A new result on global exponential robust stability of neural networks with time-varying delays

In this paper, the global exponential robust stability of neural networks with ume-varying delays is investigated. By using nonnegative matrix theory and the Halanay inequality, a new sufficient condition for global exponential robust stability is presented. It is shown that the obtained result is different from or improves some existing ones reported in the literatures. Finally, some numerical examples and a simulation are given to show the effectiveness of the obtained result.

ROBUST GLOBAL EXPONENTIAL STABILITY OF UNCERTAIN IMPULSIVE SYSTEMS

By using the quasi-Lyapunov function, some sufficient conditions of global exponential stability for impulsive systems are established, which is the basis for the following discussion. Then, by employing Riccati inequality and Hamilton-Jacobi inequality approach, some sufficient conditions of robust exponential stability for uncertain linear/nonlinear impulsive systems are derived, respectively. Finally, some examples are given to illustrate the applications of the theory.

Globally robust nonlinear PID controllers for robot manipulators with an uncertain Jacobian matrix

Based on a continuous piecewise-differentiable increasing functions vector, a class of robust nonlinear PID (RN-PID) controllers is proposed for setpoint control with uncertain Jacobian matrix. Globally asymptotic stability is guaranteed and only position and joint velocity measurements are required. And stability problem arising from integral action and integrator windup, are consequently resolved. Furthermore, RN-PID controllers can be of effective alternative for anti-integrator-wind-up, the control performance would not be very bad in the presence of rough parameter tuning.

Towards robust neural networks via a global and monotonically decreasing robustness training strategy

Robustness of deep neural networks(DNNs)has caused great concerns in the academic and industrial communities,especially in safety-critical domains.Instead of verifying whether the robustness property holds or not in certain neural networks,this paper focuses on training robust neural networks with respect to given perturbations.State-of-the-art training methods,interval bound propagation(IBP)and CROWN-IBP,perform well with respect to small perturbations,but their performance declines significantly in large perturbation cases,which is termed“drawdown risk”in this paper.Specifically,drawdown risk refers to the phenomenon that IBPfamily training methods cannot provide expected robust neural networks in larger perturbation cases,as in smaller perturbation cases.To alleviate the unexpected drawdown risk,we propose a global and monotonically decreasing robustness training strategy that takes multiple perturbations into account during each training epoch(global robustness training),and the corresponding robustness losses are combined with monotonically decreasing weights(monotonically decreasing robustness training).With experimental demonstrations,our presented strategy maintains performance on small perturbations and the drawdown risk on large perturbations is alleviated to a great extent.It is also noteworthy that our training method achieves higher model accuracy than the original training methods,which means that our presented training strategy gives more balanced consideration to robustness and accuracy.

Improve Robustness and Accuracy of Deep Neural Network with L_(2,∞) Normalization

In this paper,the L_(2,∞)normalization of the weight matrices is used to enhance the robustness and accuracy of the deep neural network(DNN)with Relu as activation functions.It is shown that the L_(2,∞)normalization leads to large dihedral angles between two adjacent faces of the DNN function graph and hence smoother DNN functions,which reduces over-fitting of the DNN.A global measure is proposed for the robustness of a classification DNN,which is the average radius of the maximal robust spheres with the training samples as centers.A lower bound for the robustness measure in terms of the L_(2,∞)norm is given.Finally,an upper bound for the Rademacher complexity of DNNs with L_(2,∞)normalization is given.An algorithm is given to train DNNs with the L_(2,∞)normalization and numerical experimental results are used to show that the L_(2,∞)normalization is effective in terms of improving the robustness and accuracy.

Robust fuzzy control of Takagi-Sugeno fuzzy neural networks with discontinuous activation functions and time delays

The problem of global robust asymptotical stability for a class of Takagi-Sugeno fuzzy neural networks（TSFNN） with discontinuous activation functions and time delays is investigated by using Lyapunov stability theory.Based on linear matrix inequalities（LMIs）,we originally propose robust fuzzy control to guarantee the global robust asymptotical stability of TSFNNs.Compared with the existing literature,this paper removes the assumptions on the neuron activations such as Lipschitz conditions,bounded,monotonic increasing property or the right-limit value is bigger than the left one at the discontinuous point.Thus,the results are more general and wider.Finally,two numerical examples are given to show the effectiveness of the proposed stability results.

