王有剛 武懷勤
摘要本文考慮了具有混合時(shí)變延時(shí)和不同時(shí)標(biāo)的混沌憶阻競爭神經(jīng)網(wǎng)絡(luò)的自適應(yīng)同步問題.使用Lyapunov泛函方法和不等式分析技術(shù),設(shè)計(jì)了一類新的具有反饋控制律的自適應(yīng)控制器以取得網(wǎng)絡(luò)同步及指數(shù)同步目的,提出了不用過多計(jì)算,如求解線性矩陣不等式或復(fù)雜代數(shù)計(jì)算的保證網(wǎng)絡(luò)同步條件;同時(shí),所獲條件也可以應(yīng)用到已有文獻(xiàn)里關(guān)于憶阻器網(wǎng)絡(luò)不同數(shù)學(xué)模型中.最后,通過實(shí)例驗(yàn)證了本文獲得的理論結(jié)果的有效和正確性.關(guān)鍵詞自適應(yīng)同步;憶阻器;競爭神經(jīng)網(wǎng)絡(luò);時(shí)間延時(shí);時(shí)標(biāo)
中圖分類號(hào)O429
文獻(xiàn)標(biāo)志碼A
0 引言
1983年,在文獻(xiàn)[1]中,Cohen和Grossberg為模擬神經(jīng)生物學(xué)中的細(xì)胞抑制現(xiàn)象提出了競爭神經(jīng)網(wǎng)絡(luò)模型.隨后,Meyer-Bse等[2]提出了具有不同時(shí)標(biāo)的競爭神經(jīng)網(wǎng)絡(luò),該網(wǎng)絡(luò)不僅模型了神經(jīng)激勵(lì)層的動(dòng)態(tài)行為——短時(shí)記憶,而且也模型了神經(jīng)突觸變化的動(dòng)力學(xué)行為——長時(shí)記憶,同時(shí),該網(wǎng)絡(luò)系統(tǒng)的狀態(tài)以兩個(gè)不同時(shí)標(biāo)在進(jìn)行變化,一個(gè)與神經(jīng)網(wǎng)絡(luò)狀態(tài)的快速變化有關(guān),另一個(gè)與外部刺激下突觸的緩慢變化有關(guān).在文獻(xiàn)[3]中,Meyer-Bse等進(jìn)一步研究了具有不同時(shí)標(biāo)的競爭神經(jīng)網(wǎng)絡(luò)的全局指數(shù)穩(wěn)定性.
為了保證相應(yīng)的信息存儲(chǔ),需要設(shè)計(jì)大型有效的神經(jīng)網(wǎng)絡(luò).隨著維數(shù)的增加,這將會(huì)占用大量的計(jì)算機(jī)內(nèi)存和硬盤空間.2008年,惠普實(shí)驗(yàn)室研究人員成功研制出了一種納米級(jí)電子設(shè)備,稱之為憶阻器[4-5]. 根據(jù)數(shù)學(xué)關(guān)系,憶阻器是一個(gè)非線性時(shí)變原件,它的值(即憶導(dǎo)值)依賴于先前通過的電流值,因而該設(shè)備擁有記憶能力,這與神經(jīng)系統(tǒng)中的突觸具有相似性.基于此特性,憶阻器已被應(yīng)用于納米記憶、計(jì)算機(jī)邏輯等領(lǐng)域[6-7].
使用憶阻器代替?zhèn)鹘y(tǒng)神經(jīng)網(wǎng)絡(luò)中的電阻,能夠設(shè)計(jì)憶阻神經(jīng)網(wǎng)絡(luò)[8-9].現(xiàn)有大量文獻(xiàn)討論了憶阻網(wǎng)絡(luò)的同步問題.在文獻(xiàn)[10]中,通過周期性的間斷控制,得到一些新的確?;趹涀璧幕煦缟窠?jīng)網(wǎng)絡(luò)的指數(shù)同步的充分性代數(shù)條件;文獻(xiàn)[11]則利用廣義的Halanay不等式和Lyapunov-Krasovskii泛函方法,提出了耦合憶阻神經(jīng)網(wǎng)絡(luò)弱的、修正的和泛函投影同步條件;基于極值分析理論,文獻(xiàn)[12]證明了具有延時(shí)的憶阻神經(jīng)網(wǎng)絡(luò)的周期解的存在性.
另一方面,具有不同時(shí)標(biāo)的競爭神經(jīng)網(wǎng)絡(luò)的同步問題也得到了研究.基于設(shè)計(jì)的反饋控制器,文獻(xiàn)[13]提出了代數(shù)和線性矩陣不等式形式的同步條件;文獻(xiàn)[14]針對具有混合時(shí)滯和不確定混合擾動(dòng)的競爭神經(jīng)網(wǎng)絡(luò),設(shè)計(jì)了一種簡單魯棒自適應(yīng)控制器,該控制器具有較好的抗干擾能力.
