Search Results (1,119)

Photovoltaic power conversion systems require maximum power point tracking (MPPT) strategies capable of fast dynamic response with low computational burden, while remaining reliable under variable environmental conditions. While neural-network-based methods have been widely investigated, their practical deployment is often limited by the availability of representative training data and by the gap between offline algorithm development and real-time converter-level validation. This paper presents a reproducible hardware-in-the-loop workflow for the development and assessment of a lightweight neural MPPT controller applied to a photovoltaic system with a two-phase interleaved boost converter. The proposed approach generates a large synthetic training dataset using the single-diode photovoltaic model, leveraging only measurable quantities (PV voltage, PV current, and module temperature) as neural network inputs. The trained network estimates the voltage and current corresponding to the maximum power point, while a proportional-integral controller drives the converter toward the predicted operating point. The trained network is deployed on an STM32 microcontroller interfaced with the Typhoon HIL platform, allowing its real-time behavior to be tested against the emulated system. The measured neural MPPT execution time on the microcontroller is around 65 $\mathsf{\mu }$s, with an overall CPU occupancy of nearly 4%, considering the PI controller stage. The implemented setup reproduces the photovoltaic generator, converter dynamics, switching behavior, and realistic irradiance and temperature profiles under repeatable real-time conditions. The interleaved boost architecture also reduces input current ripple and distributes current stress, making the setup suitable for medium-power photovoltaic applications. The main contribution of this work lies in the integrated modeling, training, control, and hardware-in-the-loop validation procedure, supporting the implementation of neural MPPT strategies. Full article

When inverter switching frequency is constrained, PMSM drives operating at relatively high fundamental electrical frequencies may exhibit a low switching-to-fundamental-frequency ratio (low carrier ratio), resulting in fewer voltage vector updates per fundamental cycle and increased current ripple and low-order harmonics. To address the difficulty of synchronizing offline selective harmonic elimination PWM (SHEPWM) switching angles with a field-oriented control (FOC) current loop, this paper proposes an angle-synchronous FOC-SHEPWM implementation based on electrical angle partitioning. A quarter-wave-symmetric SHEPWM model is first established, and the switching angles are solved using Newton iteration and homotopy continuation. A Halton low-discrepancy initial value pool is then constructed. Combined with cumulative interval mapping, admissibility screening, and continuity assessment, it yields switching angle trajectories suitable for closed-loop look-up table implementation and extends the high-modulation-index range toward six-step operation. For online implementation, the ePWM period is updated according to the electrical angular speed, and the offline angles are mapped to intra-partition compare values. Counter-zero sampling, delay angle compensation, dynamic period correction, and fundamental current extraction are integrated to realize synchronized closed-loop pulse generation. The simulation and experimental results demonstrate the stable operation of the SHEPWM-N3, SHEPWM-N5, and SHEPWM-N7 patterns. At equal numbers of switching events, the proposed patterns exhibit lower low-order harmonic content and current total harmonic distortion than ASVPWM, while the fundamental current extraction and dynamic period correction methods reduce d-q-axis current ripple and partition synchronization error, respectively. Full article

Robust control of nonlinear dynamic systems remains challenging because parametric uncertainty and external disturbances can significantly degrade tracking accuracy and transient performance. This study proposes a Quantum-Inspired Photon–Spin Control Framework (QPSCF) that integrates probabilistic state representation and interference-inspired decision-making directly into the online feedback control process while remaining fully executable on classical computing platforms. Unlike quantum-inspired approaches primarily used for offline controller tuning or heuristic optimization, the proposed framework represents multiple candidate operating states probabilistically and adaptively evaluates competing control actions according to current process conditions. The QPSCF was evaluated on a nonlinear benchmark system under parameter variations of up to ±15% and a 10% external disturbance and compared with conventional PID and Mamdani fuzzy controllers under identical simulation conditions. The proposed controller achieved a settling time of 40.70 s, an overshoot of 0.15%, an RMSE of 4.83, an IAE of 121.54, and an ISE of 1652.41. Compared with PID control, QPSCF reduced RMSE, IAE, and ISE by 19.6%, 29.6%, and 41.9%, respectively. It also reduced the maximum disturbance-induced deviation from 1.80 °C to 0.55 °C and the recovery time from 20.30 s to 5.90 s. These results demonstrate that direct probabilistic decision-making within the feedback loop can improve tracking accuracy and disturbance rejection in nonlinear systems under uncertainty. Full article