ROBUST ANALYSIS AND GLOBAL MINIMIZATION OF A CLASS OF DISCONTINUOUS FUNCTIONS (Ⅱ)

In this Paper we continue to investigate global minimization problems. An integral approach is applied to treat a global minimization problem of a discontinuous function. With the help ofthe theory of measure (Q-measure) and integration, optimality conditions of a robust function over arobust set are derived. Algorithms and their implementations for finding global minima are proposed.Numerical tests and applications show that the algorithms are effective.

Robust Global Stabilization of Spacecraft Rendezvous System via Gain Scheduling

The problem of robust global stabilization of a spacecraft circular orbit rendezvous system with input saturation and inputadditive uncertainties is studied in this paper. The relative models with saturation nonlinearity are established based on ClohesseyWiltshire equation. Considering the advantages of the recently developed parametric Lyapunov equation-based low gain feedback design method and an existing high gain scheduling technique, a new robust gain scheduling controller is proposed to solve the robust global stabilization problem. To apply the proposed gain scheduling approaches, only a scalar nonlinear equation is required to be solved.Different from the controller design, simulations have been carried out directly on the nonlinear model of the spacecraft rendezvous operation instead of a linearized one. The effectiveness of the proposed approach is shown.

GLOBAL ROBUST STABILITY OF INTERVAL HOPFIELD NEURAL NETWORKS WITH CONTINUOUSLY DISTRIBUTED DELAYS

In this paper, we implement topological degree theory and Lyapunov-functional methods to obtain the existence and uniqueness of the equilibrium point and its global robust stability for interval Hopfield neural networks with continuously distributed delays. Moreover, the methods used in judging the robust stability are proven practical and easily verifiable.

A Precise and Robust Control Strategy for Rigid Spacecraft Eigenaxis Rotation

Eigenaxis rotation is generally regarded as a near-minimum time strategy for rapid attitude maneuver due to its constitution of the shortest angular path between two orientations. In this paper, the robust control problem of rigid spacecraft eigenaxis rotation is investigated via time-varying sliding mode control （TVSMC） technique. Both external disturbance and parameter variation are taken into account. Major features of this robust eigenaxis rotation strategy are first demonstrated by a TVSMC algorithm. Global sliding phase is proved as well as the closed-loop system stability. Additionally, the necessary condition for eigenaxis rotation is provided. Subsequently, to suppress the global chattering and improve the control accuracy, a disturbance observer-based time-varying sliding mode control （DOTVSMC） algorithm is presented, where the boundary layer approach is used to soften the chattering and a disturbance observer is designed to attenuate undesired effect. The spacecraft attitude is represented by modified Rodrigues parameter （MRP） for the non-redundancy. Finally, a numerical simulation is employed to illustrate the effectiveness of the proposed strategy, where the pulse-width pulse-frequency （PWPF） technique is utilized to modulate the on-off thrusters.

Relaxed delay-dependent exponential stability condition for a class of neural networks with polytopic uncertainties and distributed delays

The global robust exponential stability of a class of neural networks with polytopic uncertainties and distributed delays is investigated in this paper.Parameter-dependent Lypaunov-Krasovskii functionals and free-weighting matrices are employed to obtain sufficient condition that guarantee the robust global exponential stability of the equilibrium point of the considered neural networks.The derived sufficient condition is proposed in terms of a set of relaxed linear matrix inequalities (LMIs),which can be checked easily by recently developed algorithms solving LMIs.A numerical example is given to demonstrate the effectiveness of the proposed criteria.

Stability Analysis of Robust Adaptive Control of Continuous-Time Systems

A robust indirect adaptive pole placement control scheme for continuous time systems with bounded disturbances and unmodelled beamics is proposed in this poper which is an extension of the fundamental work developed by the second author It is shown thai a globally asymptotically stability is established for closed loop system.

Robust stability of mixed Cohen–Grossberg neural networks with discontinuous activation functions

Purpose–The purpose of this paper is to develop a method for the existence,uniqueness and globally robust stability of the equilibrium point for Cohen–Grossberg neural networks with time-varying delays,continuous distributed delays and a kind of discontinuous activation functions.Design/methodology/approach–Basedonthe Leray–Schauderalternativetheoremand chainrule,by using a novel integral inequality dealing with monotone non-decreasing function,the authors obtain a delay-dependent sufficient condition with less conservativeness for robust stability of considered neural networks.Findings–Itturns out thattheauthors’delay-dependent sufficientcondition canbeformed intermsof linear matrix inequalities conditions.Two examples show the effectiveness of the obtained results.Originality/value–The novelty of the proposed approach lies in dealing with a new kind of discontinuous activation functions by using the Leray–Schauder alternative theorem,chain rule and a novel integral inequality on monotone non-decreasing function.