This paper develops a design criterion for continuous-time extended state observers (ESOs) that quantifies the estimation-error energy induced by step-like disturbances under stochastic uncertainty. For absolutely continuous disturbance variations, the induced estimation-error energy is shown to admit an upper bound characterized by the H2 norm of the transfer from the disturbance variation to the estimation error. A step disturbance is then treated as the limit of a sequence generated by an approximation identity, and the corresponding finite-horizon estimation-error energy is characterized exactly in this limit. The corresponding long-horizon limit is shown to be characterized by the H2 norm of the transfer function from the disturbance variation to the augmented-state estimation error. The analysis is further extended to an Ito-type stochastic system with process and measurement noise, where the finite-horizon expected estimation-error energy is decomposed into a disturbance-induced mean-response term and a noise-induced covariance term. Based on this finite-horizon characterization, a long-horizon approximate design criterion consisting of an H2 term and a steady-state covariance term is introduced. The observer gain minimizing the approximate criterion is shown to be characterized by an algebraic Riccati equation. Numerical simulations validate the proposed limiting finite-horizon characterization by comparing analytical values with Monte Carlo sample means and illustrate the approximate criterion as a design-oriented surrogate.
This paper proposes a packet-based static output-feedback control framework for cyber–physical systems under Denial-of-Service (DoS) attacks. The approach models the system under consecutive packet losses as a switched system and develops a linear matrix inequality based method to design switching-dependent control gains. The framework incorporates performance specifications through decay rate and robustness margin parameters. Numerical results show that the switching-dependent design outperforms the use of constant gains. Moreover, the practical effectiveness is demonstrated through numerical experiments where the designed controller maintains stability under DoS attacks with improved convergence time.
Coordinated frequency regulation by energy storage clusters and thermal power units is an effective way to enhance grid frequency security. This paper proposes a joint thermal power unit and energy storage cluster (TP-ESC) frequency regulation strategy based on a modified constrained Bayesian optimization (MCBO) algorithm to ensure safe frequency control. Firstly, a dynamic filter is designed to allocate frequency regulation commands within the TP-ESC hybrid system by leveraging their complementary frequency regulation characteristics. Secondly, for the energy storage cluster, a state-of-charge (SOC)-based dynamic allocation strategy is proposed to prevent overloading and frequent low-power responses while maintaining SOC consistency. In addition, dual dynamic security constraints, namely the rate of change of frequency (RoCoF) and the maximum frequency deviation (MaxFD), are enforced to enhance system stability and disturbance rejection capability. Finally, the load frequency controller parameters are optimized via the MCBO algorithm and compared with Bayesian optimization (BO) and constrained Bayesian optimization (CBO) methods. Simulation results on a two-area power system show that the control strategy based on MCBO algorithm can keep both RoCoF and MaxFD within the prescribed security constraints. Moreover, compared with the control strategy based on BO algorithm, it reduces the peak frequency deviations in area 1 and area 2 and the peak tie-line power deviation by 24.94%, 34.44%, and 31.25%, respectively These results demonstrate that the proposed method effectively enhances frequency regulation performance, system security, and operational stability.
The growing demand for wireless connectivity has increased pressure on shared spectrum, especially between satellite communication networks and emerging 5G systems. When these signals operate close to one another frequently, noticeable interference may occur and reduce signal quality. This study explores a deep learning-based interference mitigation approach that combines spectrogram-driven CNN classification with adaptive power control. The CNN model identifies three transmission cases satellite-only, terrestrial-only, and coexistence and achieved accuracy close to 98% during testing. Using the classification output, an adaptive power strategy was applied, resulting in an improvement in SINR from roughly 10 dB to over 27 dB under coexistence conditions. A feedforward neural network supported by a genetic algorithm was also used to fine-tune system parameters. The findings demonstrate that the proposed method strengthens terrestrial–satellite coexistence and offers a practical direction for interference control in future wireless systems.
