STABILITY ANALYSIS OF HOPFIELD NEURAL NETWORKS WITH TIME DELAY

ＩｎｔｒｏｄｕｃｔｉｏｎＩｎｒｅｃｅｎｔｙｅａｒｓｔｈｅｄｙｎａｍｉｃｂｅｈａｖｉｏｒｏｆｔｈｅｆｏｌｌｏｗｉｎｇＨｏｐｆｉｅｌｄｎｅｕｒａｌｎｅｔｗｏｒｋｓｍｏｄｅｌｗｉｔｈｔｉｍｅｄｅｌａｙｈａｖｅｂｅｅｎｉｎｖｅｓｔｉｇａｔｅｄｔｈｏｒｏｕｇｈｌｙＣｉ ｘｉ(ｔ) =-ｘｉＲｉ + ∑ｎｊ=1Ｔｉｊｆｊ(ｘｊ(ｔ-τｉｊ) ) +Ｉｉ　　 (ｉ=1 ,2 ,… ,ｎ) ,( 1 )ｗｈｅｒｅＲｉ,ＣｉａｎｄＩｉｒｅｐｒｅｓｅｎｔｒｅｓｉｓｔａｎｃｅ ,ｃａｐａｃｉｔａｎｃｅａｎｄｅｌｅｃｔｒｉｃｃｕｒｒｅｎｔ.Ｒｉ,Ｃｉａｒｅｐａ…     

Delay-independent exponential stability of stochastic Cohen–Grossberg neural networks with time-varying delays and reaction–diffusion terms

Different from the approaches used in the earlier papers, in this paper, the Halanay inequality technique, in combination with the Lyapunov method, is exploited to establish a delay-independent sufficient condition for the exponential stability of stochastic Cohen–Grossberg neural networks with time-varying delays and reaction–diffusion terms. Moreover, for the deterministic delayed Cohen–Grossberg neural networks, with or without reaction–diffusion terms, sufficient criteria for their global exponential stability are also obtained. The proposed results improve and extend those in the earlier literature and are easier to verify. An example is also given to illustrate the correctness of our results.     

ON THE ASYMPTOTIC BEHAVIOR OF HOPFIELD NEURAL NETWORK WITH PERIODIC INPUTS

ＩｎｔｒｏｄｕｃｔｉｏｎＳｉｎｃｅｔｈｅｓｐｅｃｉａｌｓｕｐｅｒｉｏｒｉｔｙｏｆａｒｔｉｆｉｃｉａｌｎｅｕｒａｌｎｅｔｗｏｒｋｓｔｅｃｈｎｏｌｏｇｙｉｎｖａｒｉｏｕｓｅｎｇｉｎｅｅｒｉｎｇｔｅｃｈｎｉｑｕｅｓｆｉｅｌｄｓ,ｓｕｃｈａｓｏｐｔｉｍｉｚａｔｉｏｎ ,ａｓｓｏｃｉａｔｉｖｅｍｅｍｏｒｉｅｓ,ｐａｔｔｅｒｎｒｅｃｏｇｎｉｔｉｏｎ ,ｓｉｇｎａｌｐｒｏｃｅｓｓｉｎｇａｎｄａｕｔｏｍａｔｉｃｃｏｎｔｒｏｌ,ｔｈｅｒｅｈａｓｂｅｅｎｉｎｃｒｅａｓｉｎｇｉｎｔｅｒｅｓｔｉｎｔｈｅｉｎｖｅｓｔｉｇａｔｉ…     

红外CT模拟在混凝土板内部缺陷探测中的应用

红外成像技术采用非接触式检测方法,对结构内部缺陷进行实时、快速大面积扫描探测,作为无损检测领域一项新的技术在土木工程中得到越来越广泛的应用。但是,这种技术的主要缺点是只能通过成像技术显示缺陷表面状况,却无法获知缺陷深度和厚度。本文以一维有缺陷混凝土板为研究对象,采用有限差分法对混凝土板进行热传导数值模拟分析,获得每一点物体表面温度差与缺陷深度及其厚度的非线性对应关系。在此基础上,采用人工神经网络算法,实现对混凝土板内部缺陷的三维重构,即红外CT模拟。本文提出的方法可以同时获得缺陷深度和厚度,并适用于任意形状的缺陷。     

