In the aerospace context, advanced composite materials are subjected to severe broadband dynamic loads. NiTiNOL shape memory alloys (NiTi-SMA) and novel NiTi steel wire ropes (NiTi-ST) provide a smart pathway for nonlinear vibration suppression through superelastic hysteresis. This review systematically presents the progress and perspectives of nonlinear dynamics and smart vibration control in NiTi-SMA-composite coupled systems, mapping a comprehensive technical framework from material modeling, structural dynamics, reduced-order solutions, physical validation to future technological pathways. This review paper compiles advanced structural mechanics frameworks for various composite configurations. It includes nonlinear hysteretic constitutive models into multi-type structural governing equations, and assesses the theoretical benefits of Galerkin truncation and frequency-time dual-track analytical mechanisms to describe complex topologies, e.g., resonance peak deflection, multi-stability. Most importantly, it emphasizes the systematic physical capture and validation of strong nonlinear phenomena, highlighting key experimental demonstrations of the nonlinear vibration reduction potential of NiTi-SMA and NiTi-ST in composite structures. State of the art testing methodologies such as slow bidirectional sweeps and dwell control are studied for an accurate calibration of jump boundaries and multi-stable topological evolution. Cross-verification shows small discrepancies between theoretical and physical realities. In addition, the survey proposes the future development routes such as multiphysics coupling, active/semi-active controllable damping, and data-driven optimization, which can provide an important theoretical reference for the next-generation intelligent aerospace equipment.
In this paper, the practical prescribed-time optimal control (PPT-OC) problem is investigated for input-saturated strict-feedback nonlinear leader-follower multi-agent systems (LF-MASs), where the leader is driven by an external input with an unknown upper bound. A novel PPT-OC framework is developed by integrating a bounded time-varying gain, optimal backstepping design, radial basis function neural networks (RBF-NNs), and an actor-critic NN architecture. The nonautonomous Hamilton-Jacobi-Bellman (HJB) equations are formulated with the time-varying gain as an augmented variable. RBF-NNs are adopted to approximate the unknown nonlinear terms including the saturation-dependent uncertainty, on prescribed compact design domains. Under an explicit actuator-feasibility and approximation-domain assumption, Lyapunov analysis establishes PPT convergence of the LF consensus errors and boundedness of all closed-loop signals. The actor-critic weight difference is also shown to converge to zero in a fixed time without a persistent-excitation condition. Expanded simulations on an electromechanical LF-MAS evaluate different prescribed times, saturation bounds, leader-input magnitudes, a fixed-time variant, saturation compensation, accumulated cost, and parameter sensitivity.
This article investigates the problem of finite-time stability (FTS) for nonlinear systems subject to impulsive disturbances, where the impulse instants are uncertain. By fully extracting the uncertain information in impulsive disturbance instants, a relationship between uncertain impulse density, impulsive disturbance strength, and system dynamics is established. Based on this, some robustness criteria that can ensure local FTS of nonlinear systems even in the presence of impulsive disturbances with uncertain instants are proposed, and a convergence domain estimation is provided. Moreover, the actions of impulsive disturbances can be nonlinear. Finally, two numerical examples are given to illustrate the effectiveness of the theoretical results.
This paper investigates Zeno behaviors (an infinite number of discrete transitions in a finite period) for a novel class of hybrid systems involving interactions between differential dynamics and evolutionary games. Zeno behaviors frequently arise in hybrid dynamical systems and play an important role in the analysis of system dynamics. To the best of our knowledge, Zeno behaviors in this class of systems are studied for the first time. First, we conduct a fundamental analysis of system dynamics when the hybrid system model exhibits Zeno behaviors. Secondly, on Zeno behaviors we investigate several properties of the positive limit set, which can unify the theory under Zeno and non-Zeno cases for this class of hybrid systems. Thirdly, we present two classes of prolonged trajectories beyond Zeno time, and utilize them to study the interconnection of two subsystems when exhibiting Zeno behaviors. Finally, we carry out numerical examples to show the validity of our main results.