Adaptive feedforward control of multi-harmonic disturbances in rotating machinery faces a fundamental challenge when harmonic amplitudes exhibit high dynamic range (HDR >50dB): standard adaptive controllers suffer from severe mode-wise convergence imbalance, where weak harmonics remain unregulated while dominant modes saturate. This control performance disparity stems from ill-conditioning (condition number κ∼105) in the filtered reference correlation matrix, causing convergence time constants to differ by four orders of magnitude across modes.Within the PNANC block-diagonal structure, per-channel normalization and steady-state sinusoidal mapping can be integrated into a unified framework that simultaneously achieves balanced mode regulation and computational efficiency. The resulting low-complexity eigenvalue-equalized (LC-EE) algorithm exploits the block-diagonal eigenstructure for structured per-channel normalization and the sinusoidal reference structure for steady-state replacement of FIR convolution by O(K) matrix operations. This restores uniform controllability by compressing the condition number from κ≈9×104 toward unity through diagonal preconditioning, while reducing arithmetic complexity by 98.4% compared to conventional implementations.Performance is validated through controlled simulation and semi-synthetic experiments using authentic industrial broadband noise from the MIMII dataset. The integrated framework achieves 23.4× convergence acceleration and 9.3 dB improvement in average steady-state disturbance attenuation relative to standard PNANC. Per-harmonic convergence analysis confirms that all K=8 controlled harmonics converge uniformly under eigenvalue equalization, whereas only the fundamental mode adapts under standard PNANC. Statistical analysis across 12 recordings from 4 distinct machines confirms significance (p<0.001). Robust stability is maintained under severe broadband turbulence (SNR to −5dB), secondary-path magnitude modeling errors up to 30%, and frequency jitter up to 0.5 Hz RMS. Extended mismatch testing across seven scenarios establishes practical limits: combined magnitude, phase, and delay errors should remain within approximately ±10%, ±15°, and ±5 samples, respectively, for guaranteed stability. These results suggest that the observed performance improvements stem primarily from structural properties of the PNANC framework rather than parameter retuning, and indicate the potential of the proposed realization for low-cost embedded implementation in HDR disturbance control applications.
Urban overheating calls for building materials with climate adaptation impacts. The textile-based HydroSKIN facade with spacer fabric can, among other things, evaporate water on hot days to achieve surface temperature reductions of up to 20K. As groundwork for the development of operational and control strategies for such systems, this study presents a coupled transport and storage model describing water flow within the element. Sensitivity analysis and experimental validation demonstrate the models ability to quantitatively capture and represent the dominant mechanisms with deviations in range of 50 grams during irrigation.
Alzheimer's disease (AD) is increasingly seen as both structural degeneration and disruption of brain network dynamics. Yet, quantitative neurovascular dynamics characterization is limited. We propose a control-theoretic framework to examine neurovascular complexity using resting-state fNIRS. Data were collected from 83 participants: healthy controls (HC, n = 27), mild cognitive impairment (MCI, n = 37), and AD (n = 19). Sliding-window nonlinear complexity measures - Higuchi's fractal dimension, spectral entropy, and wavelet entropy - were computed for each channel to construct complexity-coupling networks based on inter-channel temporal coordination. Static topology, dynamic descriptors (mean, variability, range, temporal dependence), and state-based metrics from k-means clustering were analyzed. Task-rest reconfiguration and cognitive performance associations were evaluated. AD showed reduced network integration and selective loss of positive complexity coupling. Decreased temporal variability and state entropy, with fewer visited states, indicate diminished dynamic flexibility and state space contraction. Additionally, AD exhibited reduced task-induced network reconfiguration, reflecting impaired neurovascular adaptability. Among measures, spectral entropybased networks, especially in deoxyhemoglobin signals, showed the strongest group discrimination and robust cognitive score associations. These findings show AD involves collapse of neurovascular dynamic complexity, with reduced spectral diversity, increased temporal rigidity, and constrained state-space dynamics. The framework offers a sensitive, mechanistically interpretable biomarker for tracking disease progression. (c) 2026 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