Automated vision-based structural health monitoring (SHM) of catenary support components is critical for ensuring railway operational safety. However, achieving reliable structural damage identification in practical engineering scenarios is hindered by complex environmental and operational variations (EOVs) in aerial imagery, the multi-scale nature of structural degradation, and the scarcity of damage samples. To address these SHM challenges, this paper proposes a high-precision unsupervised damage detection framework named Mask-Guided Asymmetrical Flow (MGAF). First, to mitigate the impact of EOVs, a SAM-guided preprocessing strategy is introduced to explicitly suppress background interference and extract effective structural Regions of Interest (ROIs). Second, an Asymmetrical Parallel Flow architecture is designed to balance damage detection sensitivity and processing latency. By optimizing the flow depth for high-resolution features, this architecture prevents overfitting to high-frequency environmental noise while preserving deep semantic information. Furthermore, a Cross-Scale Feature Fusion (CSFF) module is developed to enhance the detection capability for early-stage structural damages (e.g., fatigue micro-cracks and fastener looseness) by integrating global structural semantics with local textural damage indicators. Finally, a Morphological Edge Suppression (MES) mechanism is employed to eliminate boundary artifacts, thereby reducing the false alarm rate in pixel-level damage localization. Extensive experiments on the self-constructed CSCUD dataset demonstrate that the proposed system achieves competitive anomaly detection performance compared with representative unsupervised methods with an Image-level AUROC of 98.9% and a Pixel-level AP of 40.1%. During online inference, MGAF achieves a processing time of 0.077 s per frame on an NVIDIA RTX 4090 GPU, excluding the offline GSI-Net localization and SAM-based structural ROI extraction stages. This demonstrates the efficiency of the proposed anomaly detection module for practical railway inspection scenarios. Additional experiments on selected categories from the MVTec AD benchmark provide preliminary evidence of the transferability of MGAF to visually similar industrial anomaly detection scenarios. Full article

Connecting IT and OT in manufacturing has become a core requirement for realizing Industry 4.0 smart factory architectures. This paper reports on the implementation of a smart factory built on the Festo CP Factory platform, covering three main contributions: IT/OT network architecture, MES-driven production control, and offline production optimization using Discrete Event Simulation (DES). The system comprises four stations—an Automatic Storage and Retrieval System (ASRS), a Robot Assembly Station (RASS) with a Mitsubishi RV-4FRL six-axis robot, a Magazine Station, and a Muscle Press—linked through a LINEAR conveyor network. A layered communication stack was deployed: PROFINET at the field level, OPC-UA for MES4 connectivity, and MQTT via Node-RED for IoT monitoring. An offline DES model was built in Siemens Tecnomatix Plant Simulation Environment (version 2302; Siemens Digital Industries Software, Munich, Germany) to study how carrier count affects throughput and station utilization across three carrier configurations. Station processing times were taken directly from the physical line: 86 s for Robot Assembly, 15 s for ASRS, 4 s for Magazine, and 6 s for Muscle Press. The Robot Assembly Station reaches near-full utilization (94.49%, event-based) at three carriers, confirming it as the system bottleneck. Two carriers offer the most practical operating point, delivering 35.8 pallets per hour while keeping the bottleneck station below full saturation. Fuse count also affects cycle time. Zero-fuse assemblies complete production up to 9.3% faster than two-fuse runs due to the shorter Robot Assembly cycle. Testing on the physical CP Factory to validate the developed simulation model covered 27 configurations combining three carrier counts, three batch sizes, and three fuse levels, with simulation errors ranging from 2.9% to 10.0%. The mean error stays close to commonly cited manufacturing DES validation thresholds at low carrier counts, reaching 4.78% at one carrier. It rises at higher carrier counts, reaching 8.64% at three carriers. This progression is consistent with carrier-queuing effects that intensify with carrier count. The results confirm that complete Industry 4.0 implementation is achievable on a modular CP Factory platform and provide a practical reference for similar deployments. Full article