Online social networks have facilitated rapid rumor diffusion, posing serious threats to public trust and social stability. To investigate rumor propagation under media intervention, a Susceptible-Asymptomatic-Infected-Recovered model with dual media competition (SAIR-dM) is proposed on heterogeneous networks, where negative media reports promote rumor spreading while positive fact-checking reports suppress propagation through state transition dynamics. Based on mean-field theory, the dynamical equations, basic reproduction number, and the existence and stability conditions of four media competition equilibria are derived. Explicit threshold expressions reveal the regulatory mechanism of media competition on rumor outbreaks. Numerical results show that the net media effect, defined by the competition between positive and negative reports, exhibits significant bidirectional regulation on rumor prevalence: positive dominance suppresses propagation, whereas negative dominance amplifies rumor outbreaks. Moreover, dynamic media competition induces time-varying reproduction characteristics, highlighting the necessity of full-cycle intervention strategies, especially rapid early-stage authoritative responses. This work reveals the mechanism of dual-media competition in heterogeneous rumor dynamics and provides a theoretical basis for optimizing media intervention and online public opinion governance.
For vector ultrashort pulses in inhomogeneous optical fibers, the variable-coefficient coupled Hirota system provides an integrable model for higher-order dynamics. Within this model, the objective is to distinguish the respective roles of accumulated coefficient functions, spectral multiplicity, polarization parameters, and background amplitudes in higher-order degenerate wave dynamics and mixed interactions. The generalized Darboux transformation yields determinant representations for arbitrary-order positons, breather-positons, and mixed interaction solutions. Analysis of positons on the zero background shows that accumulated coefficient functions govern trajectory modulation, whereas spectral multiplicity determines order-dependent structure. For interactions on the zero background, component-wise asymptotics classify positon-soliton collisions as elastic or shape-changing according to the polarization configuration. Quantitative diagnostics of a positon-molecule collision show that velocity locking is preserved and distinguish collision-induced structural change from polarization-dependent redistribution of local component energies. On plane wave backgrounds, analysis of characteristic roots and parameter dependence shows that background amplitudes and polarization parameters govern the branch structure and component intensity patterns of breather-positons. Accumulated coefficient functions modulate branch trajectories, whereas interactions between rogue waves and breather-positons reshape local intensity profiles. These results may inform dispersion and polarization management for vector pulse shaping in inhomogeneous optical media.
To clarify nonlinear instability in the coaxial orthogonal face gear power-split and confluence transmission system, this study develops a 21-degree-of-freedom multi-backlash bending-torsion coupled dynamic model based on the lumped parameter method. The model incorporates piecewise backlash nonlinearity, time-varying mesh stiffness and tooth surface friction, and is solved using a nondimensional variable-step fourth-order Runge–Kutta scheme. Combined with bifurcation diagrams, Lyapunov exponents and other analytical methods, the influence laws of rotational speed, tooth backlash, damping coefficient and friction coefficient on the bifurcation evolution and chaos of the transmission system were revealed. The results show that increasing backlash from 0 to 50 μm expands the sensitive resonance region by about 37
Studying the collaborative effects among different functional brain regions helps reveal the working mechanisms of electrical activity in the nervous system. However, existing studies on memristive Hopfield neural networks mainly focus on single-network structures or general coupling forms, while the information exchange mechanisms between different subnetworks and their effects on controllable multiscroll dynamics remain insufficiently explored. Therefore, this paper proposes a new multi-segment linear memristor to simulate a neural synapse and constructs a dual neural network heterogeneous coupled complex multiscroll system (DNNHCMS) by connecting two asymmetric sub-neural networks through synaptic coupling. Numerical results reveal that the coupling strength between the subnetworks determines the dynamical characteristics of the DNNHCMS. Within specific parameter ranges, the system can generate multiple double-scroll attractors, the number of which can be controlled by adjusting the segment parameter of the memristor. Meanwhile, the system exhibits significant extreme multistability and flexible amplitude control, with circuit simulations and hardware implementations further validating the reliability of the numerical analysis. Finally, based on the controllable multiscroll and extreme multistability characteristics of the DNNHCMS, this paper constructs 10 plaintext-related dynamic S-boxes using the highly complex chaotic sequences generated by the system, and proposes a new image encryption scheme. In each round of permutation and diffusion, chaotic sequences are used to dynamically select S-boxes for substitution operations, thereby reducing redundant computations and significantly improving the security and efficiency of the encryption process. Security analysis demonstrates that the proposed scheme possesses robust security performance.