Proportional-integral-derivative (PID) control remains ubiquitous in industrial applications, yet its performance heavily depends on proper parameter tuning. In practical scenarios where obtaining an accurate mathematical model is difficult and costly, data-driven tuning emerges as a highly effective alternative. However, existing data-driven methods typically optimize either set-point tracking (servo performance) or disturbance rejection (regulator performance) independently, lacking the flexibility to systematically adjust the critical trade-off between them. To address this limitation, we propose a systematic, frequency-domain data-driven approach for optimizing PID controller parameters under strictly guaranteed stability margins. Utilizing a single set of measured input-output data, the method estimates the plant's frequency response without requiring a parametric model. Within this framework, we define a unified cost function based on the error between the reference and the closed-loop output response. By introducing a flexible trade-off parameter, the design explicitly optimizes the balance between servo and regulator performances. The key contribution is the incorporation of a prescribed maximum sensitivity constraint into this objective, ensuring robust operation against uncertainties while maximizing the achievable performance balance. The effectiveness of the proposed method is validated through numerical simulations on a time-delay plant. Comparative evaluations demonstrate that our approach achieves control performance and robustness equivalent to conventional modelbased trade-off methods. Notably, in the presence of modeling errors, the proposed data-driven method completely avoids the performance degradation seen in conventional approaches, providing a powerful and highly practical tool for real-world control design. (c) 2026 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
This paper presents Service station-enhanced Mainstream Traffic Flow Control (ST-MTFC), a novel traffic management strategy that integrates Variable Speed Limit (VSL)-based Mainstream Traffic Feedback Control (MTFC) with Infrastructure-dependent Control (IDC) actions. The coordinated approach combines mainstream regulation with ALINEA ramp-metering at service station exits and route guidance at service station entrances, while modeling service stations as dynamic storage elements, providing additional degrees of freedom and operational flexibility for control design, particularly under high-demand scenarios. The control design employs an advanced version of the METANET model, METANET-s, which incorporates service station dynamics to capture complex traffic phenomena with greater accuracy. ST-MTFC is conceived as an improved alternative to classical MTFC, applicable to freeway segments that include service stations. By utilizing the presence of service stations and their associated entrance and exit controls, ST-MTFC extends the ability of MTFC to manage congestion at bottlenecks, protect ramp accessibility, and reduce excessive queuing in upstream VSL-controlled links. The ST-MTFC strategy delivers significant improvements in travel time savings and congestion reduction. The efficacy of the proposed approach is validated through extensive simulations conducted with the AIMSUN Next micro-simulation platform.
This paper addresses robust output preservation for uncertain nonlinear feedback systems subject to structured perturbations and bounded restorative intervention. Motivated by phenotype maintenance in biological regulatory networks, we formulate pathological anomalies as admissible uncertainty channels that deform the equilibrium output of a nonlinear system, while restorative action is modeled as a bounded input acting on the feedback interconnection. The proposed framework establishes positivity, boundedness, existence of positive equilibria, local equilibrium continuation, and compact-set local stability over uncertainty–intervention families. It further provides a computable sufficient condition for robust output-threshold preservation and an upper bound on the minimal intervention effort required to recover a prescribed safety level. A key consequence is that open-loop perturbation severity and restorative difficulty are generally distinct, since the intervention gain may vary across perturbation channels. The framework is illustrated on a reduced p53–MDM2 feedback model, where TP53 loss-of-function, MDM2 amplification, and impaired upstream signaling are represented as structured pathological perturbations. Simulations show that these perturbations may degrade the tumor-suppressive phenotype without destabilizing the equilibrium, and that bounded intervention restores phenotype safety with channel-dependent effort. The results position biological pathway degradation as a robust nonlinear output-preservation problem, with potential relevance for uncertainty-aware analysis of safety-critical feedback systems.