With the transition of China’s energy structure toward clean and low-carbon development, the rapidly expanding deployment of renewable energy resources, particularly wind and photovoltaic power, has continued to increase. However, the volatility and intermittency of renewable energy output can disturb the dynamic active-power symmetry between the generation and load sides, leading to frequency deviations and posing significant challenges to the secure and stable operation of power grids. In grid-forming wind-storage systems, the grid-side inverter adopts virtual synchronous generator (VSG) control to reproduce the dynamic behavior of synchronous machines, thereby enabling virtual inertia, damping, and active-power support to mitigate source–load power asymmetry and facilitate the restoration of active-power symmetry following disturbances. Nevertheless, most existing tuning methods for VSG active-power control parameters have difficulty simultaneously satisfied small-signal stability requirements and achieved satisfactory frequency-support performance, thereby limiting the capability of VSGs to restore active-power symmetry. To address this issue, this paper proposes an offline–online coordinated tuning method for VSG active-power parameters. By incorporating small-signal stability constraints derived from eigenvalue and root-locus analysis together with frequency-support constraints, a feasible parameter region is first obtained. Within this region, offline optimization is performed with the minimization of the peak frequency deviation as the objective to determine the nominal parameters, and an online tuning strategy is further introduced to adaptively adjust the active-power droop coefficient according to the disturbance condition. Finally, simulation results show that, for all investigated load disturbances ranging from 0.08 to 0.14 pu, the proposed tuning method increases the frequency nadir by at least 0.038 Hz and by up to 0.064 Hz compared with the original parameter set. Meanwhile, the maximum RoCoF is reduced by at least 66.3%, with a maximum reduction of 73.8%. Under the 0.14 pu disturbance, the online tuning strategy limits the VSG primary frequency regulation power to 0.0178 pu, representing a reduction of approximately 16.0% from the 0.0212 pu obtained with the fixed offline-optimized parameters. Consequently, the prescribed capacity limit of 0.02 pu is satisfied, with an available power margin of approximately 11.0%. These results demonstrate that the proposed method improves the system frequency response and enables effective utilization of the VSG primary frequency regulation capability while satisfying the small-signal stability requirements. Full article

The automation of material receiving processes is central to ensuring traceability, process control, and data consistency in warehouse and logistics environments. This paper presents an integrated automation framework combining on-device optical character recognition (OCR), offline-first data capture, and Discrete Event System (DES) formal verification to address inbound registration failures at 3PL receiving docks. The framework comprises an Android mobile application for dock-side registration and a centralised web platform for validation and monitoring. The DES model, represented as a finite automaton, was applied pre-deployment to verify behavioural completeness, identifying three design gaps before operation began. An eight-week deployment at a Third-Party Logistics (3PL) facility showed statistically significant improvements over a two-week manual baseline (percentage points, p.p.): registration completeness +27.4 p.p. (${\chi }^2=48.0$, $p<0.001$), information gap rate −17.2 p.p. (${\chi }^2=27.1$, $p<0.001$), average registration time −56.3% ($t\left(201\right)=20.0$, $p<0.001$), and traceability coverage +57.0 p.p. (reaching 100%). The study contributes a reference architecture for inbound monitoring systems and demonstrates the practical value of DES-based pre-deployment verification. Full article

In multi-AUV collaborative underwater coverage operations, the fundamental principle of load balancing is temporally rather than spatially defined. Traditional partition schemes that equalize area or grid counts suffer from a critical fallacy: they assume a linear mapping from geometry to working time, which collapses in the presence of irregular obstacles. This paper demonstrates that temporal load imbalance causes some AUVs to finish prematurely and remain idle—consuming power and fighting currents while their counterparts struggle with topologically complex sub-regions. To address this, we propose a dual-layer nested Particle Swarm Optimization (PSO) framework for Voronoi partitioning, where the outer layer coarsely initializes grid counts, while the inner layer directly minimizes the makespan and time variance derived from actual BINN-simulated coverage paths. This two-stage PSO architecture effectively resolves the nonlinear mismatch between geometric partitioning and real operation time, redefining load balancing from a geometrical abstraction to a temporally grounded, operationally relevant metric. Furthermore, we rigorously distinguish the intrinsic nature of coverage path planning (CPP) from Traveling Salesman Problem (TSP)-based point routing—the latter generates discontinuous, sharp-turning trajectories that violate the kinematic constraints of side-scan sonar payloads and cause critical data gaps. The proposed architecture yields a standard deviation of mission times that is an order of magnitude lower than area-based heuristics, while maintaining kinematically feasible continuous sweeps. Crucially, we explicitly delineate the operational boundary of this framework: it is purpose-built for static, pre-surveyed environments where offline computational overhead (approximately 8 min) is a justifiable investment against a 3 h optimal field execution. Full article