To investigate dynamic contact loads in high-speed train axle-box bearings under wheel polygonal excitation and evaluate the mitigation effect of rail grinding, a sequential co-simulation framework integrating vehicle-track coupled dynamics and a refined axle-box bearing dynamic model is developed. Field-measured wheel polygonal wear data and rail profiles before and after grinding are incorporated, while axle-box boundary-load histories are transferred unidirectionally from the vehicle-track model to the bearing model, extending the analysis from macroscopic wheel-rail responses to internal roller-raceway contact loads. Under 20th-order excitation, the RMS contact loads of the inner and outer raceways increase by 39.8 and 26.2
Existing research on patch structured models primarily focuses on fundamental theoretical aspects such as well-posedness and equilibrium stability, while the exploration of their bifurcation dynamics remains an advanced frontier with both theoretical appeal and profound challenges. In this article, we consider the effect of delay in a patch structured system arising in the houseflies model contemplating nonconstant fertility and develop basic theory of patch structured system with delay. Specifically, we yield the existence of Hopf bifurcation and consider the effect of the dispersal rate on the Hopf bifurcation. Particularly, if each patch is beneficial to the houseflies, when the dispersal rate approaches to zero, the limit of Hopf bifurcation value is the minimum of local Hopf bifurcation values over all patches; when the dispersal rate approaches to infinity, the Hopf bifurcation value tends to that of the average model. To validate the theoretical findings and illustrate how varying fertility rates and patch connectivity influence oscillatory behavior, we give some numerical simulations.
This paper proposes a safety-critical hierarchical target encirclement control strategy for multiple autonomous underwater vehicles (multi-AUVs) with bearing-only measurements. The proposed control scheme consists of three layers: the estimation layer, the guidance layer, and the control layer. In the estimation layer, target position and velocity estimators are designed for each pursuing AUV using the relative bearing angle. In the guidance layer, leveraging estimated information, a nominal guidance law is proposed based on the line-of-sight (LOS) guidance method for the pursuing AUVs to encircle the target. Then, based on the safety constraints derived from the control barrier function (CBF), a distributed quadratic programming problem is formulated to compute the optimal guidance law that satisfies the safety constraints. In the control layer, an event-triggered predefined-time integral sliding mode controller (EPISMC) combined with an adaptive predefined-time disturbance observer (APDO) and anti-windup compensation is developed, which reduces communication frequency, mitigates saturation, and compensates for environmental disturbances. The overall stability of the closed-loop system is proved using Lyapunov theory. Finally, numerical simulations and semi-physical experiments are conducted to demonstrate the effectiveness of the proposed control scheme.
This paper investigates the problem of obstacle avoidance planning and control for unmanned ground vehicles (UGV) based on event-triggered model predictive control (EMPC). A unified framework comprising planning and control is established and the control strategy is designed to realize obstacle avoidance planning, which enables the vehicle to automatically calculate obstacle avoidance path in real-time when detecting obstacles. In addition, in order to reduce the computational complexity of model predictive control, an event-triggered mechanism is introduced, which can effectively reduce the computational burden of model predictive control. The stability of the system and the feasibility of the solution are demonstrated by rigorous derivations. The reference trajectory is generated using a point cloud dataset that matches the real map, and the effectiveness of the algorithm is verified by simulation of obstacle avoidance and trajectory planning.