Digital twin (DT) models are increasingly used to support personalized therapy design in type 1 diabetes (T1D). This study extends previous in-silico validation of a DT-based decision support system by presenting a retrospective evaluation of the DTs generated during a clinical trial (Clinical Trials Registration: NCT05610111). The dataset includes 72 individuals with T1D followed over 16 weeks. Every biweek, DTs were generated and used to obtain therapy recommendations to be implemented by the participant in the following biweek. This work aims to assess the DT fidelity and predictive capability when deployed as a decision-support tool in a human-in-the-loop setting. Across 541 eligible biweekly windows, DT replay under original therapy settings achieved a median (25th-75th percentile) RMSE of 10.7 (5.6-17.8) mg/dL. Predictive consistency was evaluated by comparing DT-predicted and observed biweekly changes in CGM-derived metrics, including mean glucose, glucose variability, and time in glucose ranges. When the recommended therapy profile was predominantly applied, predicted and observed changes in these metrics were statistically equivalent within pre-specified clinical margins. These results indicate that DTs can maintain high predictive capability in free-living conditions, establishing them as a reliable and robust tool for personalizing insulin therapy in real-world clinical practice. (c) 2026 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
In the recent work (Li et al., 2024), a neural network-based control framework with provable stability guarantees was proposed for nonlinear systems. While effective, a key practical challenge in deploying this framework lies in solving the optimization problems required for generating training data. This short communication complements (Li et al., 2024) by presenting a block-alternating iterative procedure for addressing these optimization problems. Basic theoretical properties of the proposed algorithm, including feasibility preservation and convergence to stationary points, are established. The effectiveness of the approach is illustrated through a temperature regulation experiment. (c) 2026 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).
This paper presents the design and implementation of a lightweight, portable hand exoskeleton controlled via electroencephalography (EEG) signals to support individuals with hand impairments in performing activities of daily living (ADLs). The device is fabricated using 3D-printed PLA material and actuated by a linear motor under Arduino Uno control. With two degrees of freedom (DoFs) per finger, the prototype achieves a total of eight DoFs across four fingers. EEG signals - specifically eyeblink detection - serve as intuitive control inputs, enabling users to initiate flexion and extension movements. In this study, EEG signals are used as event-based commands rather than continuous control inputs for system dynamics. A closed-loop position regulation mechanism using potentiometer feedback ensures stable and safe actuator displacement during operation. Experimental validation demonstrates effective assistance in hand opening and closing, with a maximum grasping force of 11.88 N and a range of motion of 67 degrees. Compared to existing designs, the proposed exoskeleton achieves reduced weight (298 g including controller and battery) while ensuring safety, comfort, and functional grasping capability. These findings suggest that the integration of EEG-based control with feedback, combined with a cost-effective 3D-printed rigid structure, provides a promising approach for enhancing autonomy and independence in patients with impaired hand function. The presented results represent a preliminary proof-of-concept validation conducted on a single healthy subject. (c) 2026 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
Achieving high control performance while minimizing resource usage is a key challenge in industrial systems. Fractional-order PID (FOPID) controllers offer enhanced performance over integer-order PID. Meanwhile, event-based control architectures reduce resource use and costs. However, integrating FOPID control with event-based architectures is challenging, as standard discrete-time approximations' accuracy of fractional dynamics degrades under non-uniform sampling. This paper presents a practical method to implement FOPI and FOPID controllers in an event-based framework. Building on industrial PIPlus and PIDPlus architectures, the approach uses state-space realizations of fractional operators, enabling the controller's internal states and output to be updated in closed form only at event times. A simulation study compares the proposed method against an existing direct-discretization approach, which represents the only alternative available in the literature to implement event-based fractionalorder controllers, by considering different controller configurations. The results show that the proposed method reproduces the response of the reference time-based controller more accurately. In particular, it achieves substantial reductions in the integrated absolute error between the event-based response and the reference response obtained with standard periodic sampling (IAEP index), with average improvements ranging from 38% to 68% across the different controller configurations considered. These findings suggest that the proposed state-space approach is a reliable solution for deploying existing fractional-order controllers in event-based architectures while preserving their intended closed-loop dynamic behavior. (c) 2026 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
