Scholarly article on topic 'Exponential stability criteria for fuzzy bidirectional associative memory Cohen-Grossberg neural networks with mixed delays and impulses'

Exponential stability criteria for fuzzy bidirectional associative memory Cohen-Grossberg neural networks with mixed delays and impulses Academic research paper on "Mathematics"

CC BY
0
0
Share paper
Academic journal
Adv Diff Equ
OECD Field of science
Keywords
{""}

Academic research paper on topic "Exponential stability criteria for fuzzy bidirectional associative memory Cohen-Grossberg neural networks with mixed delays and impulses"

He and Chu Advances in Difference Equations (2017) 2017:61 DOI 10.1186/s13662-017-1082-9

0 Advances in Difference Equations

a SpringerOpen Journal

RESEARCH

Open Access

Exponential stability criteria for fuzzy bidirectional associative memory Cohen-Grossberg neural networks with mixed delays and impulses

CrossMark

Weina He and Longxian Chu*

Correspondence: chulongxian_pdsu@126.com Software College, Pingdingshan University, Pingdingshan,467000, PRChina

Abstract

This paper is concerned with fuzzy bidirectional associative memory (BAM) Cohen-Grossberg neural networks with mixed delays and impulses. By constructing an appropriate Lyapunov function and a new differential inequality, we obtain some sufficient conditions which ensure the existence and global exponential stability of a periodic solution of the model. The results in this paper extend and complement the previous publications. An example is given to illustrate the effectiveness of our obtained results.

MSC: 34C20; 34K13; 92B20

Keywords: fuzzy BAM Cohen-Grossberg neural networks; exponential stability; mixed delays; periodic solution; impulse

ft Spri

ringer

1 Introduction

In recent years, considerable attention has been paid to bidirectional associative memory (BAM) Cohen-Grossberg neural networks [1] due to their potential applications in various fields such as neural biology, pattern recognition, classification of patterns, parallel computation and so on [2-4]. In real life, numerous application examples appear, for example, emerging parallel/distributed architectures were explored for the digital VLSI implementation of adaptive bidirectional associative memory (BAM) [5], Teddy and Ng [6] applied a novel local learning model of the pseudo self-evolving cerebellar model articulation controller (PSECMAC) associative memory network to produce accurate forecasts of ATM cash demands. Chang et al. [7] proposed a maximum-likelihood-criterion based on BAM networks to evaluate the similarity between a template and a matching region. Sudo et al. [8] proposed a novel associative memory that operated in noisy environments and performed well in online incremental learning applying self-organizing incremental neural networks. On the one hand, the existence and stability of the equilibrium point of BAM Cohen-Grossberg neural networks plays an important role in practical application. On the other hand, time delay is inevitable due to the finite switching speed of amplifiers in the electronic implementation of analog neural networks, moreover, time delays may have important effect on the stability of neural networks and lead to periodic oscillation,

© The Author(s) 2017. This article is distributed under the terms of the Creative Commons Attribution 4.0 InternationalLicense (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use,distribution, and reproduction in any medium, provided you give appropriate credit to the originalauthor(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

bifurcation, chaos and so on [3, 9, 10]. Thus many interesting stability results on BAM Cohen-Grossberg neural networks with delays have been available [11-30].

As is well known, numerous dynamical systems of electronic networks, biological neural networks, and engineering fields often undergo abrupt change at certain moments due to instantaneous perturbations which leads to impulsive effects [4,19, 20, 31-36]. Many scholars [37,38] think that uncertainty or vagueness often appear in mathematical modeling of real world problems, thus it is necessary to take vagueness into consideration. Fuzzy neural networks (FNNs) pay an important role in image processing and pattern recognition [37] and some results have been reported on stability and periodicity of FNNs [11,3941]. Here we would like to point out that most neural networks involve negative feedback terms and do not possess amplification functions or behaved functions. The model (1.1) of this paper has amplifications function and behaved functions which differ from most neural networks with negative feedback term. Up to now, there are rare papers that consider exponential stability of this kind of fuzzy bidirectional associative memory CohenGrossberg neural networks with mixed delays and impulses.

Inspired by the discussion above, in this paper, we are to consider the following fuzzy bidirectional associative memory Cohen-Grossberg neural networks with mixed delays and impulses,

xi(t) = n(xl(t))[-ai(t,xi(t)) + Yj=1 cJi(t)fj(yJ(t- r(t)))

+ AA aji(t)flœKji(t - s)fj(yJ(s)) ds + A^i Tjiuj + V™iHjiuj + VA j(t)f-œKji(t - s)fj(yj(s)) ds + Ii(t)], t = tk,l e I, Axi(tk) = xi(tk) — Xi(t-) = -YikXi(t-) + EA eij(t-)Ej(yj(t- - r)), k e Z+ yj(t) = Vj(yj(t))[-bj(t,yj(t)) + dij(t)gi(xi(t- r(t)))

