This paper presents a controlled and reproducible comparison between two duty-cycle computation routes for space vector pulse width modulation (SVPWM) applied to three-phase two-level inverters: the classical trigonometric formulation and the algebraic g, h approach. Both methods were implemented in MATLAB/Simulink under identical operating conditions, including the same reference signals, switching sequence, sampling configuration, and spectral evaluation window, enabling isolation of the computational stage effects. The results show that both approaches produce identical duty ratios within numerical precision and exhibit equivalent spectral performance, with individual harmonic distortion indices at the post-filter measurement point remaining practically unchanged, typically around 3.10 ≈ 14.16
Traveling-wave-based (TW-based) protection devices for fault detection and location on transmission lines are becoming increasingly prevalent in modern power systems due to their superior accuracy and fast operating performance. This study investigates modern fault detection and location techniques based on TWs using a commercial Intelligent Electronic Device (IED). The research focuses on a specific type of contingency, namely evolving faults. During single-phase faults with a 0^∘ fault inception angle, the amplitude of the generated TWs can be minimal or even imperceptible, hindering their detection and the accurate localization of the fault point by TW-based methods. However, if the fault evolves into a three-phase or bi-phase short circuit, the TW protection algorithm may successfully detect and locate the fault. Therefore, the objective of this research is to analyze this specific fault condition and evaluate how TW-based protection methods respond. To conduct the study, a real-world 50 Hz transmission line from the Paraguayan interconnected power system was modeled on the RTDS platform, and two SEL-T401L IEDs were connected at its opposite terminals in a Hardware-in-the-Loop (HIL) configuration. A total of 945 tests were performed, varying the location, the time of evolution and the fault resistance. The results show that the SEL-T401L may fail to detect or locate faults using traveling-wave-based principles when the fault evolution time exceeds 35
In this work, we propose a Quality of Experience (QoE)-driven framework for the automatic tuning of MPC schemes acting as energy management systems for renewable microgrids. Building upon the receding-horizon auto-tuning paradigm, we assess how normalised QoE metrics can be utilised to dynamically adapt the predictive control law during the implementation. The practical interest of this modular framework is that a local microgrid operator can adapt energy and demand coordination with respect to high-level goals, without requiring any technical knowledge related to MPC synthesis or manual matrix calibration. High-fidelity simulation results of two different microgrids with green-hydrogen energy are provided to illustrate the advantages of the proposed technique.
The problem of quasi-data-driven dynamic output feedback control design is addressed in this work by considering input-state-output data from linear time-invariant systems. The new design is achieved directly from the collected data using state data from the open-loop offline experiments. This distinguishes this work from other approaches that employ equality constraints, which makes the design more conservative. Then, the result is extended for the case where the state data are not available, presenting an approach for the input–output problem. Moreover, this work also addresses the problem of piecewise constant reference tracking by using a reference input scheme. The design of the scheme is done by using offline data from the open-loop system, providing a way to render the steady tracking error to zero. The effectiveness of the proposed methodology is illustrated in a case study of a real-world setup of a twin-rotor aerodynamic system, where both input-state-output and input–output problems are implemented and compared.
This paper proposes an integral-action sliding mode control (IA-SMC) strategy for the grid-side converter (GSC) of a doubly fed induction generator (DFIG)-based wind energy system. The studied system integrates the turbine, gearbox, DFIG, DC bus, a rotor-side converter (RSC) and a grid-side converter (GSC). The proposed controller addresses the nonlinear and coupled dynamics of the system, ensuring robust DC-link voltage regulation and accurate reactive power control while mitigating chattering effects. The performance of the IA-SMC is evaluated and compared with conventional PI and classic sliding mode control (SMC) under wind variations, grid-voltage disturbances, and DC-link capacitance changes. The results show that IA-SMC significantly outperforms both controllers. Compared to PI, the DC-link voltage overshoot is reduced by up to 91% under wind disturbances and 83% under grid-voltage variations, with a settling time reduction of about 32% and a voltage drop reduction of 66% during capacitance variations. Compared to the classic SMC, the overshoot is reduced by up to 50% under wind disturbances and 67% under grid-voltage variations, while the voltage drop is reduced by about 50% . Moreover, IA-SMC reduces current overshoot by about 69% and attenuates reactive power fluctuations by up to 96% compared to PI and 56% compared to SMC. These results confirm the effectiveness and robustness of the proposed approach.