Electric vehicles exhibit variable speed motion when encountering road obstacles. This inevitably makes the electric powertrain in a non-inertial system. Most existing electric powertrain models overlook the non-inertial effects induced by vehicle motion. This study proposes a modeling method for the electromechanical rigid-flexible coupling of the electric powertrain in a non-inertial system. Additional inertial forces are derived from vehicle motion parameters. A vibration test is launched to verify model effectiveness. Speed bumps, potholes, and randomly uneven roads serve as road excitations to analyze the vibration response and load characteristics of the powertrain. The results show that the equilibrium positions of components are significantly offset when encountering speed bumps and potholes. The position offset leads to fluctuations in bearing force. Under conditions of a randomly uneven road, random components are introduced into the bearing vibration signals and motor electrical signals, resulting in a increase in amplitude fluctuations. Furthermore, the structural parameters of speed bumps, potholes, and road profiles greatly influence the bearing force and vibration acceleration. The research results provide theoretical support for the structural design and life prediction of electric powertrains.
The low-speed, high-power permanent magnet semi-direct-drive sprocket system (PM-SDDS) exhibits electromechanical coupling characteristics, and its vibration analysis is complex due to internal and external excitations and varying operating conditions. This study integrates the effects of controller parameters, inverter and flux linkage nonlinearities, gear meshing nonlinearities, and the sprocket polygonal effect (SPE), establishing a state-space-based torsional dynamics model applicable to various operating conditions. The impact of controller parameters and load variation on system modal characteristics is analyzed, along with vibration responses under typical conditions. The results show that changes in the speed-loop proportional gain Kpw affect the first-order natural frequency fn,1 through the electromagnetic (EM) effect, while load variations impact fn,1 by altering the system's mechanical characteristics. The SPE introduces low-frequency excitations into the system, with its intensity varying under changes in lumped mass and motor speed. Transient free vibrations at fn,1 are observed both during heavy-load startup and upon sudden load changes. The second harmonic of the meshing frequency 2fm is the dominant frequency for exciting second- and third-order torsional resonance, with modulation sidebands |fe±2fm| appearing in the stator current. Under sudden load changes, sideband clusters modulated by fn,1 appear near the gear meshing harmonics, accompanied by |fe±fn,1| modulation sidebands in the current spectrum. This work provides a theoretical guidance for torsional resonance monitoring and dynamic characteristics analysis of the PM-SDDS.
The fractional chaotic system exhibits long-memory effects, weak singularity and high sensitivity, which make accurate long-time prediction particularly challenging for machine learning methods. A fractional difference equation is used to address weakly singular problems of fractional differential equations. Then the Kolmogorov–Arnold network is employed and trained pointwise. The proposed method is tested on three fractional-order systems with stable, periodic, and chaotic dynamical behaviors. The machine learning method achieves higher accuracy in longer time domains than the existing methods. The fractional Lorenz system is used as a representative chaotic benchmark. These results indicate that the proposed method can provide learned analytical representations for fractional chaotic systems. Its main novelty lies in combining a fractional difference formulation with Kolmogorov–Arnold network training to control the accumulation of long-memory errors.
This article proposes a reinforcement learning (RL) framework for the online self-tuning of active disturbance rejection control (ADRC) for permanent magnet synchronous motor (PMSM) drives. The proposed framework is built upon a structured two-stage curriculum learning pipeline. A central component of this pipeline is the physics-guided generation for domain randomization (PIGEON) method, which constructs a challenging yet physically plausible training curriculum from motor priors, rather than relying on blind or unstructured parameter sampling. To support policy learning across different operating phases, a self-calibrating context-adaptive reward activation (CARA) mechanism is introduced, providing phase-aware guidance for transient response, disturbance rejection, and steady-state tracking. The framework is validated through both high-fidelity simulations and hardware experiments. Simulation results demonstrate improved robustness under broad electrical and mechanical parameter variations, speed-switching profiles, and load disturbances. Structured hardware experiments further show that the controller trained in simulation can be deployed to the target platform without additional on-hardware retraining, and outperforms controllers based on nominal-only training, offline fixed-gain robust tuning, and conventional domain randomization.