+ A n=i Pij (t)f-œ Nij (t - s)gi (xi(s)) ds + AL SijUi + VL LijUi + vn=i qij(t)flœNij(t - s)gi(xi(s)) ds + /j(t)], t = tk,j e h, Ayj(tk) =yj(tk) - yj(t-) = Sjkyi(tk) + Eti hji(t-)Hl(Xl(t— - r)), k e Z

with initial conditions

|xi(s) = ^ii(s), s e (-TO,0],i e I, I yj(s) = <fo(s), s e (-to, 0],j e h,

where n and m correspond to the number of neurons in X-layer and Y-layer, respectively. xi(t) and yj(t) are the activations of the ith neuron and the jth neurons, respectively. ii(-) and ûj(-) are the abstract amplification functions, ai (t, •) and bj(t, •) stand for the rate functions with which the ith neuron and jth neuron will reset its potential to the resting state in isolation when disconnected from the network and external inputs; aji(t), Pji (t), Tji and Hji are elements of fuzzy feedback MIN template and fuzzy feedback MAX template, fuzzy feed-forward MIN template and fuzzy feed-forward MAX template in X-layer, respectively; Pij(t), qij(t), Sij and Lij are elements of fuzzy feedback MIN template and fuzzy feedback MAX template, fuzzy feed-forward MIN template and fuzzy feed-forward MAX template in Y-layer, respectively; /\ and \/ denote the fuzzy AND and fuzzy OR operation, respectively; uj, ui denote external input of the ith neurons in X-layer and external input of the jth neurons in Y-layer, respectively; Ii(t) and Jj (t) are external bias of X-layer and Y-layer, respectively,f (0 andg^-) are signal transmission functions, Kji(t) andNj(t) are delay

kernels, i = {1,2,...,«}, h = {1,2,..., m}, Z+ denotes the set of positive integral numbers, the impulse times tk satisfy 0 = to < t1 < t2 < ••• < tk < •••, limk^TO tk = to, <ii(), <2j() e C, where C denotes real-valued continuous functions defined on (-to, 0], t(t) is the transmission delay such that 0 < t(t) < t, t is a positive constant, eij(t-) represents impulsive perturbations of the ith unit at time tk, hji(t-) represents impulsive perturbations of the jth unit at time tk, Ej(y,(t-)) represents impulsive perturbations of the jth unit at time tk and yj(t-) denotes impulsive perturbations of the jth unit at time tk caused by the transmission delays, Hi(xi(t-)) represents impulsive perturbations of the ith unit at time tk and xi(t-) denotes impulsive perturbations of the ith unit at time tk which caused by the transmission delays. For details, see [42-44].

The main purpose of this paper is to investigate the existence and global exponential stability of a periodic solution of fuzzy BAM Cohen-Grossberg neural networks with mixed delays and impulses. By constructing a suitable Lyapunov function and a new differential inequality, we establish some sufficient conditions to ensure the existence and global exponential stability of a periodic solution of the model (1.1). The results obtained in this paper extend and complement the previous studies in [4,10]. Two examples are given to illustrate the effectiveness of our theoretical findings. To the best of our knowledge, there are very few papers that deal with this aspect. Therefore we think that the study of the fuzzy BAM Cohen-Grossberg neural networks with mixed delays and impulses has important theoretical and practical value. Here we shall mention that since the existence of amplifications function and behaved functions in model (1.1), thus there are some difficulties in dealing with the exponential stability. We will apply some inequality techniques, meanwhile, the construction of Lyapunov function is a key issue.

The remaining part of this paper is organized as follows. In Section 2, the necessary definitions and lemmas are introduced. In Section 3, we present some new sufficient conditions to ensure the existence and global exponential stability of a periodic solution of model (1.1). In Section 4, an illustrative example is given to show the effectiveness of the proposed method. A brief conclusion is drawn in Section 5.

2 Preliminaries

Let R denote the set of real number, Rn the «-dimensional real space equipped with the Euclidean norm | R+ the set of positive numbers. Denote PC(R, R+) = {< : R ^ Rn : <(t) is continuous for t = tk, <(t+), <£(t-) e Rn and ^(t-) = 0(tk)}.

Throughout this paper, we make the following assumptions:

(HI) For i e i, j e h, c,,(t), aji(t), j(t), etj(t), dj(t),pij(t), qij(t), hji(t), t(t), h(t) and Jj(t) are all continuously periodic functions defined on t e [0, to) with common period rn >0.

(H2) For i e i, j e h, there exist positive constants Lf, LE, Lg and LH such that

l/j(«)-Jj(v)| < Lfj |u - v|, |Ej(u)-Ej(v)| < LE|u - v|, gi(u)-gi(v)| < Lg|u- v|, |Hi(u)-Hi(v)| < LH|u - v|

for all u, v e R.

(H3) For i e i, j e h, t^-) and ^y(^) are continuous and satisfy 0 < ii < ti(^) < li, 0 < < (•) < , where ii, ii, are some positive constants.

(H4) For i e t, j e h, there exist continuous positive «-periodic functions ^(t) and aj(t) such that

ai(t, u)-at(t, v) bj(t, u)-bj(t, v)

> Qi(t), -:—;-> Oj(t)

for all u, v e R.

(H5) For i e I, j e ft, the delay kernels Kij(■), N'KO e C(R+, R+) are piecewise continuous and satisfy Kj(s) < K(s) andNji(s) < K(s) for all s e R+, where K(s) e C(R+,R+) and integrable, satisfying K(s)e¡s ds < to, in which the constant ¡ denotes some positive number.

(H6) For i e I, j e ftv >0, there exists q e Z+ such that tk + v = tk+q and yik = yi(k+q), Sjk = 5j(k+q), k e Z+.

(H7) For i e j e ft, j = maxte[o,®] |c/i(t)|, a- = maxte[o,®] |a/i(t)|, P- = maxte[o,®] |^/i(t)|, e- = maxte[o,ffl] |eij(t)|, d- = maxte[o,®] |dj(t)|,p- = maxte[o,®] pij(t)|, q- = maxte[o,ffl] |qij(t)|, h- = maxte[o,®] |hji(t)|, e* = minted |gi(t)|,

a* = minte[o,v] |oj(t)|.

