Abstract Unmanned Aerial Vehicles (UAVs) can be used to track growing and moving boundaries such as those of wildfires where the boundary cannot be prespecified. Towards this, we first present a Model Predictive Control (MPC) formulation for this task, which systematically incorporates vehicle dynamics, evolving boundary models, and input constraints to enable precise tracking. While effective, solving the nonlinear optimization online incurs high computational cost, limiting real-time deployment. To address this, we propose a novel receding-horizon guidance law that replaces the optimization step with a closed-form solution based on steady-turn motion primitives embedded in a receding-horizon framework. This approach generates circular-arc trajectories in lieu of the computationally expensive optimization routine, while preserving the predictive nature of the formulation and enabling real-time onboard implementation. Simulation studies validate the method across varying UAV initial conditions, prediction horizons, and fire model parameters, demonstrating that it achieves tracking performance comparable to MPC while reducing computation time by several orders of magnitude.

ABSTRACT In this study, an event-based nonlinear model predictive control (EANMPC) method is developed to address the trajectory tracking problem of unmanned hovercraft subjected to unknown time-varying environmental disturbances and actuator saturation. Initially, the EANMPC method is applied to enhance tracking performance through rolling optimisation and online optimisation. Furthermore, it is difficult to transfer control signals to controlled systems, while the model predictive control (MPC) method occupies huge computational resources and makes it difficult to ensure timeliness. Moreover, the paper provides a proof of the adaptive control horizon’s stability and presents the simulation results of two cases. The results and tables indicate that the proposed method has better tracking performance and less computer resource cost.

This article investigates the distributed online optimization problem in a zero-sum game between two distinct time-varying multiagent networks. At each iteration, the agents not only communicate with their neighbors but also gather information about agents in the opposing network through a time-varying network, assigning weights accordingly. Moreover, we consider quantized communication and bandit feedback mechanisms, with agents transmitting quantized information and adopting one-point estimators. At each iteration, agents make and submit decisions and then receive the cost function values near their decision points rather than the full cost function information. To guarantee the payoff of each network, we design an algorithm named quantized distributed online bandit optimization in two-network (QDOBO-TN). We use dynamic Nash equilibrium regret to measure the positive payoff discrepancy between the decision sequence produced by Algorithm QDOBO-TN and the Nash equilibrium sequence. Furthermore, we propose a multiepoch version of Algorithm QDOBO-TN. The regret bounds for both algorithms are sublinear with respect to the iteration count T. Finally, we conduct a series of simulation experiments that further validate the effectiveness of the algorithms.

An embodied intelligence-based online optimization methodology for injection molding process using multi-cavity hot-runner

Multi-objective online optimization of building energy systems for improved control smoothness and efficiency

As an actuation mechanism for achieving precision attitude control in aircraft, the electromechanical actuator (EMA) plays a critical role in ensuring flight safety and stability. However, the EMA is subject to unmeasurable unknown disturbances that act through mismatched channels relative to the system’s control input. To address this, this paper employs feedback linearization to transform the existing model. The transformed model effectively recasts the unknown disturbance into the same channel as the control input, thereby enabling active disturbance rejection via control law design. Furthermore, to overcome the challenge of immeasurable disturbances, an extended state observer (ESO) is designed to estimate the unknown disturbance; the estimated value is then directly utilized in the control law synthesis. Subsequently, a fuzzy logic system (FLS) is developed to perform real-time online adaptation and optimization of the controller parameters. Finally, extensive simulation results are provided to validate the effectiveness of the proposed algorithm.

