Presents a recurrent neural network for solving the Sylvester equation with time-varying coefficient matrices. The implicit dynamics is deliberately developed in way that its trajectory guaranteed to converge exponentially solution of given equation. Theoretical results convergence and sensitivity analysis are presented show desirable properties network. Simulation matrix inversion online nonlinear output regulation via pole assignment ball beam system inverted pendulum on cart also included...

In this paper, the existence and uniqueness of equilibrium point its global asymptotic stability are discussed for a general class recurrent neural networks with time-varying delays Lipschitz continuous activation functions. The network model considered includes delayed Hopfield networks, bidirectional associative memory cellular as special cases. Several new sufficient conditions ascertaining existence, uniqueness, such obtained by using theory topological degree properties nonsingular...

In this paper, two related problems, global asymptotic stability (GAS) and robust (GRS) of neural networks with time delays, are studied. First, GAS delayed is discussed based on Lyapunov method linear matrix inequality. New criteria given to ascertain the networks. designs applications networks, it necessary consider deviation effects bounded perturbations network parameters. case, a must be formulated as interval model. Several sufficient conditions derived for existence, uniqueness, GRS...

Autonomous surface vehicles (ASVs) are marine vessels capable of performing various operations without a crew in variety cluttered and hostile water/ocean environments. For complex missions, there increasing needs for deploying fleet ASVs instead single one to complete difficult tasks. Cooperative with offer great advantages enhanced capability efficacy. Despite application potentials, coordinated motion control pose challenges due the multiplicity ASVs, complexity intravehicle interactions...

Cyber-physical systems (CPSs) empower the integration of physical processes and cyber infrastructure with aid ubiquitous computation resources communication capabilities. CPSs have permeated modern society found extensive applications in a wide variety areas, including energy, transportation, advanced manufacturing, medical health. The security against cyberattacks has been regarded as long-standing concern. However, suffer from extendable vulnerabilities that are beyond classical networked...

Techniques for automatically designing deep neural network architectures such as reinforcement learning based approaches have recently shown promising results. However, their success is on vast computational resources (e.g. hundreds of GPUs), making them difficult to be widely used. A noticeable limitation that they still design and train each from scratch during the exploration architecture space, which highly inefficient. In this paper, we propose a new framework toward efficient search by...

This technical note presents a second-order multi-agent network for distributed optimization with sum of convex objective functions subject to bound constraints. In the network, agents connect each others locally as an undirected graph and know only their own objectives The is proved be able reach consensus optimal solution under mild assumptions. Moreover, converted convergence dynamical system, which using Lyapunov method. Compared existing networks optimization, herein capable solving...

This brief is concerned with the distributed maneuvering of multiple autonomous surface vehicles guided by a virtual leader moving along parameterized path. In guidance loop, law developed incorporating constant bearing strategy into path-maneuvering design such that prescribed formation pattern can be reached. To optimize signal under velocity constraint as well minimize control torque during transient phase, an optimization-based command governor employed to generate optimal for vehicle...

Since the last decade, several complex-valued neural networks have been developed and applied in various research areas. As an extension of real-valued recurrent networks, use states, connection weights, or activation functions with much more complicated properties than ones. This paper presents sufficient conditions derived to ascertain existence unique equilibrium, global asymptotic stability, exponential stability delayed two classes functions. Simulation results three numerical examples...

This paper presents a design method for output-feedback path-following control of under-actuated autonomous underwater vehicles moving in vertical plane without using surge, heave, and pitch velocities. Specifically, an extended state observer (ESO) is developed to recover the unmeasured velocities as well estimate total uncertainty induced by internal model external disturbance. At kinematic level, commanded guidance law based on line-of-sight scheme observed To optimize signals,...

In this paper, a robust recurrent neural network is presented in Bayesian framework based on echo state mechanisms. Since the new model capable of handling outliers training data set, it termed as (RESN). The RESN inherits basic idea ESN learning framework, but replaces commonly used Gaussian distribution with Laplace one, which more to outliers, likelihood function output. Moreover, facilitated by employing bound optimization algorithm, which, proper surrogate derived and approximated while...

In this paper, a neurodynamics-based output feedback scheme is proposed for distributed containment maneuvering of marine vessels guided by multiple parameterized paths without using velocity measurements. Each vessel subject to internal model uncertainties and external disturbances induced wind, waves, ocean currents. order recover unmeasured information as well identify unknown dynamics, an echo state network (ESN) based observer recorded input-output data each vessel. Based on the...

Multivariable time series prediction has been widely studied in power energy, aerology, meteorology, finance, transportation, etc. Traditional modeling methods have complex patterns and are inefficient to capture long-term multivariate dependencies of data for desired forecasting accuracy. To address such concerns, various deep learning models based on Recurrent Neural Network (RNN) Convolutional (CNN) proposed. improve the accuracy minimize dependence aperiodic data, this article, Beijing...

This technical note presents a continuous-time multi-agent system for distributed optimization with an additive objective function composed of individual functions subject to bound, equality, and inequality constraints. Each is assumed be convex in the region defined by its local bound constraints only without need globally convex. All agents communicate using proportional-integral protocol their output information instead state reduce communication bandwidth. It proved that all any initial...

In this paper, a design method is presented for path-following control of underactuated autonomous underwater vehicles subject to velocity and input constraints, as well internal external disturbances. the guidance loop, kinematic law desired surge speed pitch rate derived based on backstepping technique line-of-sight principle. an extended state observer developed estimate composed unknown dynamics Then, disturbance rejection constructed using observer. To bridge loop reference governor...

In this paper, we present a neurodynamic approach to model predictive control (MPC) of unknown nonlinear dynamical systems based on two recurrent neural networks (RNNs). The echo state network (ESN) and simplified dual (SDN) are adopted for system identification dynamic optimization, respectively. First, the is identified ESN with input-output training testing samples. Then, resulting nonconvex optimization problem associated MPC decomposed via Taylor expansion. To estimate higher order term...

In this paper, distributed optimization is addressed based on a continuous-time multiagent system in the presence of time-varying communication delays. First, relationship between optimal solutions and equilibrium points with time delay revealed. Next, delay-dependent delay-independent sufficient conditions form linear matrix inequality are derived for ascertaining convergence to solutions, cases slow-varying fast-varying delay. Furthermore, set also obtained delay-free case. addition,...

This paper presents a one-layer projection neural network for solving nonsmooth optimization problems with generalized convex objective functions and subject to linear equalities bound constraints. The proposed is designed based on two operators: equality constraints, function in the problem can be any which not restricted but required (pseudoconvex) set defined by Compared existing recurrent networks optimization, model does have design parameter, more convenient implementation. It proved...

This paper presents a collective neurodynamic approach with multiple interconnected recurrent neural networks (RNNs) for distributed constrained optimization. The objective function of the optimization problems to be solved is sum local convex functions, which may nonsmooth. Subject its constraints, each minimized individually by using an RNN, consensus among others. In contrast existing continuous-time methods, proposed capable solving more general problems. Simulation results on three...

This paper presents a predictor-based neural dynamic surface control (PNDSC) design method for class of uncertain nonlinear systems in strict-feedback form. In contrast to existing NDSC approaches where the tracking errors are commonly used update network weights, predictor is proposed every subsystem, and prediction employed adaptation laws. The scheme enables smooth fast identification system dynamics without incurring high-frequency oscillations, which unavoidable using classical methods....