In this paper, we use the following norm of Rn+m:

= £ x + £ j i=l j=i

se(-TO,o]

£Ms)| + J2 l^2j(s)| . i=l j=l

for u = (1, X2,...,Xn, yi, y2,---,ym)T e Rn+m, 0 = (011, 012,...,01n, 021, 022,-.., 02m) T e Cn Lemma 2.1 ([37]) Let x and y be two states of system (1.1). Then

/\ aij(t)gj(x) -f\ aij(t)gj(y) j=i j=i

|aj(t)|g (x)-gj(y)|

\J Pij(t)gj(x)-\J Pij(t)gj(y) j=i j=i

<£ |Aj(t)|^gj(x)-gj(y)|.

Lemma 2.2 ([4]) Letp, q, r and r denote nonnegative constants andf e PC(R, R+) satisfies the scalar impulsive differential inequality

j D+f(t) < -p/(t) + q supt_T<s<tf(s) + k(s)f (t - s) ds, t = tk, t > to, f (tk) < a/(t-) + bkf(t- - r), k e Z+,

where 0 < a < +c, au, bk e R, k(-) e PC([0, a], R+) satisfies k(s)ems ds < cfor some positive constant no > 0 in this case when a = +c. Moreover, when a = +cc, the interval [t - a, t] is understood to be replaced by (-c, t]. Assume that (i) p > q + r J0 k(s) ds. (ii) There exist constant M > 0, n >0 such that

max{i, ak + bkekr} < Men(tn-to), n e Z+,

where X e (0, no) satisfies

X < p - qeXT - r i k(s)eXs ds. Jo

f (t) < Mf (to)e-(X-n)(t-to), t > to,

wheref (to) = SUPto-max{ff ,T f(s).

3 Global exponential stability of the periodic solution

In this section, we will discuss the global exponential stability of the periodic solution for (1.1).

Theorem 3.1 Assume that (H1)-(H7) hold, then there exists a unique a>--periodic solution of system (1.1) which is globally exponentially stable if the following conditions are fulfilled. (H8)

minfe*, a*} minielJeh{ii, ûj} maxig/ygR {ii, û j}

V max âd, Y^ max d*:Lg > ..f je» j j —^ id j J

In n mm]

V max aiLf, Y^ max â,ÎLf, Y^ max »*Lg, Y^ max q*Lg >

je» ji j ^ jeh ji j ^ iel ij i ^ iel ij i

i=1 ' i=1 ' j=1 j=1 J

x / K (s) ds. Jo

(H9) There exist constants M > 1, X e (0, X0) and n e (0, X) such that nn=1 max{1, x;} < Mentn /or a// n e Z+ ho/ds and

minfe*, a*} mimeiJeh {îp É.,} [A f A J Ar

X <---——-— - max I max cjiLJJ , max dlL° ^ eXT

maxieiyeRU;,ûj} jeR ' ' i

In n m m A

V max afiLf, Y^ max ß ÎLf, Y^ max »*-Lf, Y^ max öi-Lf

jeR ji j ^ jeR ji j iel ij ^ iel ij i

i=l ' i=l ' j=l j=l J

x / K (s) ds, Jo

= m^,^ûj} max\\l-yu|, \l-j| + maxi Vmaxe^jLf, Vmax¿iLfleXT).

miniei,jeR{ii, ûj} ieljeR [ j [ ^ jeR ij j -j-^ iel ji i J J

Proof Assume that u(t) = (xd(t, <1), X2(t, <1),...,x„(t, <1), y:(t, <2), y2(t, <i),...,ym(t, <2))T is an arbitrary solution of system (1.1) through (t,<1,<2), where <1 = (<11,<12,...,<1n)T,

02 = (021,022,..., 02m)T. Define

xi(t + «, 01)=<1i, t < 0, i e t, yj(t + «, 02) = <2j, t < 0, j e h,

then < = (<u, ^12,..., ^1n)T e Cn, < = (<21, <22,..., <2m)T e Cm. Now we construct the following Lyapunov function:

n nxi(t+m) 1

V(t) = £ / — ds Sgn[x;(t + «) -Xi(t)]

^x;(t) ii(s) nyj(t+m) 1

j=1 j j s) It is easy to see that

ds sgnj + «)—y,(t)].

1, ^ I IV k (t +«)- xi(t) | + V |yj (t +«) — y (t) | ietje4 ii j\tr tr y

< V(t) < maxj p ^J X>(t + «)-xi(t)| + £ |y (t + «) — yj (t)| j. (3.2)

When t = tk, calculating the derivative of D+V(t) along the solution of (1.1), we have " D+xi(t + «) D+xi(t)'

D+V(t) < £

Ji(xi(t + «)) li(xi(t))_ D+yj(t + «) D+yj(t)

sgn[xi(t + «)—xi(t)] sgnj + «)—y;(t)]

L^j (yj (t + «)) fy (yj(t))J

—ai(t, xi(t + «)) + ai(t, xi(t)) + £ cji(t)fj(yj(t + « — r (t + «)))

m m ,, t+«

J2cji(t)fj(yj(t — r(t)))+Aaji(t) Kji(t + « — s)fj(yj(s)) ds j=1 j=1 •/—c

m p t m p t+« f\aji(t) I Kji(t — s)f (y,(s)) ds + \/ jt) Kji(t + « — s)f (yj(s)) ds

J—c J — C

\JPji(t) / Ki(t — s)j(yj(s)) ds

•1 «/—C

sgn[xi(t + «) — xi(t)]