INTRODUCTION: In wind power generation systems, the unstable variability of wind energy significantly affects control quality and power stability. Conventional PID controllers often show limitations in nonlinear systems or systems with time-varying parameters, especially when integral windup and degraded transient performance occur. OBJECTIVES: This paper proposes an online optimization method for PID parameters based on a Genetic Algorithm (GA), applied to a simplified dynamic model of a wind power generation system, in order to improve the system response quality. METHODS: The studied system is modeled by a second-order transfer function representing the system’s inertia and friction characteristics. The GA is implemented in a real-time optimization manner, using an objective function based on the ITAE criterion to evaluate and select the optimal PID parameter set. RESULTS: Simulation results show that the proposed online GA–PID approach improves settling time, reduces overshoot, and eliminates steady-state error more effectively than fixed PID and conventional anti-windup PID controllers. CONCLUSION: The proposed online GA–PID method is suitable for energy systems with high variability and adaptive control requirements, especially in wind power generation applications.

Compact heavy-duty skid-steer robots are increasingly used for city logistics and intralogistics tasks where high payload capacity and stability are required. However, their limited maneuverability and non-negligible turning radius challenge conventional waypoint-tracking controllers that assume unconstrained motion. This paper proposes a curvature-constrained trajectory planning and control framework that guarantees geometrically feasible motion for such platforms. The controller integrates an explicit curvature limit into a finite-state machine, ensuring smooth heading transitions without in-place rotation. The overall architecture integrates GNSS-RTK and IMU localization, modular ROS 2 nodes for trajectory execution, and a supervisory interface developed in Foxglove Studio for intuitive mission planning. Field trials on a custom four-wheel-drive skid-steer platform demonstrate centimeter-scale waypoint accuracy on straight and curved trajectories, with stable curvature compliance across all tested scenarios. The proposed method achieves the smoothness required by most applications while maintaining the computational simplicity of geometric followers. Computational simplicity is reflected in the absence of online optimization or trajectory reparameterization; the controller executes a constant-time geometric update per cycle, independent of waypoint count. The results confirm that curvature-aware control enables reliable navigation of compact heavy-duty robots in semi-structured outdoor environments and provides a practical foundation for future extensions.

Thermodynamic Performance Online Optimization and Constraint Analysis of a Multi-Combustion Chamber Pre-Cooled Variable Cycle Engine

While tremendous progress has been made towards achieving highly efficient and practical Private Set Intersection (PSI) protocols during the last decade, the development of post-quantum PSI is still far from satisfactory. Existing post-quantum PSI protocols encounter a dilemma: while those based on fully homomorphic encryption (FHE) achieve low online communication, they suffer from significant online computation; conversely, protocols based on post-quantum Oblivious Pseudorandom Functions (OPRFs) exhibit excellent online computational performance but incur substantially high online communication. To overcome this dilemma, we present a lattice-based PSI protocol that achieves optimal online performance in both communication and computation. Our solution introduces two core innovations: a robust signal comparison algorithm based on RLWE key exchange, which determines the intersection through signal consistency rather than direct shared key comparison, and an optimized Oblivious Key–Value Stores (OKVS) implementation featuring a composite key–value mapping for efficient handling of high-dimensional RLWE polynomials. We implement the protocol and conduct extensive benchmarks in both symmetric and asymmetric set-size settings. The results show that our construction achieves the lowest online overhead in both computation and communication among all tests. For example, with asymmetric set sizes (212,11041), the online phase requires only 0.132 s, yielding 19× and 282× improvements over FHE-based (CCS’21) and OPRF-based (EUROCRYPT’25) protocols, respectively. Even at (224,11041), our online communication time is only 0.201 s, which is 226× and 184× that of FHE-based and OPRF-based PSI, respectively. Additionally, our online communication overhead is the lowest in all tests; however, this comes at the cost of heavy offline communication overhead for very large set sizes, revealing a clear trade-off between pre-computation and online efficiency. This work addresses a critical gap in post-quantum PSI by delivering a protocol that achieves balanced online communication and computational overhead, thereby enabling broader practical deployment.