Abstract In this paper, we devise sampled-data state-feedback controllers for uncertain linear systems that ensure robust stability and $$\mathcal {H}_\infty $$ H ∞ performance. We model the closed-loop system as a hybrid linear dynamic system with polytopic or interval uncertainties, capturing continuous and discrete-time behaviours in a unified framework. This approach avoids the proliferation of uncertain parameters typically encountered in discretization methods. Using this hybrid model, we derive computationally tractable control design conditions that guarantee an upper bound for the associated closed-loop $$\mathcal {H}_\infty $$ H ∞ norm. Our formulation enables designers to select between interval methods that offer computational efficiency for systems with numerous independent uncertainties, or polytopic methods that provide less conservative results at higher computational cost. Numerical examples with detailed simulation verification demonstrate the effectiveness of our proposed techniques and validate the theoretical performance guarantees across the entire uncertainty space.
Four-wheel-independent-steering (4WIS) vehicles, owing to the independent controllability of all four wheels, significantly enhance vehicle maneuverability in confined spaces and improve stability under high-speed driving conditions. However, existing vehicle dynamic models for path tracking primarily focus on suppressing the sideslip angle at the center of gravity to follow a given trajectory. This modeling paradigm, to some extent, weakens the vehicle’s capability to perceive and exploit lateral motion, and may induce pronounced yaw motion in high-speed, large-curvature cornering scenarios. As a result, the vehicle is more likely to enter highly nonlinear operating regions, thereby increasing the risk of instability. To address this issue, this paper proposes a side-move-aware vehicle dynamic model that explicitly incorporates lateral motion perception. The influence of the proposed model on the vehicle stability region is investigated using phase-plane analysis. Based on the side-move-aware model, a tube-based robust model predictive control (RMPC) strategy is developed to enhance control robustness in the presence of modeling-parameter uncertainties and external disturbances. In addition, a support-function-based state constraint construction method is introduced to facilitate efficient constraint handling. The feasibility and effectiveness of the proposed control scheme are evaluated using a dSPACE–FPGA hardware-in-the-loop platform under multiple vehicle speeds, external disturbances, and modeling-parameter uncertainties. The results show that, compared with the conventional vehicle dynamic model, the proposed side-move-aware model reduces the average tracking-error indices by approximately 55.7% . Compared with the robust H_∞ and sliding-mode controllers, the proposed RMPC reduces the average error indices by 11.8– 19.1% under crosswind disturbances and by 14.7– 28.5% under modeling-parameter uncertainties.
This work proposes and experimentally evaluates a noninvasive method for estimating the speed of hermetic compressors using vibration measurements. The approach employs an all-phase fast Fourier transform (ApFFT) combined with a rate limiter filter and a correction procedure to obtain the dominant frequency regardless of the frequency resolution, which enables the use of reduced acquisition windows. In this way, a rapid response is achieved at low computational cost, making it suitable for embedded systems and applications with transient behavior, such as operation under varying speed or fault conditions. Experimental tests evaluating compressor startup, steady-state operation, and stall conditions demonstrated that the proposed method generally outperforms the conventional FFT. Notably, the proposed method reduced the mean absolute error by 29
This paper presents a full Newton-based methodology for incorporating battery energy storage systems (BESS) into power flow analysis with explicit modeling of primary frequency regulation. The main contribution is the integration of frequency-dependent BESS behavior into the conventional power flow formulation, enabling the representation of active power adjustments in response to frequency deviations within a steady-state framework. Two operating modes are addressed: isolated (off-grid) operation and grid-connected (on-grid) operation in conjunction with conventional synchronous generators. The proposed formulation is embedded directly into the full Newton power flow algorithm, preserving its numerical characteristics while extending its capability to capture frequency regulation effects. The methodology is validated using two representative test systems: a 6-bus system for the off-grid scenario and the New England test system for the on-grid scenario. The results demonstrate the robustness and accuracy of the proposed approach and confirm its effectiveness in representing frequency-dependent behavior and the contribution of BESS units to primary frequency regulation under different operating conditions.