—bj(t, yj(t + «)) + bj(t, yj(t) + £ dij (t)gi(xi{t + « — r (t + «)))

n n p t+« — £ dij (t)gi(xi(t — r (t))) + /\ pf (t) / Nj(t + « — s)gi(xi(s)) ds

i=1 i=1

n p t n p t+« /\pij (t) Nij (t — s)gi(xi(s)) ds + \/ qij (t) Nij(t + « — s)gi(xi(s)) ds

• 1 «/—C r_1 «/—C

" pt+w

"V qij(t) Nij(t + w - s)gi(xi(s)) ds sgnj + w)-yj(t)]

i=1 •/-to j

< - mi e*J2 Ht + w)- Xi (t) | + CjiLfj j + w - t (t)) - yj (t - t (t)) |

i i=1 i=1 j=1

n m /. to

+ Z Z O/iL/ / K(s) |yj(t + w - s)-yj(t - s) | ds i=1 j=1

n m /. to

+ E E ^jiLf / K(s) |yj(t + w - s)-yj(t - s) | ds

i=1 j=1

- mR 0* Z|y(t + w)-yj(t)| + ^ J2dijLg№ + w - t(t)) - xt(t - t(t)) |

je j=1 j=1 i=1

m n /. to

+ ZZ PjLg // (s) |xi (t + w - s)-Xi(t - s)| ds

j=1 i=1

m n /. to

+ qi-Lg^ / K(s)|xi(t + w -s)-Xi(t-s)| ds

j=1 i=1

< -min{e*,a*} £(xi(t + w)-xi(t)) + j + w) -y,(t)) . i=1 j=1 nm

+ Z mmjx |yj(t + w - t(t)) - yj(t - t(t)) | i=1 je j=1

n j> to

+ y^ max a-iLfj J K(s) |y;(t + w - s)- yj (t - s) | ds

n „ to m

+ J2 max jj / K(s)J^ j + w - s)- y;(t - s) | ds i=1 ' j0 j=1 mn

+ y^maxd*jLg^2^(t + w - t(t)) -x^t- t(t))|

j=1 iei i=1

m /. to n

+ y^maxp*jLg I K(s)^2|xi(t + w-s)-xi(t-s)|ds

j=1 iei i=1

m /. to n

y^maxq*jLg I K(s) ^|xi(t + w - s)-xi(t - s)| ds

iei i=1

E(xi(t + w)-xi(t)) + y]j + w) -yj(t))

< - min{e*, a/}

max c - Lf , V max d-Lg

^ jeR ji ' iel ij i

Y}xi{t + w - T(t))- xi(t - T(t)) | + ^|yj(t + w - T(t))-yj(t - T(t)) | L i=1 j=1

In n m m

V max alLf, Y^ max P iLf, Y^ max p*,Lg, Y^ max q*Lf

/eR ji j ^ jeh ji ' ^ iet ij i ^ iet ' i i=1 i=1 j=1 j=1

x / K(s) J0

^ |xi(t + « — s)—xi(t — s)| + ^(t + « — s)— y (t — s)|

L i=1 j=1

ds. (3.3)

In view of (3.2), it follows from (3.3) that

D+ V(t) < — miniQ*i,a*} min [iu,±i}V(t)

' iet,jeh '

y maxclLf,, Y^maxd*iLgi\ max [ii, 1j}V(t — r(t))

jeh ' ^^ iet ' J iet,jeh

In n m m \

V max alLf, Y^ max PJiLf, Y^ max »*Lg, Y^ max q*,Lg I jeh ji j jeh ji j iet ij i iet ij i

i=1 ' i=1 ' j=1 j=1 J

x / K(s)V(t — s)ds. (.4)

When t = tk, in view of (H6), (H7), and (3.2), we get

V(tk) < ji", ^£|xi(tk + «) — xi(tk)| + £|yj(tk + «) —yj(tk)| j = max 11, -11 ( V|xi(tk+q) —xi(tk)| + V|yj(tk+q) —yj(tk)| )

ietjeH ii jvtr ^ /

< imjh{ i1' |} {V11—Yik 1 |xi(t—+q} — xi(t—} |

+ EE ^ |yj(t—+q — r) — yj(t— — ^ |

i=1 j=1

m m n 1

+ E |1 — jI fc^+q) — yj(t—) | + E J2 hjL Htlt+q — r) — xi(t— — r) |

j=1 j=1 i=1

11 W n

| -v k / xil

U —j) I i=1

< ma^ i max |1 —^I VUiitf, + «) — x^t—)|

iet,jeh i. 1A iet '

Emahx jf J2 |Уj(t- + « — r) — yj(t— — r) |

i=1 je j=1

+ max 11 — j ^ |yj (t— + «) — yj(t—)|

+ meatx h^Yl\x{*k+«—r)— x^t——r)|

i=1 ie j=1

< ma^ 1, V [ max {11 — YikI, |1 —j|}

iet,jeh \ i 1 iet,jeh \ -j. —) f

x |xi(tk+w)- xi(t-)| + J2 yj (t-+w)- »(£)|

\ i=1 j=1 I

Y max ejLf,^^ max h-LH

*r H I

^ jeR ' iel

I i=1 i=1

x |xi(tk+w)- x^t-)| + J2 jk +w)- yj (t-)|

\ i=1 j=1

maxiel,jeR{li, &j} (|1 . M . ,u.ux

—--—— max} |1-№ |, |1- Sjk |} V(t-)

miniei,jeR{li,} ieljeR ' ' yk'