This paper presents a data-driven framework for verifying static state-feedback stabilisers for control-affine nonlinear systems using a Koopman-lifted behavioural representation. The state is embedded into a finite-dimensional observable space, and one-step lifted data matrices are constructed from persistently exciting exploratory trajectories collected under an auxiliary pre-stabilising feedback. To reduce the conservatism associated with global-fit residual bias, a two-stage protocol is introduced. In Stage (i), the exploratory lifted data are used to construct a behavioural representation of a candidate closed loop via the Koopman-lifted extension of Willems’ fundamental lemma, to screen candidate feedback gains. In Stage (ii), dedicated near-equilibrium closed-loop perturbation data are used to identify a local closed-loop Jacobian, from which a quadratic Lyapunov function is computed via a discrete Lyapunov equation. Under the local embedding and residual assumptions, this lifted Lyapunov function induces a local Riemannian contraction metric on the underlying nonlinear manifold. Building on this construction, the paper develops two complementary certificates: a deterministic local Region-of-Attraction inner bound, together with Local Input-to-State Stability margins against unmodelled dynamics, obtained from a local residual bound; and a scenario-certified probabilistic one-step Lyapunov decrease set obtained using the Scenario Approach applied directly to the nonlinear closed-loop dynamics. The framework is validated on an inverted-pendulum benchmark and shown to be competitive with strong baseline controllers while additionally providing explicit stability certificates grounded in measured trajectory data.
We study the control of opinion dynamics in large populations when a strategic decision maker repeatedly broadcasts public messages that shape the evolution of opinions. The population state is modeled as a probability distribution over finitely many ideological bins, and its evolution defines a decision-dependent Markov chain on the probability simplex. On the theoretical side, we introduce a class of contractive opinion dynamics in which the linear part combines inertia and peer influence, while an action-dependent source term models the direct impact of messaging. Under explicit contractivity and regularity assumptions, we prove that the deterministic dynamics associated with any fixed policy admit a unique steady-state distribution and extend this result to the stochastic case by showing the existence and uniqueness of an invariant measure on the simplex. On the empirical side, we employ Proximal Policy Optimization (PPO) as a reinforcement learning mechanism for the decision maker. Numerical experiments indicate that PPO can learn communication strategies that substantially reduce long-run polarization and keep the opinion distribution near moderately centralized regimes, although no general convergence guarantees are provided. The simulations further show that the qualitative long-term behavior of the population depends strongly on the messaging pattern: fact-checking and moderate messages tend to promote centralization, whereas predominantly provocative messaging sustains bimodal and more polarized steady or weakly oscillatory distributions. The results illustrate how contraction-based analysis and policy-gradient methods can be combined to study decision-dependent stochastic systems on the space of distributions. (c) 2026 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
This study demonstrated that simulation based experiential learning pedagogy using MATLAB Simscape and Copilot is an effective tool to reduce the theory practice gap in control systems education. Integrating Kolb's cycle with AI-enhanced, segmented activities allowed students to understand the mechanical-electrical analogies and PID tuning, thereby preparing them for industry skillsets. With engineering curricula changing in the context of AI era, these kinds of pedagogies provide a scalable solution to the issue of rote learning. Due to the strong understanding of fundamental concepts, the students develop critical thinking to solve complex systems and face the challenges of the industry relevant problems with confidence and creativity. Finally, it is not the equations on the blackboard, but hands that transform virtual world into real knowledge, makes true teachers. (c) 2026 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