This paper presents a robotic operating system (ROS)/Gazebo-based synthetic data generation and evaluation workflow for offshore wind-turbine inspection using UAV imagery, combining a digital-twin synthetic data generation pipeline with a controlled experimental assessment of hybrid real–synthetic training strategies for damage detection. The core perception task is structural damage detection, addressed with a You Only Look Once version 11 (YOLOv11) object detector. To mitigate the scarcity and acquisition cost of annotated offshore data, we generate synthetic inspection imagery in simulation and further expand it through a compositional copy-paste augmentation strategy that increases scene diversity and reduces context bias. We perform a controlled comparison across four training scenarios (100
This paper addresses a fundamental challenge in nonlinear control: ensuring safety through state constraints and performance through prescribed tracking accuracy for uncertain systems with actuator faults, while guaranteeing practical fixed-time convergence independent of initial conditions. The primary contribution is a unified control framework that integrates four critical capabilities: fixed-time convergence with a priori known settling time to a residual set, strict enforcement of time-varying full-state constraints, prescribed performance tracking without requiring prior knowledge of the reference trajectory, and robustness to actuator faults and system uncertainties. Two key innovations enable these capabilities. First, a novel prescribed performance boundary is constructed using a dynamically scaled function that automatically adapts to the available state space, eliminating the conventional requirement for a priori reference knowledge. Second, newly designed tangent barrier Lyapunov functions with high-order stabilization terms rigorously ensure time-varying state constraints are never violated while enabling fixed-time stability analysis. Neural networks online compensate for system uncertainties and actuator faults. Lyapunov analysis proves all closed-loop signals remain bounded and converge to a residual set within a fixed time. Simulations on a second-order nonlinear system under time-varying constraints and a severe actuator fault demonstrate superior performance compared to existing methods, while strictly maintaining all constraints.
Distributed state estimation under a hard communication budget is limited not only by local packet generation but also by the network-level decision of which candidate packets can actually be admitted during congestion. This paper develops a fairness-regularized value-based admission strategy for bandwidth-constrained cooperative state estimation. Each active node runs a standard local Kalman filter, forms a candidate packet according to an innovation trigger, and reports an uncertainty-reduction value together with its transmission cost. A starvation-sensitive compensation factor reshapes the packet utility so that short-term information gain and long-horizon access fairness are balanced under a strict stepwise byte budget. The resulting admission problem is a binary knapsack problem that is solved online by a low-complexity value-density scheduler and coupled with a covariance-intersection backend for conservative fusion under unknown cross-correlation. The analysis includes a bounded-admission-gap sufficient condition for covariance boundedness and a practical tuning rule for the fairness parameters. The evaluation covers adaptive and exact-knapsack baselines, scalability tests for N=8,12,20,30 nodes, two-dimensional β – τ _max sensitivity, and a time-varying-budget scenario with content-dependent packet costs. The results show that the proposed method keeps RMSE and ANEES comparable to utility-only and exact utility-only admission, while clearly reducing long admission gaps and improving Jain fairness. In the default congested evaluation window, FRUA reduces the maximum silence length from 18.2 to 14.0 steps and improves the Jain index from 0.952 to 0.974 relative to UOA. When the network size increases to N=30 , FRUA keeps the maximum silence length below 18 steps, whereas UOA and exact UOA exceed 150 steps. These results indicate that fairness-aware admission is a useful system-level complement to local communication-efficient estimation when congestion, packet competition, and unknown cross-correlation coexist.