Finite-control-set model predictive control provides excellent torque–speed regulation for switched reluctance motor drives but requires an online combinatorial search at every control instant, making low-cost embedded implementation challenging. This article investigates whether offline reinforcement learning can distill policies from multiple classical controllers into a single feedforward policy requiring neither online optimization nor controller gain tuning. A replay buffer is populated with trajectories generated by three demonstrators—hysteresis current control, proportional–integral control with pulse-width modulation, and finite-control-set model predictive control—using a finite-element model of a four-phase 8/6 switched reluctance machine parameterized from measurements of the physical drive. An implicit Q-learning agent then learns a control policy without evaluating actions outside the offline dataset. The central finding is that demonstration diversity governs the stability of offline reinforcement learning on this problem: policies trained from a single demonstrator experience early mode collapse in all fifteen runs, whereas two- or three-demonstrator datasets converge stably in all fifteen. Behavior cloning trained on the identical buffer, split, architecture, and deployed controller provides the reference point for interpreting this result. It matches the offline RL policy on torque quality and improves on its speed regulation, exhibiting none of the seed-to-seed fragility seen at no load while requiring roughly $8\%$ more switching transitions. The stability requirement therefore appears to be a property of the advantage-weighted offline RL objective rather than the control task, and the measured benefit of that objective on this problem is confined to switching effort. We report this rather than claim a broader advantage. The characterization of the distilled controller shows that it generalizes to operating points that are not included in the training dataset, gains nothing systematic beyond approximately 60% of the replay buffer, remains insensitive to $\pm 20\%$ perturbations of all reward weights, and degrades gracefully under measurement noise while the current mask enforces the peak-current constraint throughout. A deployment analysis shows that the 18,432 multiply–accumulate policy meets a $50\mathsf{\mu }$s control period in its existing form at a measured cost of about $2\%$ in torque ripple. All the results are simulation-based on a finite-element model parameterized from a physical machine. Full article

Large language model (LLM) agents can perform prognostics and health management (PHM) tasks such as bearing fault diagnosis, but most published systems rely on large, paid, cloud-hosted models that a small or medium enterprise (SME) cannot self-host. The diagnostic accuracy that can be achieved on free, offline, commodity hardware is a practical question. Using an execution-based evaluation (Pass@1, Pass all 3, macro F1), this paper benchmarks eight small, publicly accessible, locally runnable LLMs (1B–9B parameters, via Ollama) on vibration-derived features for three-class bearing fault diagnosis (Healthy, Outer race, and Inner race). The proposed work is evaluated on three independent datasets, Paderborn, CWRU, and HUST, across a 0–5-shot ablation and at two decoding temperatures to separate accuracy from reliability. The findings replicate across all three datasets, namely, a model-capability gate that only models near 7B parameters and above clears the majority-class floor; few-shot prompting is non-monotonic; accuracy and reliability are distinct axes; and a single free model, gemma2:9b (5.4 GB), is the most accurate and among the most reliable, with no task-specific training. We further propose ensemble agreement gating, a training-free reliability rule that withholds predictions when several free local models disagree, raising accuracy on the answered subset (e.g., 0.77 → 0.87 on Paderborn). In a head-to-head on identical prompts and data, the free local models match a current frontier model in the settings tested at zero cost and fully offline, and we characterize where such a training-free local approach is and is not appropriate for resource-constrained operators of rotating machinery. On the same features, however, a simple supervised baseline such as logistic regression, and even an untrained physics rule, outperform all eight LLMs, so we present this as a cautionary benchmark: the contribution of the free local approach is training-free deployment and a reliability gating rule, not classification accuracy. Full article

Low-switching-frequency operation is essential for high-power modular multilevel converters (MMCs), but it makes the simultaneous achievement of low harmonics, fast transient response, and reliable internal control more difficult. Conventional lookup-table selective harmonic elimination PWM (SHEPWM) preserves steady-state harmonic optimization, yet its fundamental-period pattern update limits transient flexibility; predictive pulse-pattern methods improve current or flux tracking, but do not directly coordinate the MMC execution chain from arm-level pulse displacement to physical submodule gating. This paper proposes a coordinated switching-event control scheme for SHEPWM-based MMCs. The offline harmonic-optimized pattern is used as the steady-state backbone, while selected time-stamped events are corrected online through a virtual-flux formulation. The same events are then processed by causal pulse-edge compensation, paired upper/lower-arm displacement for circulating-current suppression, and threshold-based asynchronous submodule scheduling. Simulations of startup, active-power steps, and power-flow reversal show that more than 90% of the virtual-flux error is compensated within approximately 5 ms. Device-level co-simulation reduces current THD from 1.5881% to 0.702%, and paired-event control reduces arm-current THD from 21.600% to 4.9266%. At rated steady state, the method achieves 1.25% grid-current THD with a 50 Hz average submodule switching frequency, supporting low-loss MMC operation without abandoning SHEPWM harmonic optimization. Full article

