[7] Ma Q Z, Wang S P, Chen X B. Uniform compact attractors for the coupled suspension bridge equations. Appl Math Comput, 2011, 15: 6604–6615.
[8] Park J Y, Kang J R. Pullback D-attractors for non-autonomous suspension bridge equations. Nonlinear Anal, 2009, 71: 4618–4623. DOI:10.1016/j.na.2009.03.025
[9] Ma Q Z, Wang B L. Existence of pullback attractors for the coupled suspension bridge equations. Electronic J Diff Equ, 2011, 16: 1–10.
[10] Kang J R. Pullback attractors for the non-autonomous coupled suspension bridge equations. Appl Math Comput, 2013, 219: 8747–8758.
[11] Park J Y, Kang J R. Global attractor for suspension bridge equation with nonlinear damping. Quart Appl Math, 2011, 69: 465–475. DOI:10.1090/qam/2011-69-03
[12] Kang J R. Long-time behavior of a suspension bridge equations with past history. Appl Math Comput, 2015, 265: 509–519.
[13] Pata V, Zucchi A. Attractors for a damped hyperbolic equation with linear memory. Advances Math Sci Appl, 2001, 2: 505–529.
[14] Giorgi C, Muñoz Rivera J E, Pata V. Global attractors for a semilinear hyperbolic equation in viscoelasticity. J Math Anal Appl, 2001, 260: 83–99. DOI:10.1006/jmaa.2001.7437
[15] Fabrizio M, Giorgi C, Pata V. A new approach to equations with memory. Arch Rati Mech Anal, 2010, 198: 189–232. DOI:10.1007/s00205-010-0300-3
[16] Sun C Y, Cao D M. Non-autonomous wave dynamics with memory-asymptotic regularity and uniform attractor. Discrete Continuous Dyn Syst Ser B, 2008, 9: 743–761. DOI:10.3934/dcdsb
[17] Zhong C K, Yang M H, Sun C Y. The existence of global attractors for the norm-to-weak continuous semigroup and application to the nonlinear reaction-diffusion equations. J Diff Equ, 2006, 223(2): 367–399. DOI:10.1016/j.jde.2005.06.008
[18] Sun C Y, Cao D M, Duan J Q. Non-autonomous dynamics of wave equations with nonlinear damping and critical nonlinearity. Nonlinear Anal, 2006, 19: 2645–2665. DOI:10.1088/0951-7715/19/11/008
[19] Khanmamedov A Kh. Global attractors for wave equations with nonlinear interior damping and critical exponents. J Diff Equ, 2006, 230: 702–719. DOI:10.1016/j.jde.2006.06.001
[20] Chueshov I, Lasiecka I. Von Karman Evolution Equations:Well-Posedness and Long-Time Dynamics. New York:Springer, 2010 http://www.springer.com/us/book/9780387877112
[21] Kang J R. Global attractors for an extensible beam equation with localized nonlinear dmping and linear memory. Math Meth Appl Sci, 2011, 34: 1430–1439. DOI:10.1002/mma.v34.12