This paper deals with the input-to-state (IS) stabilization of switched cyber-physical systems (SCPSs) that involve two communication networks (sampler-to-buffer and controller-to-actuator). Transmission delays in both communication networks are considered. Two data buffers, used to store the latest sampled data and the latest received data, are employed in the control unit. To avoid using disordered data packets received by the control unit, a proactive strategy to drop the disordered data packets is proposed. A switched state feedback controller is designed, which can develop a closed-loop switched system with delays and asynchronous switching. Although the actuator cannot identify the disordered data of the control input, the proposed control scheme is still capable of reducing the number of disordered switching events and weakening the adverse effects of the disordered data on the system performance. A sufficient condition to ensure the solvability of the IS stabilization problem is derived using the mode-dependent average dwell time (MDADT) method and the Lyapunov functional technique. A step-by-step procedure for obtaining a feasible solution is also provided. Lastly, the model of a mass-switched unmanned marine vehicle is utilized to show the effectiveness of the proposed results.
In this paper, a robust finite control set model predictive control (FCS-MPC) strategy is proposed for active power injection into the grid using a single-phase DC/AC inverter with an LCL filter. The overall system is modeled as a hybrid system within the linear complementarity (LC) framework, thereby preserving the power converter nonlinear and switching behavior. Within the FCS-MPC formulation, a bipolar switching policy and a one-step prediction horizon are adopted to minimize a quadratic objective function based on the grid-current tracking error, with the reference synchronized to the grid voltage. For online parameter identification, a simplified linear representation of the LC-based model is derived, while the converter’s discrete switching nature is retained in the control law. A recursive least squares (RLS) algorithm is integrated into the predictive control framework to estimate the model parameters online, enhancing robustness against uncertainties in the LCL filter parameters and DC-side voltage variations. The performance of the proposed control strategy is validated through numerical simulations and experimental implementation, demonstrating accurate current tracking and robust active power delivery to the grid under parameter variations and grid disturbances.
The increasing photovoltaic (PV) penetration in distribution systems poses technical and economic challenges for both operation and planning. This paper proposes a short-term planning model formulated as a mixed-integer linear programming (MILP) problem that coordinates line reconductoring, allocation of fixed and switched capacitor banks (CBs), and installation of voltage regulators (VRs). The model incorporates PV inverters with Volt-VAR control (VVC), in accordance with the IEEE 1547-2018 standard, and represents the progressive evolution of load and distributed generation across three operational stages. The framework is validated on two distribution test systems with distinct characteristics: a 135-bus and the IEEE 123-bus feeders. The results show that, although both systems benefit from VVC, the effects on investment decisions are system-dependent. In the 135-bus feeder, the reactive support provided by smart inverters allows VRs to be eliminated once PV multiplying factor exceeds 1.6, while also reducing energy loss costs. In the IEEE 123-bus system, which exhibits a more pronounced voltage drop, the VR is installed regardless of VVC considerations; nonetheless, adopting PV-provided VVC allows deferral of investment in switched CBs, thereby reducing total planning costs. The proposed model, therefore, offers a flexible decision-support tool that adapts its reinforcement recommendations to each network.
Reinforcement learning (RL) is a promising approach for adaptive flight control of fixed-wing unmanned aerial vehicles (UAVs) operating under nonlinear dynamics and disturbances. However, the influence of RL training strategies on hierarchical flight control performance remains insufficiently explored. This study compares three RL-based outer-loop strategies for altitude and heading tracking, while conventional PID controllers stabilize inner-loop angular rates. The RL agent observes altitude and heading errors, their derivatives and integrals, and selected aircraft states, and outputs continuous roll, pitch, and yaw angle commands. The reward function penalizes tracking errors, oscillations, and excessive control effort to ensure smooth and stable responses. Three training paradigms are evaluated using DDPG: centralized training with decentralized execution (CTDE) using cooperative agents, fully decentralized task-specific agents, and a single-agent multi-task (SAMT) framework that learns a unified policy. Under identical disturbance conditions, the decentralized strategy achieves the highest steady-state accuracy (1.34 m altitude error, 0.59^∘ heading error) but requires ∼ 30,000 episodes. The centralized strategy converges in ∼ 5000 episodes with higher RMSE ( ∼ 2.48 m). The SAMT approach converges in fewer than 1000 episodes while maintaining low RMSE ( ∼ 1.92 m). These results demonstrate that training architecture significantly affects learning efficiency, robustness, and control smoothness.