The Internet of Drones (IoD) enables efficient communications among unmanned aerial vehicles (UAV) via wireless links. Restricted by limited onboard computing resources and complicated operating environments, such a network is vulnerable to external or internal security threats. It is therefore urgent to deploy an effective authentication mechanism to isolate malicious attackers prior to UAV communications. Nevertheless, the conventional identity-based IoD authentication schemes must account for total computation and communication overhead, regardless of the effective instant of authentication messages. In this work, we consider the offline precomputation in the authentication phase before the message is available and add cloud servers as an extra component in IoD to assist limited onboard resources to share partial authentication calculation. As a result, this work proposes a new cloud- and offline-computing-assisted authentication framework, known as COAu-IoD, to provide lightweight communication between a UAV and ground station by transferring heavy computation overhead to the offline phase. We give a security model for this authentication framework and provide an efficient construction based on this framework and a known identity-based signature scheme. What is more, we give its security proof and evaluate its performance using the standard and recent state-of-the-art IoD authentication schemes. The results demonstrate that our COAu-IoD scheme achieves lower online computation and communication overhead for UAVs, at the cost of acceptable offline precomputation overhead. Full article

Wastewater treatment plants (WWTPs) are crucial infrastructure. Continuous monitoring of plant conditions using sensor technology has enabled real-time monitoring, forecasting and control. WWTP processes, namely, those occurring in biological reactors, exhibit highly non-linear relationships, multiscale seasonality, and dependency on hydrodynamics and sensor status. WWTP operators may benefit from automated detection of changepoints, i.e., points in time such that WWTP parameters clearly differ before and after. A challenge is the high cost of manual annotation of changepoints for training and evaluation of models. In this study we contribute an “offline-to-online” framework for annotation and detection of changepoints. We first detect changepoints in WWTP data a posteriori (offline) using a dynamic programming method with several cost functions, filtering for high-consensus points. These are reviewed by the authors and verified independently by WWTP experts. This algorithm-assisted annotation reduces the expert annotation burden. The verified labels are then used to test a suite of real-time (online) changepoint detection methods. Full article

To enhance the fuel economy and operational adaptability of hybrid tractors under variable plowing loads, this paper proposes a slip-ratio-aware equivalent consumption minimization strategy (ECMS) calibrated via dynamic programming (DP). A resistance–slip ratio prediction model was first identified using plowing resistance and slip ratio data collected from soil-bin tests. The predicted slip ratio was integrated into the demand power model to quantify slip-induced traction losses. Offline DP was subsequently applied to generate globally optimized power split trajectories and establish a baseline equivalence-factor map indexed by plowing resistance level and battery state of charge (SOC). For real-time operation, the equivalence factor is dynamically adjusted via SOC feedback and normalized slip ratio deviation, enabling coordinated power distribution among the engine, MG1, and MG2. Powertrain bench tests were conducted by reproducing variable plowing loads using a dynamometer. The equivalent plowing resistance was calculated from measured load torque, and the corresponding slip ratio was estimated using the identified prediction model. Compared with A-ECMS, the proposed strategy reduced equivalent fuel consumption by 14.02% in simulation and 7.33% in bench tests. The proportion of engine operation in the high-efficiency region increased from 61% to 80% in simulation and from 65% to 77% in the bench test, while the corresponding proportion for the electric motors increased from 87% to 92% and from 88% to 90%, respectively. The SOC deviation decreased from 3.03% to 2.26% in simulation and from 3.07% to 2.43% in the bench test. These results demonstrate that the proposed strategy improves fuel economy, SOC regulation, and component operating efficiency under variable plowing loads. Full article

Three-view 3D Gaussian Splatting is prone to insufficient geometric constraints, overfitting to sparse color observations, and uneven allocation of Gaussian representation capacity, leading to structural distortions and local artifacts. We propose a DNGaussian-based framework that combines branch-decoupled regularization with edge-aware selective densification. Monocular relative depth maps are generated offline using Depth Anything V2 with a ViT-S backbone and incorporated as relative geometric supervision. DropGaussian is applied only to the color-rendering branch, whereas Hard-depth and Soft-depth updates use the complete Gaussian set, preventing stochastic suppression from directly perturbing depth supervision. Sobel responses from the three training views are accumulated as cross-view Gaussian edge statistics and used solely to reduce the effective screen-space gradient threshold for Gaussians with high cumulative mean edge scores; cloning and splitting follow the base DNGaussian operations. Relative to DNGaussian, the proposed method improves mean PSNR by 1.50 dB on LLFF and 1.01 dB on DTU, while maintaining a comparable mean Gaussian count on LLFF. These results demonstrate its effectiveness for three-view novel view synthesis. Full article