+ maxiel,jeR{l^ maJ y max ejL, V maxh-iLH ivft- - t)

miniei,jeR{li,£;} jeR ij j , iei ji i V k '

In view of (3.4)-(3.5) and (H8)-(H9), using Lemma 2.2, we have

V(t) < MV(0)ek(Akn)t, t > 0, where V(0) = SUpkTO<s<o V(s). It follows from (3.6) that

£>(t + w) -xi(t)| + ^|yj(t + w) -yj(t)| < v||< - ^Уek(Akn)t, t > 0, i=1 j=1

maxiel jeR{li, &j}

v = m-i ,jeRL jJ > 1,

miniei,jeR{li, &j}

X satisfies the condition (H9). Notice that

xi(t + kw) =xi(t) + ^[xi(t + Iw) -x^t + (l- 1)w)], i e i,

yj(t + kw) = yj(t) + j + lw) - yj(t + (l - 1)w)], j e h.

In view of (3.7), we have

^[xi(t + lw) -x^t +(l - 1)w)]

= lim ^[xi(t + lw)-xi(t +(l-1)w)]

k—>to '

< v ||< - lim y

' k—TO ^

< v ||< - v ||ek(Akn)^ e^-^'-1^ < TO, as k — to,

for any given t > 0. By (3.8), we know that limk^TOxi(t + km) exists. Similarly, we know that limk^TOyj(t + km) also exists.

Set (x* (t), y* (t))T = (xi(t), xl(t),...,x*n(t), yi(t), f2(t),...,/m(t))T, where x* = lim^TO xi(t + km), y* = limk^TOyj(t + km), then (x*(t),y*(t))T is a periodic function with period m for system (1.1).

Assume that system (1.1) has another m-periodic solution (x**(t),y**(t))T as follows: (x* *(t, ti), y* *(t, f2))T = (x* *(t, ti), x* *(t, t0,..., xn* (t, t0,

y* *(t, t*), y* *(t, t*),..., ym (t, t*))T,

where t1 e Cn, t* e Cm. It follows from (3.7) that

£>*(t) -x**(t)| + £|yj(t) -yj*(t)|

i=1 j=1

= |x * (t + km) - x** (t + km) | + ^ |y* (t + km) - y** (t + km) |

i=1 j=1

< v- t ||e(A^)(t+km), t > 0. (3.9)

Let k ^ to, then x*(t) = x**(t), yj(t) = y**(t), t > 0. Thus we can conclude that system (1.1) has a unique m-periodic solution which is globally exponentially stable. The proof of Theorem 3.1 is complete. □

Remark 3.1 Li [4] investigated the existence and global exponential stability of a periodic solution for impulsive Cohen-Grossberg-type BAM neural networks with continuously distributed delays, the model in [4] is not concerned with fuzzy terms. Bao [10] discussed the existence and exponential stability of a periodic solution for BAM fuzzy CohenGrossberg neural networks with mixed delays, the model in [10] is not concerned with impulsive effects. Yang [16] considered the periodic solution for fuzzy Cohen-Grossberg BAM neural networks with both time-varying and distributed delays and variable coefficients, the model in [16] is not concerned with impulsive effect and distributed delays. Balasubramaniam et al. [38] analyzed the global asymptotic stability of stochastic fuzzy cellular neural networks with multiple time-varying delays, the model in [38] is not concerned with impulsive effect and distributed delays, Balasubramaniam and Vembarasan [41] studied the robust stability of uncertain fuzzy BAM neural networks of neutral-type with Markovian jumping parameters and impulses, the authors did not discuss the existence and global exponential stability of a periodic solution of neural networks and the model in [41] is also not concerned with distributed delays. In this paper, we study the exponential stability for fuzzy bidirectional associative memory Cohen-Grossberg neural networks with mixed delays and impulses. All the obtained results in [4, 10, 16, 38, 41] cannot be applicable to model (1.1) to obtain the exponential stability of model (1.1). From this viewpoint, our results on the exponential stability for fuzzy bidirectional associative memory Cohen-Grossberg neural networks with mixed delays and impulses are essentially new and complement earlier works to some extent.

4 Examples

In this section, we consider the following neural networks with mixed delays and impulses

X i(t) = il(xi(t))[-«i(t, Xi(t)) + e;=i ^my^ - r (t)))

+ Am a/i (t) /_TO A/i(t - s)f (yj(s)) ds + A/= + V;=i ^/i^/ + V2=i A'iW/l Kji(t _ s)f(y(s)) ds + /i(t)j, t = t*, XX2(t) = l2(X2(t))[_a2(t, x2(t)) + E;=i Cj2(t)fj(yj(t _ r(t))) + Am a/2(t) /_TO K/2(t _ s)f'(y (s)) ds + V?=i ^«/l K'2(t _ s)f(yi(s)) ds + /2(t)j, t = t*, AXi(t*) = xi(t*) _ xi(t_) = _yikXi(t_) + E/U el/'(t_)£/'(У/'(t_ _ r)), * e Z+, Ax2(t*) = X2(t*) _X2(t_) = _Y2*X2(t_) + E)Li e't^E^ ¡(t_ _ r)), * e Z+, (4 i) yyi(t) = #i(yi(t))[_6i(t, y(t)) + E2=i dii(t)gi(xi(t _ r (t))) '

+ ALiPii(t) Ntt(t _s)gi(Xi(s)) ds + A2=iSiiM + V2=iLiiUi + V2=i ifl (t) (t _ s)gi (Xi (s)) ds + Ji (t)], t = t*,

y2(t) = ^2(y2(t))[_^2(t,y(t)) + E2=i di2(t)gi(Xi(t _ r(t)))

