Operational temperature effect on MR dampers in stay cable vibration control : an adaptive RL-based control strategy

Adaptive control strategy for VSG virtual inertia based on RBF neural network

This paper presents a hierarchical control scheme for virtual synchronous generators (VSGs) to mitigate power oscillations and steady-state errors induced by power command variations and grid frequency fluctuations. Unlike conventional adaptive methods that operate at the parameter level, the proposed architecture addresses the inherent structural coupling between damping and droop characteristics from a control-structure perspective. It integrates two complementary modules: a lower-level differential feedforward compensation (DFC) loop that restructures the active-power control channel to achieve structural decoupling between damping and droop, thereby eliminating steady-state deviations under frequency disturbances; and an upper-level RBF neural network that uses frequency deviation and its rate of change to dynamically regulate the virtual inertia J-the sole optimized parameter-reducing the control dimension while enhancing transient response and frequency stability. The uniform ultimate boundedness of the closed-loop system is proved via Lyapunov theory. The strategy is validated through four typical disturbance scenarios via Starsim hardware-in-the-loop experiments, along with comprehensive sensitivity analyses covering system parameters (short-circuit ratio from 1 to 5) and controller parameters (DFC coefficient and filter time constant deviations), as well as RBF initialization uncertainties. Results demonstrate that the DFC-RBF-VSG achieves zero overshoot, settling time within 0.1 s, zero steady-state error under frequency disturbances, total harmonic distortion (THD) reduced to 2.44%–2.95%, and peak inrush current limited within 300 A, consistently outperforming conventional Fixed-VSG, linear-adaptive VSG (Linear-VSG), fuzzy-based adaptive VSG (Fuzzy-VSG), and feedback-compensated VSG (FBC-VSG). The method strikes a favorable balance among dynamic responsiveness, steady-state precision, and robustness, offering an effective solution for high-penetration renewable energy systems.

Reinforcement learning-based adaptive gyroscopic control for torsional response mitigation: a proof-of-concept study

Torsional vibrations induced by seismic excitation can significantly increase structural demand, particularly in systems with geometric or stiffness irregularities. This proof-of-concept study proposes a hybrid adaptive control framework that combines passive gyroscopic damping with reinforcement learning (RL)-based adaptive torque control for torsional response mitigation. A simplified three-degree-of-freedom (3DOF) structural model is evaluated under bidirectional earthquake loading, while the control policy is trained using the Proximal Policy Optimization (PPO) algorithm to generate adaptive torque directly from measured system states without explicit system identification. Numerical simulations using recorded earthquake ground motions show that the proposed approach reduces root-mean-square torsional displacement to 0.012 ± 0.001 rad, representing a 65.7% reduction relative to the uncontrolled case and improved performance over passive gyroscopic control alone. Sensitivity analyses involving variations in mass, stiffness, and damping indicate stable controller performance across the structural parameter ranges considered. The trained controller achieved sub-millisecond inference time with minimal computational overhead, supporting its computational feasibility for future real-time implementation. The results demonstrate the potential of integrating physically grounded gyroscopic damping with data-driven adaptive control for intelligent torsional vibration mitigation.

Hybrid Mathematical Modelling and Embedded Learning for Semi-Active Friction Damping in Non-Smooth Vibration Control

Gain-Scheduled Linear Parameter-Varying Control with Aerodynamic Torque Observation for PMSG in WTCS

This study proposes a gain-scheduled linear parameter-varying (LPV) control strategy for rotor speed regulation of a permanent magnet synchronous generator (PMSG) in wind turbine conversion systems (WTCS). The nonlinear electromechanical dynamics are reformulated into a speed-dependent LPV model over the operating region. To support reference generation without direct aerodynamic torque measurement, a first-order aerodynamic torque observer is introduced. Based on a set of local linear quadratic regulators designed at selected operating points, a smooth gain-scheduling law is constructed through a neural-network-based convex interpolation mechanism. The resulting controller preserves continuity across operating conditions and improves dynamic response compared with a conventional PI-PI strategy. Closed-loop stability is discussed using a common Lyapunov theory for the local closed-loop models, and simulation studies under nominal and parameter-perturbed conditions are carried out to evaluate the proposed method. The simulation results indicate that the proposed strategy can provide satisfactory performance.

A vibration control method for chatter mitigation in milling process based on sliding mode control and reinforcement learning

Chatter is a self-excited vibration phenomenon that limits the productivity, surface quality and tool life of milling operations. In this study, an active chatter mitigation strategy is proposed by integrating sliding mode control (SMC) with reinforcement learning (RL) and an active vibration damper (AVD). A two-degree-of-freedom milling model is first formulated to describe tool vibration in the feed and normal directions, including regenerative cutting-force effects, nonlinear force components and damper-friction dynamics. A continuous-time sliding mode controller is then developed to provide robust suppression of chatter under bounded nonlinearities and modelling uncertainties. To reduce the conservative switching action and improve adaptive compensation, an actor-critic reinforcement learning component is incorporated as a bounded auxiliary control signal. The RL agent uses vibration states, sliding variables and previous control information to learn compensation forces that reduce residual chatter while penalising excessive control effort and abrupt force variations. A Lyapunov-based boundedness theorem is established to show that the sliding surface, vibration error and closed-loop milling states remain uniformly ultimately bounded when the switching gain dominates the lumped uncertainty and bounded RL compensation. Numerical simulations are conducted using cutting and structural parameters extracted from established nonlinear milling chatter studies. The proposed SMC-RL controller is compared with an uncontrolled case and a conventional PID controller. The results show that the proposed method achieves faster vibration decay and lower residual chatter in both vibration directions. Based on the mean squared error indicator, the proposed controller achieves vibration attenuation of 86.27% in the x-direction and 87.09% in the y-direction, outperforming the PID benchmark. These findings demonstrate that combining SMC robustness with RL-based adaptive compensation provides a promising framework for active chatter suppression in high-productivity milling processes.

Unified Experimentally Constrained PID/LQR Optimization for MRD-Based Semi-Active Suspension Control in Electric Vehicles

The rapid adoption of electric vehicles, together with increased battery mass and altered load distribution, is placing greater demands on ride comfort and suspension adaptability, while controller optimization may still request forces beyond the instantaneous capability of the physical semi-active actuator if experimentally supported force limits are not explicitly enforced. This study proposes a unified experimentally constrained optimization framework for a magnetorheological damper (MRD)-based semi-active suspension system using a two-degree-of-freedom quarter-car model. The damper is characterized at eleven current levels and represented by a branch-dependent lookup model that provides the zero-current baseline and instantaneous feasible force range. Proportional–integral–derivative (PID) and linear quadratic regulator (LQR) controllers are independently tuned using a genetic algorithm (GA) and particle swarm optimization (PSO) under identical vehicle dynamics, objective functions, tuning excitation, and MRD force constraints. Each candidate force demand is projected onto the experimentally derived feasible range throughout optimization. The controllers are tuned on a composite B–C–D profile and subsequently evaluated over nine road–speed scenarios. PID-PSO reduces the RMS sprung-mass acceleration by 15.91% and achieves the best acceleration performance in six cases, whereas LQR-PSO provides more balanced improvements in body motion, suspension travel, tire response, and force feasibility. The proposed framework therefore provides a more physically constrained basis for the comparative design and evaluation of MRD-based semi-active suspension control.