在本文中,我們研究具有混合時(shí)變延時(shí)和不同時(shí)標(biāo)的憶阻神經(jīng)網(wǎng)絡(luò)的同步問題.利用Lyapunov泛函方法與不等式分析技術(shù),通過設(shè)計(jì)兩個(gè)新的簡單有效的自適應(yīng)控制器,給出了完全同步與指數(shù)同步
的條件.所設(shè)計(jì)的自適應(yīng)控制器能夠適用于其他具有不同數(shù)學(xué)模型的憶阻神經(jīng)網(wǎng)絡(luò).與文獻(xiàn)[15-21]的結(jié)果相比較,本文建立的同步條件的優(yōu)點(diǎn)是不需求解線性矩陣不等式或計(jì)算代數(shù)方程等過多的復(fù)雜計(jì)算.
參考文獻(xiàn)
References
[1] Cohen M A,Grossberg S.Absolute stability of global pattern formation and parallel memory storage by competitive neural networks[J].IEEE Transactions on Systems,Man and Cybernetics,1983,SMC-13(5):815-826
[2] Meyer-Bse A,Ohl F,Scheich H.Singular perturbation analysis of competitive neural networks with different time scales[J].Neural Computation,1996,8(8):1731-1742
[3] Meyer-Bse A,Pilyugin S S,Chen Y.Global exponential stability of competitive neural networks with different time scales[J].IEEE Transactions on Neural Networks,2003,14(3):716-719
[4] Strukov D B,Snider G S,Stewart D R,et al.The missing memristor found[J].Nature,2008,453(7191):80-83
[5] Tour J M,He T.Electronics:The fourth element[J].Nature,2008,453(7191):42-43
[6] Ventra M D,Pershin Y,Chua L O.Circuit elements with memory:Memristors,memcapacitors,and meminductors[J].Proceedings of the IEEE,2009,97(10):1717-1724
[7] Kim K H,Gaba S,Wheeler D,et al.A functional hybrid memristor crossbar-array/CMOS system for data storage and neuromorphic applications[J].Nano Letters,2011,12(1):389-395
[8] Pershin Y V,Ventra M D.Experimental demonstration of associative memory with memristive neural networks[J].Neural Networks,2010,23(7):881-886
[9] Itoh M,Chua L O.Memristor cellular automata and memristor discrete-time cellular neural networks[J].International Journal of Bifurcation and Chaos,2011,19(11):1-8
[10] Zhang G D,Shen Y.Exponential synchronization of delayed memristor-based chaotic neural networks via periodically intermittent control[J].Neural Networks,2014,55:1-10
[11] Wu H Q,Li R X,Yao R,et al.Weak,modified and function projective synchronization of chaotic memristive neural networks with time delays[J].Neurocomputing,2015,149(22):667-676
[12] Wu H Q,Li R X,Ding S B,et al.Complete periodic adaptive antisynchronization of memristor-based neural networks with mixed time-varying delays[J].Canadian Journal of Physics,2014,92(11):1337-1349
[13] Lou X Y,Cui B T.Synchronization of competitive neural networks with different time scales[J].Physica A:Statistical Mechanics and its Applications,2007,380(1):563-576
[14] Shi Y C,Zhu P Y.Synchronization of memristive competitive neural networks with different time scales[J].Neural Computing and Applications,2014,25(5):1163-1168
[15] Yang X S,Cao J D,Yu W W.Exponential synchronization of memristive Cohen-Crossberg neural networks with mixed delays[J].Cognitive Neurodynamics,2014,8(3):239-249
[16] Guo Z Y,Yang S Y,Wang J.Global exponential synchronization of multiple memristive neural networks with time delay via nonlinear coupling[J].IEEE Transactions on Neural Networks and Learning Systems,2015,26(6):1300-1311
[17] Zhang G D,Hu J H,Shen Y.Exponential lag synchronization for delayed memristive recurrent neural networks[J].Neurocomputing,2014,154(22):86-93
[18] Wang L M,Shen Y,Yin Q,et al.Adaptive synchronization of memristor-based neural networks with time-varying delays[J].IEEE Transactions on Neural Networks and Learning Systems,2015,26(9):2033-2042
[19] Filippov A.Differential equations with discontinuous right-hand side[J].Journal of Mathematics and Its Applications,1991,154(2):377-390
[20] Zhang G D,Shen Y.New algebraic criteria for synchronization stability of chaotic memristive neural networks with time-varying delays[J].IEEE Transactions on Neural Networks and Learning Systems,2013,24(10):1701-1707
[21] Zhang G D,Shen Y,Wang L M.Global anti-synchronization of a class of chaotic memristive neural networks with time-varying delays[J].Neural Networks,2013,46(11):1-8