+ ALPi2 (t) Nz(t _ s)gi(Xi(s)) ds +/\2=i 5i2Mi + V2=i Li2Mi + V2=i ia (t) Na(t _ s)gi (Xi (s)) ds + J2 (t)], t = t*, Ayi(t*)=yi(t*) -yl(t(() = _5i*yi(t_) + E2=ihii(t_)Hi(Xi(t_ _ r)), * e Z+, Ay2(t*) = y2(t*) _y2(t_) = S2*y2(t_) + E2=ih2i(t_)Hi(Xi(t_ _ r)), * e Z+,

cii(t) C2i(t)

C2i(t) C22(t)

dii(t) d2i(t)

di2(t) d22(t)

aii(t) a2i(t)

ai2(t) a22(t)

Pii(t) P2i(t)

Pi2 (t) P22(t)_

>ii(t) ^2i(t)'

Mt) fe(t)

?ii(t) ?2i(t)

?i2(t) <722 (t)

Tii T2i

Ti2 t22

0.9 + 0.2 sin t 0.5 + 0.4 cos t

0.4 + 0.2 sin t 0.5 + 0.i cos t

0.5 + 0.2 sin t 0.4 + 0.2 cos t

0.3 + 0.i sin t 0.5 + 0.2 cos t

0.5 + 0.2 cos t 0.5 + 0.2 sin t

0.4 + 0.2 cos t 0.4 + 0.2 sin t

0.5 + 0.2 cos t 0.4 + 0.2 sin t

0.3 + 0.i cos t 0.3 + 0.2 sin t

0.4 + 0.i sin t 0.6 + 0.3 cos t

0.3 + 0.2 sin t 0.5 + 0.3 cos t

0.5 + 0.3 cos t 0.4 + 0.i sin t

0.3 + 0.2 cos t 0.5 + 0.2 sin t

i i "Sn Sii' i i

i i , > S22 i i

Hii H2i i i "¿ii ¿21 i i

Hi2 H22 i i , Li2 ¿22 i i

i + sin t i + cos t i + sin t i + cos t

en(i) 612(f)

621(f) 622(f)

¿11 (f) ¿12(f)

¿21 (f) ¿22(f)

tanh(0.2f) tanh(0.2f) tanh(0.2f) tanh(0.2f)

tanh(0.3f) tanh(0.3f) tanh(0.3f) tanh(0.3f)

Y1k(s) Y2k(s) 81k (s) hk (s)

t!(*1(f)) ^1^1(f))

«1(f, X1(f)) b1(f, y1(f))

l2(X2(f)) &2(y2(f))

«2(f, X2 (f)) b2(f,y2(f))

0.55 + 0.05 cos(x1(f)) 4X1 (f)

0.55 + 0.05 sin(y1(f)) 3y1(f)

0.55 + 0.05 sin(x2 (f)) 4x2 (f) 0.55 + 0.05 cos(y2(f)) 3y2(f)

/<u(s) K21 (s) K12 (s) <22 (s)

e 3s e 3s

Nn(s) N21(s) N12(s) N22 (s)

„-3s

„ -3s

f^f- - T)) £2(y2(f- -1)) H^f- -T)) H2(X2(f- -T))

tanh(0.2y1(f- - 0.2)) tanh(0.2y2(f- - 0.2)) tanh(0.2X1(f- - 0.2)) tanh(0.2X2(f- - 0.2))

Let fk = 0.5nk, t(f) = 0.21 sin2f|,/1(u) = |u + 1|, g\_(u) = |u - 1|, then we get i1 = 0.6, i1 = 0.5, &1 = 0.6, = 0.5,12 = 0.6, i2 = 0.5, $2 = 0.6, = 0.5, c*n = 1.1, c21 = 0.9, ci2 = 0.6, c22 = 0.6, d* = 0.6, d*21 = 0.6, d*2 = 0.4, d*22 = 0.6, a* = 0.7, a^ = 0.7, a*2 = 0.6, a^ = 0.6, p* = 0.6, p21 = 0.6, = 0.4, p22 = 0.5, = 0.6, ^2*1 = 0.7, = 0.5, ^2 = 0.8, q* = 0.4, q21 = 0.9, q12 = 0.5, q22 = 0.7, h*1 = 1, ¿21 = 1, h*2 = 1, ¿22 = 1, t = 0.2, K(s) = e-3s, X0 = 1.5, w = 2n, L{ = if = 1, if = 0.2, Lf = 0.2, LH = 0.2, LH = 0.2. It is easy to check that

maxiei,jeR[Ti,

in m 1

V max c *L{, Y^ max d*Lg

Z—I ji j Z—I ij i

= 7.5,

I ^—' jgh

Emax a*L{, Y^ max jL, Y^ max p *.Lg, Y^ max q *iLg \ I K(s) ds

g ji j ¿—^ jgh ji j ^ igl 'j 1 igl ij ' | Jo

I z—' jgh

6.4237.

Choose X = 0.9 < X0 such that

X < 7.5 - 2e0,2X -1.4 / e(X-3)sds.

Then we obtain

maxiei,jeh[ii, & j}

minigiygh^i, &••}

x ma^ |1 - Yi/1, 11 - 8ji \ + ma^^ max e*:Lf, Y^ max h;*LH [ e'

iczp iczh I I ' ^ /eh II t J ici j I

e-3s e-3s

Figure 1 Numerical solutions of system (3.1): times series of x1.