Robust and efficient power regulation is essential for the safe operation of molten salt breeder reactors (MSBRs), which exhibit strong nonlinearities, time-varying dynamics, and significant modeling uncertainties. This paper develops an event-triggered H_∞ -based fractional-order proportional–integral–derivative (FOPID) control framework for MSBR power regulation under mixed additive and multiplicative output uncertainty. The uncertainty structure explicitly captures sensor imperfections, unmodeled output dynamics, and process drift. A graphical H_∞ tuning methodology is employed to determine stabilizing controller parameters with explicit robustness margins. An event-triggered execution mechanism is incorporated to generate aperiodic control updates, thereby reducing computational and communication burden while ensuring the existence of a strictly positive minimum inter-event time. A Lyapunov-based analysis establishes input-to-state stability of the closed-loop system under non-periodic updates. To improve adaptability under time-varying operating conditions, a long short-term memory (LSTM) network is integrated to predict fractional orders and update weights online within stability-admissible bounds. Simulation studies demonstrate accurate power tracking, effective disturbance rejection, reduced settling time compared with benchmark methods, and robust performance under actuator faults and large parametric uncertainties. Monte Carlo analysis further confirms stable operation and consistent robustness across the considered uncertainty range.
This paper addresses the complex path-planning problem for automated guided vehicles (AGVs) in semiconductor fabrication facilities, which are characterized by dense layouts, narrow corridors, and dynamic obstacles. To address these challenges, we propose the hierarchical state feature-driven deep reinforcement learning (HSF-DRL) framework, which leverages the options framework in hierarchical reinforcement learning (HRL) to decompose navigation into a two-tier decision-making process. Specifically, a rule-based high-level meta-controller selects temporally extended options (e.g., global navigation, dynamic avoidance, and precise docking) by integrating online heuristic search with context-specific features; meanwhile, a low-level executor, implemented with a Deep Q-Network (DQN), generates primitive actions. A key contribution is the option-conditioned dynamic feature-fusion mechanism, where the weights of environmental, procedural, and heuristic features are conditioned on the active high-level option. This enables context-aware perception and mitigates the fixed-input limitations of conventional DRL. Evaluations conducted exclusively within a 2D grid-based semiconductor-fab simulation demonstrate that HSF-DRL outperforms traditional DQN and Dyna-Q baselines in path optimality, convergence speed, and stability under highly dynamic scenarios. Overall, this work provides a structured architectural approach for AGV navigation, establishing a foundation for future physical deployments in complex industrial settings.
In this paper, we propose a novel criterion for practical synchronization of discrete-time Lur’e-type complex dynamic networks (CDNs) with event-triggered control. In particular, we consider that each node is a Lur’e system subject to control saturation and persistent disturbances, operating with an asynchronous relaxed event-triggering mechanism (ETM). Based on the Lyapunov Theory, we derive constructive conditions expressed in terms of quasi-linear matrix inequalities to guarantee that the synchronization error of the CDN is ultimately bounded considering prescribed sets of admissible initial errors and disturbances. These conditions are cast into a semidefinite programming problem to codesign the decentralized controllers and the parameters of the ETMs aiming at reducing the frequency of control updates compared to the time-triggered strategy. The synchronization error norm in the ultimate bounded regime is upper bounded by a positive constant specified by the designer. A numerical example is presented to illustrate the proposed method.