Addressing the current scarcity of research on the nonlinear dynamics of functionally graded porous (FGP) circular and annular plates, particularly regarding amplitude-frequency response characteristics, this paper develops a novel theoretical model to capture the nonlinear dynamic behavior of such plates featuring thickness-dependent pore distribution supported by nonlinear elastic foundation, and investigates the regulatory laws governing the resulting nonlinear dynamic responses of geometry-, porosity- and foundation-related parameters. Building upon the large deformation assumption and first-order shear deformation formulation, nonlinear dynamic equations are derived using energy methods, and a hybrid solution framework integrating the incremental harmonic balance method and the spectral element method is established. The validity and versatility of the proposed approach are rigorously proven via comparisons with finite element simulations, benchmark literature, and numerical solutions for both linear and nonlinear cases. Parametric studies reveal strong correlations between the amplitude-frequency and amplitude-time responses and the aforementioned design parameters, particularly with foundation parameters exerting complex regulatory effects. This study provides an efficient analytical pathway for examining the nonlinear dynamics of FGP circular/annular plates, and the identified parameter-response relationships offer preliminary recommendations for the design and performance tuning of such structural elements.
The nonlinear energy harvesting system enhanced by a resonance mechanism is proposed, which consists of a piezoelectric cantilever beam supported by a nonlinear boundary oscillator. The nonlinear boundary oscillator comprises three springs and a lumped mass, which introduces tunable nonlinear stiffness and modifies the overall dynamic characteristics. The dynamic equations for the electromechanically coupled system are derived using the extended Hamilton principle. The direct multiple-scale method is employed to derive the modulation equations governing the slow dynamics of amplitude and phase. The perturbation solutions are validated against numerical simulations via the Runge–Kutta method. At primary resonance, the coupled system exhibits nonlinear behaviors, including hardening spring behavior, jump phenomena, and multiple solutions. Parametric studies indicate that the mass ratio between the boundary oscillator and the cantilever beam significantly influences the frequency response, enabling enhancement of the electrical output and adjustment of the effective harvesting bandwidth. The nonlinear stiffness parameter primarily affects the extent of hardening nonlinearity, while the load resistance largely determines the electrical output characteristics. Compared with a conventional single piezoelectric cantilever beam, the coupled system yields larger vibration amplitudes and higher electrical output, demonstrating its potential for efficient broadband vibration energy harvesting.
Astrocytes are increasingly recognized as active regulators of neural dynamics and computations, rather than merely passive support cells. Through multifaceted interactions with neurons, synapses and neuromodulatory systems, astrocytes profoundly modulate neural dynamics and computations that underlie a broad range of physiological functions, most notably adaptive behaviors related to cognitive processing and motor control. Despite accumulating evidence of astrocytic engagement in these processes, the correspondence between experimentally validated astrocytic biophysical mechanisms and computational rules governing these processes remains incompletely characterized. In this review, we synthesize experimental observations across molecular, cellular and population levels. We center our analysis on three core functional domains, namely working memory, decision-making and motor control, and systematically elucidate how astrocytic biophysical mechanisms can be mapped onto fundamental computational operations. These operations facilitate the context-dependent optimization of working memory flexibility, decision-making accuracy and motor precision. This review can advance the mechanistic understanding of adaptive neural computations across cognitive and motor processes.
Safe and efficient coordination of multiple articulated wheel loaders in confined and obstacle-rich environments remains a critical challenge for autonomous mining operations. This paper proposes a lightweight closed-form control strategy for collision-free navigation of multiple articulated wheel loaders. The variable structure control framework drives the loader state toward the switching surface, stabilizing the collision risk function’s rate of change and mitigating unnecessary conservative avoidance behaviors. Additionally, dual avoidance regions enable precise spatial modeling, enhanced maneuverability in narrow spaces, and accommodated articulated motion characteristics, thereby providing enhanced operational flexibility in complex environments. The integration of virtual leaders facilitates the adaptation of articulated loaders to dense obstacle scenarios. A rigorous Lyapunov-based stability analysis is conducted to demonstrate that the system states maintain bounded tracking errors relative to the reference trajectories while guaranteeing collision avoidance. The efficacy and robustness of the proposed algorithm are validated through extensive simulations. The proposed method provides a computationally efficient and practically feasible solution for the safe and coordinated operation of multiple articulated wheel loaders in autonomous mining environments.