Figure 2 Numerical solutions of system (3.1): times series of x2.

Thus we can choose n = 0.95 < X such that ]"["=i max{i, xi} = i.9503n < 2.5857" ^ ent" for all n e Z+. Then all the conditions of Theorem 3.i hold. Thus (4.i) has exactly one 2n-periodic solution which is globally exponentially stable. These results are illustrated in Figures i, 2, 3, 4.

5 Conclusions

In this article, we have analyzed the global exponential stability of fuzzy bidirectional associative memory Cohen-Grossberg neural networks with mixed delays and impulses. By constructing a suitable Lyapunov function and a new differential inequality, some sufficient criteria which ensure the existence and global exponential stability of a periodic solution of the model have been established. The obtained conditions are easy to check in practice. The results in this paper extend and complement some previous studies. Finally,

Figure 3 Numerical solutions of system (3.1): times series of y1.

Figure 4 Numerical solutions of system (3.1): times series of y2.

an example with their numerical simulations is carried out to illustrate the correctness. To the best of our knowledge, there are only rare results on the exponential stability for fuzzy bidirectional associative memory Cohen-Grossberg neural networks with proportional delays, which will be our future research direction.

Competing interests

The authors declare that there is no conflict of interest regarding the publication of this paper. Authors' contributions

The authors have equally made contributions. All authors read and approved the final manuscript. Acknowledgements

The authors would like to thank the referees and the editor for helpful suggestions incorporated into this paper. Received: 10 September 2016 Accepted:4 January 2017 Published online: 21 February 2017

References

1. Cohen, MA, Grossberg, S: Absolute stability of globalpattern formation and parallelmemory storage by competitive neuralnetworks. IEEE Trans. Syst. Man Cybern. 13, 815-826 (1983)

2. Kang, HY, Fu, XC, Sun, ZL: Globalexponentialstability of a periodic solutions for impulsive Cohen-Grossberg neural networks with delays. Appl. Math. Model. 39,1526-1535 (2015)

3. Wu, HX, Liao, XF, Feng, W, Guo, ST: Mean square stability of uncertain stochastic BAM neuralnetworks with interval time-varying delays. Cogn. Neurodyn. 6,443-458 (2012)

4. Li, XD: Existence and globalexponentialstability of periodic solution for impulsive Cohen-Grossberg-type BAM neuralnetworks with continuously distributed delays. Appl. Math. Comput. 215, 292-307 (2009)

5. Hasan, S, Siong, N: A parallelprocessing VLSIBAM engine. IEEE Trans. NeuralNetw. 8,424-436 (1997)

6. Teddy, SD, Ng, SK: Forecasting ATM cash demands using a locallearning modelof cerebellar associative memory network. Int. J. Forecast. 27, 760-776 (2011)

7. Chang, H, Feng, Z, Wei, X: Object recognition and tracking with maximum likelihood bidirectionalassociative memory networks. Neurocomputing 72, 278-292 (2008)

8. Sudo, A, Sato, A, Hasegawa, O: Associative memory for online learning in noisy environments using self-organizing incrementalneuralnetwork. IEEE Trans. NeuralNetw. 20, 964-972 (2009)

9. Zhou, LQ, Chen, XB, Yang, YX: Asymptotic stability of cellular neuralnetworks with multiple proportionaldelays. Appl. Math. Comput. 229,457-466 (2014)

10. Bao, HM: Existence and exponentialstability of periodic solution for BAM fuzzy Cohen-Grossberg neuralnetworks with mixed delays. NeuralProcess. Lett. 43, 871-885 (2016)

11. Sakthivel, R, Arunkumar, A, Mathiyalagan, K, MarshalAnthoni, S: Robust passivity analysis of fuzzy Cohen-Grossberg BAM neuralnetworks with time-varying delays. Appl. Math. Comput. 218, 3799-3809 (2011)

12. Zhang, ZQ, Liu, WB, Zhou, DM: Globalasymptotic stability to a generalized Cohen-Grossberg BAM neuralnetworks of neutraltype delays. NeuralNetw. 25, 94-105 (2012)

13. Zhou, DM, Yu, SH, Zhang, ZQ: New LMI-based conditions for globalexponentialstability to a class of Cohen-Grossberg BAM networks with delays. Neurocomputing 121, 512-522 (2013)

14. Jian, JQ, Wang, BX: GlobalLagrange stability for neutral-type Cohen-Grossberg BAM neuralnetworks with mixed time-varying delays. Math. Comput. Simul. 116,1-25 (2015)

15. Wang, DS, Huang, LH, Cai, ZW: On the periodic dynamics of a generalCohen-Grossberg BAM neuralnetworks via differentialinclusions. Neurocomputing 118, 203-214 (2013)

16. Yang, WG: Periodic solution for fuzzy Cohen-Grossberg BAM neuralnetworks with both time-varying and distributed delays and variable coefficients. NeuralProcess. Lett. 40, 51-73 (2014)

17. Balasubramaniam, P, Syed Ali, M: Stability analysis of Takagi-Sugeno fuzzy Cohen-Grossberg BAM neuralnetworks with discrete and distributed time-varying delays. Math. Comput. Model. 53, 151-160(2011)

18. Li, XD, Fu, XL: Globalasymptotic stability of stochastic Cohen-Grossberg-type BAM neuralnetworks with mixed delays: an LMIapproach. J. Comput. Appl. Math. 235, 3385-3394 (2011)

19. Li, YK, Chen, XR, Zhao, L: Stability and existence of a periodic solutions to delayed Cohen-Grossberg BAM neural networks with impulses on time scales. Neurocomputing 72,1621-1630 (2009)