With the development of smart agriculture, agricultural production scheduling optimization has become the key to improving resource utilization efficiency and economic benefits. However, traditional methods are difficult to cope with complex decision-making needs in multi-objective and dynamic environments. In this regard, this study proposes a hybrid optimization model that integrates INSGA-III (Improved Non-dominated Sorting Genetic Algorithm III) and DynaQ (Dynamic Q-learning) to achieve multi-objective collaborative optimization and dynamic adaptability. The model adopts a dual layer architecture of "offline optimization online correction", where the upper layer generates global non supported solutions through the introduction of adaptive crossover mutation operator INSGA-III to solve multi-objective optimization problems of maximizing production, minimizing costs, and reducing carbon emissions. The lower layer uses DynaQ dynamic adjustment strategy based on MDP (Markov Decision Process) modeling to adapt to environmental changes; The scheduling rules are designed around the dual objectives of "workpiece selection machine allocation". There are three types of rules for workpiece selection, including priority for low completion, and three types of strategies for machine allocation, including efficiency priority. These are combined into nine complete rules and are based on six standardized state characteristics such as average processing completion rate and machine utilization rate for decision-making. The experiment is based on actual data from wheat planting areas, with constraints such as a water limit of 1200m ³/ha and a 15-day sowing cycle. Using adaptive genetic algorithm as a control, the optimal parameters are determined through orthogonal analysis (NIND for medium and large-scale problems is 90), and dynamic interference scenarios are introduced for verification. The results showed that compared with traditional NSGA-III, the Pareto frontier distribution index (spacing measure) of this model increased by 18.7%, the comprehensive satisfaction of the objective function reached 92.3%, the scheduling stability in dynamic environment improved by 34.5%, and the convergence speed within 100 iterations accelerated by 22%, fully demonstrating its efficiency and robustness, providing a new path for intelligent agricultural dynamic scheduling, and possessing both theoretical value and practical significance.

Open-winding permanent-magnet synchronous motors feature flexible control and a high fault-tolerance capability, making them widely used in high-reliability and high-power scenarios such as military equipment and electric locomotives. To address the issues that traditional model predictive control fails to balance, such as zero-sequence current suppression, system loss optimization and the reliance of weight parameter design on experience (with online optimization consuming excessive resources), this paper proposes an OW-PMSM MPC strategy for loss optimization and a weight design method based on a residual neural network. Specifically, the former strategy adds a zero-sequence current suppression term and a loss quantification term to the MPC cost function, enabling coordinated control of the two objectives; the latter establishes a mapping between weight parameters and motor performance via ResNet (which avoids the gradient vanishing problem in deep networks) and outputs optimal weight parameters offline to save online computing resources. Comparative experiments under two operating conditions show that the improved MPC strategy reduces system loss by 25%, while the ResNet-based weight design improves the performance of the drive system by 30%, fully verifying the effectiveness of the proposed methods.

ABSTRACT The pressure filtration operation is a key stage in coal preparation, where the moisture content of filter cake critically affects subsequent process regulation and product quality. Conventional moisture detection mainly relies on offline measurements, which suffer from poor timeliness and considerable errors, making them inadequate for precise on-site monitoring and control. To overcome these limitations, this study presents a soft measurement method for filter cake moisture based on a Double Regularized Broad Learning System (DR-BLS). Through mechanistic analysis, key influencing factors in the filtration process are identified, and a soft sensing framework is established. The proposed method incorporates a double regularization strategy into the Broad Learning System to improve generalization capability and robustness against noise. Moreover, an online learning mechanism is introduced to enhance adaptability to dynamic operating conditions, ensuring high-precision and real-time prediction of moisture content. Industrial field experiments demonstrate that the DR-BLS method significantly outperforms traditional soft measurement techniques in prediction accuracy and model stability, providing reliable and timely data support for the online optimization and intelligent control of pressure filtration processes in coal preparation plants.

Accelerating multi-energy system online optimization via integer state variable prediction with operation strategy learning

Distributed adaptive fault-tolerant cooperative control for fixed-wing UAVs with actuator faults and input constraints.