The increasing penetration of renewable energy imposes increased flexibility requirements on supercritical coal-fired power plants (SCFPPs). To address these challenges, this study proposes a stochastic model predictive control (SMPC) strategy enhanced by Gaussian process regression (GPR), wherein the GPR model captures unmodeled dynamics and compensates the nominal model. A dynamic model of a SCFPP was constructed with its accuracy verified against field data. Model uncertainties are inferred during the prediction process using GPR and the propagated uncertainty information is incorporated into the MPC framework, thereby forming a Gaussian process (GP)-based stochastic model predictive control strategy. Simulation results demonstrate that under load-rising conditions, GP-SMPC outperforms the PID in multi-loop tracking performance, achieving an 9.93% improvement in the IAE index. Furthermore, external disturbance simulations verify that GP-SMPC delivers improved control performance over the PID, yielding smaller ME and faster responses, with average IAE reductions of 66.54%, 77.92%, and 61.35%, while the average settling times are all reduced to within 92s. Finally, closed-loop tests validate the practical applicability of the proposed strategy. By compensating model uncertainties and enhancing disturbance rejection, GP-SMPC improves SCFPPs flexibility and stability under high renewable penetration. (c) 2026 Published by Elsevier Ltd.
Aquaponics research is often limited by heterogeneous measurement practices and inconsistent data structures, which complicate comparative analysis and model-based optimization across experimental systems. This paper proposes a unified digital infrastructure combining (i) a standardized measurement protocol defining monitored variables, acquisition frequencies, and harmonized data formats, and (ii) a transient MATLAB-based dynamic model for simulating biomass growth and nutrient cycling in recirculating aquaponic systems. The model is implemented using an explicit Euler-Forward scheme and incorporates fish growth, nitrogen transformations, solids removal, and plant nutrient uptake within a modular framework. Scenario-based simulations illustrate how feeding intensity, plant cultivation area, and nitrification efficiency influence biomass production and nitrogen dynamics. A baseline sensitivity and uncertainty analysis further evaluates the robustness of the proposed framework. The results identify feeding intensity as the primary driver of biomass growth, while plant area and nitrification efficiency regulate nitrate accumulation and removal. The study introduces three compact performance indicators-residual nitrate load, uptake fraction, and feed-based efficiency-to support benchmarking and control-oriented interpretation of aquaponic systems. The proposed approach establishes a reproducible data-model workflow that can serve as a foundation for future aquaponic digital-twin implementations. (c) 2026 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
Existing research in aero-engine rotor assembly optimization faces an inherent dichotomy between continuous-space heuristic algorithms and the discrete nature of the assembly problem, coupled with a lack of systematic frameworks for algorithm selection in complex engineering scenarios. This paper proposes a synergistic optimization and decision-making framework to bridge this gap. Firstly, a novel discretization framework is introduced, which adapts continuous-space metaheuristics to the discrete domain by integrating differential operations with probabilistic mutation, simulated annealing, and adaptive local search strategies. Secondly, a multi-dimensional performance evaluation is conducted, assessing convergence accuracy (0.058 g mm for the discretized Harris Hawks Optimization algorithm), robustness under data uncertainty, and computational efficiency (0.021 s for Simulated Annealing algorithm). Finally, to resolve the multi-criteria trade-offs, we propose Large Language Model-informed Entropy VIKOR (LLM-E-VIKOR). This method innovatively integrates expert-level subjective weights with objective entropy weights to ensure the evaluation aligns with high-precision industrial exigencies. Experimental results demonstrate that the discretized Harris Hawks Optimization achieves the highest comprehensive score (1.0) under the proposed framework. The proposed framework effectively reconciles the trade-off between computational efficiency and optimization accuracy, offering a robust, high-precision, and expert-informed solution for aero-engine rotor assembly process planning. (c) 2026 Elsevier Ltd. All rights are reserved, including those for text and data mining, AI training, and similar technologies.