矩形板结构损伤的分区域神经网络识别方法

通过引入LM优化算法，针对矩形薄板中对称结构的损伤识别问题，提 出了一种基于神经网络的分区域分步识别方法. 对于预测输出量比较多且对预测精度要求比 较高的问题，常会出现网络训练时收敛速度慢、网络预测精度低，并且当网络训练达到目标 误差时，输出的预测量中常有某个输出量的误差还很大的情况. 针对这些问题，利用选 取的组合输入参数，提出了基于神经网络的分区域识别方法. 通过对悬臂板结构的数值模拟 结果表明：提出的分区域识别方法对结构损伤的分区和预测是可行和有效的， 其预测精度要明显的高于只用单个网络的预测结果，并且预测子网络对损伤的位置和程度是 同步输出的，从而避免了传统分步识别理论中子网络过多的问题.     

铝基复合材料高速干摩擦行为的遗传神经网络预测模型

对4种SiC颗粒增强铝基复合材料在5种速度和4种压力条件下进行了销-盘摩擦磨损试验,运用遗传神经网络技术建立了铝基复合材料在高速干滑动过程中的摩擦行为预报模型,并用该模型对铝基复合材料进行预报.结果表明,蓄热能力较大的铝基复合材料在服役条件下具有较高的摩擦系数,与实际情况相一致.用遗传神经网络建立的铝基复合材料摩擦行为预测模型为服役条件下提供了简便、可靠的优选材料方法.     

Neural network optimization of material composition of a functionally graded material plate at arbitrary temperature range and temperature rise

Summary A neural network model is applied to optimization problems of material compositions for a functionally graded material plate with arbitrarily distributed and continuously varied material properties in the thickness direction. Unsteady temperature distribution is evaluated by taking into account the bounds of the number of the layers. Thermal stress components for an infinite functionally graded material plate are formulated under traction-free mechanical conditions. As a numerical example, a plate composed of zirconium oxide and titanium alloy is considered. In the optimization problem of minimizing the thermal stress distribution, the numerical calculations are carried out making use of the neural network. The optimum material composition is determined by taking into account the effect of temperature-dependence of material properties. The results obtained by neural network and ordinary nonlinear programming method are compared. Received 3 March 1998; accepted for publication 22 May 1998     

Optimal Slewing of a Flexible Spacecraft

An optimal slewing program is designed within the class of Euler rotations for a spacecraft with elastic elements. The mathematical model constructed accounts for an arbitrary number of partial modes of elastic vibrations. An optimal reorientation problem is formulated using a nontraditional performance criterion, which minimizes the dynamic overloads of the elastic elements in relative motion. An algorithm for solving the corresponding nonlinear boundary-value problem is developed and implemented in a software package of FORTRAN-programs. A neural network is generated in the space of slew parameters; it may be trained during a preflight period. Known radial basis functions are used to model the process of fast in-flight computation of the optimal reorientation program     

Stability and Synchronization of a Ring of Identical Cells with Delayed Coupling

We consider a ring of identical elements with time delayed, nearest neighbour coupling. The individual elements are modelled by a scalar delay differential equation which includes linear decay and nonlinear delayed feedback. The linear stability of the trivial solution is completely analyzed and illustrated in the parameter space of the coupling strength and the coupling delay. Conditions for global stability of the trivial solution are also given. The bifurcation and stability of nontrivial synchronous solutions from the trivial solution is analyzed using a centre manifold construction.Dedicated to Professor Shui-Nee Chow on the occasion of his 60th birthday.     

三峡库区大石板滑坡区排水系统效果评估

结合三峡库区大石板滑坡工程实例 ,以连续 36天的降雨过程为条件 ,利用已有的地下水位观测资料 ,采用人工神经网络方法预测滑坡内地下水位在排水工程实施前后的变化 ,以此作为非饱和—饱和—非衡定地下渗流计算的特定水头边界。根据计算的地下水位进行滑坡稳定性分析 ,结合稳定分析成果对排水系统工程的效果进行评价。结果表明 ,地下排水工程的实施可较大幅度地提高滑坡的稳定性。