ABSTRACT Traditional data‐driven optimization methods using case‐based reasoning (CBR) rely on heuristic similarity matching and lack probabilistic rigor, especially in complex processes like fluid catalytic cracking (FCC) with high dimensionality and uncertainty. To address these challenges, a novel data‐driven framework that integrates compact posterior estimation with CBR is proposed. The method first identifies key variables affecting product yields through information‐theoretic dimensionality reduction. Optimal operating parameters are then inferred using a combination of K‐nearest neighbors for similarity matching and Markov Chain Monte Carlo sampling for probabilistic estimation. Industrial validation showed gasoline and total liquid yields increased by 7.31% and 6.94%, respectively, with coke yield reduced by 5.83%. This approach successfully improves computational efficiency and optimization accuracy in practical applications.

A Model-Based Approach for Online Optimization of Pneumatic Drives

Truncation dual ascending distributed online optimization of multiple virtual power plants based on energy state virtual queues

Caregivers of individuals living with dementia are at increased risk for social isolation, stress-related illnesses, and other adverse outcomes. Participation in a caregiver support group can provide connections to others in similar situations and to resources and strategies, all of which likely reduce adverse outcomes. Prior to the COVID-19 pandemic, caregiver support groups were primarily conducted in-person. With the proliferation of online support groups post-pandemic, we believe a framework for optimal online support group management and implementation is needed to prioritize caregiver safety and well-being. Due to the COVID-19 pandemic, the Boston-area frontotemporal dementia (FTD) caregiver peer support group shifted to meeting weekly via Zoom in March 2020. Trained support group facilitators (SGFs) attended every meeting to manage online meeting technology, promote adherence to group guidelines, and record meeting notes. New members were screened to confirm FTD caregiver status and approved for registration. Once registered, they were added to the group mailing list and received weekly emails with meeting links and follow-up resources. Since March 2020, 282 unique caregivers have registered for support group and 228 have attended at least one session. Attendance each year has increased from 426 group attendees in to 914 in 2024. Average weekly attendance has grown from 11 members per meeting in 2020 to 18 in 2024. Support group members attended from fourteen states in 2024, including individuals from more rural areas without local in-person support groups. Barriers to attending support group have diminished due to online access. Support group members have provided qualitative feedback about areas of support group that were most valuable to them, including receiving weekly follow-up emails, ease of technology, and sense of community with other members. The framework and guidelines developed by the SGFs for this support group create a safe and welcoming environment for caregivers to share all aspects of their caregiving journey. Providing caregivers of PLWD with a regular online support group with established guidelines for behavior can diminish the isolation that dementia caregivers often experience, promoting opportunity for high quality care for care recipients.

This comprehensive review examines the works of reinforcement learning (RL) in ship collision avoidance (SCA) from 2014 to the present, analyzing the methods designed for both single-agent and multi-agent collaborative paradigms. While prior research has demonstrated RL’s advantages in environmental adaptability, autonomous decision-making, and online optimization over traditional control methods, this study systematically addresses the algorithmic improvements, implementation challenges, and functional roles of RL techniques in SCA, such as Deep Q-Network (DQN), Proximal Policy Optimization (PPO), and Multi-Agent Reinforcement Learning (MARL). It also highlights how these technologies address critical challenges in SCA, including dynamic obstacle avoidance, compliance with Convention on the International Regulations for Preventing Collisions at Sea (COLREGs), and coordination in dense traffic scenarios, while underscoring persistent limitations such as idealized assumptions, scalability issues, and robustness in uncertain environments. Contributions include a structured analysis of recent technological evolution, and a Large Language Model (LLM) based hierarchical architecture integrating perception, communication, decision-making, and execution layers for future SCA systems, which prioritizes the development of scalable, adaptive frameworks that ensure robust and compliant autonomous navigation in complex, real-world maritime environments.