20. Li, XD: Exponentialstability of Cohen-Grossberg-type bam neuralnetworks with time-varying delays via impulsive control. Neurocomputing 73, 525-530 (2009)

21. Xiang, HJ, Cao, JD: Exponentialstability of a periodic solution to Cohen-Grossberg-type BAM networks with time-varying delays. Neurocomputing 72,1702-1711 (2009)

22. Mathiyalagan, K, Park, JH, Sakthivel, R: Synchronization for delayed memristive BAM neuralnetworks using impulsive control with random nonlinearities. Appl. Math. Comput. 259, 967-979 (2015)

23. Mathiyalagan, K, Sakthivel, R, Anthoni, SM: Exponentialstability result for discrete-time stochastic fuzzy uncertain neuralnetworks. Phys. Lett. A 376, 901-912 (2012)

24. Sakthivel, R, Su, H, Shi, P, Sakthivel, R: Exponential filtering for discrete-time switched neuralnetworks with random delays. IEEE Trans. Cybern. 45,676-687 (2015)

25. Sakthivel, R, Vadivel, P, Mathiyalagan, K, Arunkumar, A, Sivachitra, M: Design of state estimator for bidirectional associative memory neuralnetworks with leakage delays. Inf. Sci. 296, 263-274 (2015)

26. Syed Ali, M: Stability analysis of Markovian jumping stochastic Cohen-Grossberg neuralnetworks with discrete and distributed time varying delays. Chin. Phys. B 23,060702 (2014)

27. Syed Ali, M, Balasubramaniam, P, Zhu, QX: Stability of stochastic fuzzy BAM neuralnetworks with discrete and distributed time-varying delays. Int. J. Mach. Learn. Cybern. (2015, in press). doi:10.1007/s13042-014-0320-7

28. Xu, CJ, Li, PL: Exponentialstability for fuzzy BAM cellular neuralnetworks with distributed leakage delays and impulses. Adv. Differ. Equ. 2016,276 (2016)

29. Xu, CJ, Zhang, QM, Wu, YS: Existence and exponentialstability of periodic solution to fuzzy cellular neuralnetworks with distributed delays. Int. J. Fuzzy Syst. 18,41-51 (2016)

30. Xu, CJ, Zhang, QM: Existence and stability of pseudo almost periodic solutions for shunting inhibitory cellular neural networks with neutral type delays and time-varying leakage delays. Netw. Comput. NeuralSyst. 25,168-192 (2014)

31. Li, KL, Zeng, HL: Stability in impulsive Cohen-Grossberg-type BAM neuralnetworks with time-varying delays: a generalanalysis. Math. Comput. Simul. 80, 2329-2349 (2010)

32. Arbib, M: Branins, Machines, and Mathematics. Springer, New York (1987)

33. Zhou, L, Li, W, Chen, Y: Suppressing the periodic impulsein partialdischarge detection based on chaotic control. Autom. Electr. Power Syst. 28, 90-94 (2004)

34. Nagesh, V: Automatic detection and elimination of periodic pulse shaped interference in partialdischarge measurements. IEE Proc. Sci. Meas. Technol. 141, 335-339 (1994)

35. Bai, CZ: Stability analysis of Cohen-Grossberg BAM neuralnetworks with delays and impulses. Chaos Solitons Fractals 35, 263-267 (2008)

36. Xia, YH: Impulsive effect on the delayed Cohen-Grossberg-type bam neuralnetworks. Neurocomputing 73, 2754-2764(2010)

37. Yang, T, Yang, LB: The globalstability of fuzzy cellular neuralnetworks. IEEE Trans. Circuits Syst. 43,880-883(1996)

38. Balasubramaniam, P, Syed Ali, M, Arik, S: Globalasymptotic stability of stochastic fuzzy cellular neuralnetworks with multiple time-varying delays. Expert Syst. Appl. 37, 7737-7744 (2010)

39. Li, YK, Wang, C: Existence and globalexponentialstability of equilibrium for discrete-time fuzzy BAM neuralnetworks with variable delays and impulses. Fuzzy Sets Syst. 217,62-79 (2013)

40. Li, XD, Rakkiyappan, R: Stability results forTakagi-Sugeno fuzzy uncertain BAM neuralnetworks with time delays in the leakage term. NeuralComput. Appl. 22, 203-219 (2013)

41. Balasubramaniam, P, Vembarasan, V: Robust stability of uncertain fuzzy BAM neuralnetworks of neutral-type with Markovian jumping parameters and impulses. Comput. Math. Appl. 62, 1838-1861 (2011)

42. Li, YT, Wang, JY: An analysis on the globalexponentialstability and the existence of a periodic solutions for non-autonomous hybrid BAM neuralnetworks with distributed delays and impulses. Comput. Math. Appl. 56, 2256-2267 (2008)

43. Xu, DY, Yang, ZC: Impulsive delay differentialinequality and stability of neuralnetworks. J. Math. Anal. Appl. 305, 107-120 (2005)

44. Zhou, QH, Wan, L: Impulsive effects on stability of Cohen-Grossberg-type bidirectionalassociative memory neural networks with delays. Nonlinear Anal., RealWorld Appl. 10, 2531-2540 (2009)

Submit your manuscript to a SpringerOpen journal and benefit from:

► Convenient online submission

► Rigorous peer review

► Immediate publication on acceptance

► Open access: articles freely available online

► High visibility within the field

► Retaining the copyright to your article

Submit your next manuscript at ► springeropen.com