Search Results (1,125)

Wind turbine clutter (WTC) can severely degrade the quality of meteorological radar data, particularly under complex weather conditions. Because WTC echoes can substantially overlap with precipitation echoes, they may introduce considerable errors into weather identification and forecasting. To address this issue, a deep learning model termed WTC-CNXGRN is proposed for automatic WTC detection. The model is built on the ConvNeXtV2 architecture and is designed to address characteristic WTC features, including unstable spatial structure, pronounced local texture variations, and complex multi-channel feature coupling. Multi-parameter dual-polarization weather radar data are used as input, while local statistical features and weighted polarimetric features are incorporated to strengthen the representation of spatial echo variations. In addition, a Global Response Normalization (GRN) mechanism is integrated into the network to adaptively recalibrate channel-wise responses, thereby enhancing the representation of critical discriminative features. On the test dataset, the proposed model achieves an accuracy of 98.74%, a precision of 98.41%, a probability of detection (POD) of 98.34%, and a false alarm ratio (FAR) of 1.59%, achieving overall better performance than the evaluated baseline models. Benefiting from its efficient architecture and improved detection strategy, the model provides low computational cost and fast inference, making it suitable for near-real-time operational applications. Qualitative tests using radar data from the Nantong region further provide preliminary evidence of the model’s cross-site applicability. Full article

In design applications, simulations enable rapid iterations and adjustments to device architecture, reducing the need for physical prototyping. The simulation of piezoelectric devices has traditionally relied on the Finite Element Method (FEM). Physics-Informed Neural Networks (PINNs) offer an alternative based on governing equations without requiring labeled solution data. This study develops a mixed PINN architecture for the static direct and converse piezoelectric responses of a polyvinylidene fluoride (PVDF) bimorph cantilever. The network predicts two mechanical displacements, electric potential, three stress components, and two electric-displacement components. The methodology integrates the piezoelectric governing equations into a first-order loss formulation. The models are evaluated against coupled FEM solutions. For the converse effect, the relative L² errors in horizontal displacement, vertical displacement, and electric potential are 0.107, 0.150, and 0.045, respectively. For the direct effect, they are 0.117, 0.153, and 0.067. The mixed configurations give lower displacement errors than the tested networks predicting only displacement and electric potential. However, they also use hard traction enforcement in the direct problem and mechanical gradient routing in the converse problem, so the improvement cannot be attributed to the additional outputs alone. Discrepancies between independently predicted stresses and those reconstructed from displacement and potential derivatives from the PINN reveal incomplete physical consistency. These results support approximate displacement and potential prediction while identifying constitutive consistency as a remaining limitation. Full article

Offshore wind turbines equipped with full-power converters (FPCs) require coordinated control of power tracking, electrical safety, and mechanical load mitigation. Nonlinear model predictive control (NMPC) provides a systematic framework for this multi-objective problem; however, standard implementations solve the nonlinear program to full convergence at every control step, incurring computational times incompatible with the 10–100 ms control periods of commercial turbines. Existing formulations also commonly omit electrical hard constraints and rely on expensive lidar for wind speed feedforward. To address these limitations, this paper proposes a real-time iteration NMPC (RTI-NMPC) framework in which each control step performs a truncated real-time iteration, i.e., a single inexact SQP step realized by at most five interior-point (IPOPT) Newton iterations on the parametric NLP, combined with warm-start initialization and a solution-shift strategy. Torque command amplitude bounds and a DC-link voltage deadband are embedded as hard bounds on the predicted trajectory within the OCP, while the torque rate limit is enforced through a dual mechanism consisting of a quadratic penalty in the cost function and a ±15 kN·m/s hard saturation at the solver output. A second-order autoregressive predictor supplies wind speed feedforward over a 2 s horizon, and the power-tracking weight is adaptively adjusted according to the prediction confidence. Ablation experiments confirm that the warm-start solution-shift mechanism is decisive for closed-loop performance: removing it degrades the power-tracking RMSE by up to 370% under grid load-drop transients, collapsing to the level of the higher-iteration-budget standard NMPC, while the adaptive weighting is shown to be intrinsically coupled to the predictor and remains inactive under persistence forecasting. Comparative simulations under four operating scenarios show that the proposed controller achieves an average per-step computation time of 3.796–4.313 ms, approximately one order of magnitude faster than standard NMPC, thereby satisfying the real-time requirement within the present simulation setting. Under a grid load-drop scenario, the root-mean-square power-tracking error is reduced by 78.1% and 78.5% relative to a PI controller and standard NMPC, respectively; under model mismatch, the corresponding reductions are 18.0% and 17.5%. This advantage is scenario- and model mismatch-dependent and partly arises from reduced commitment to an imperfect internal prediction model, rather than the general superiority of truncated optimization over a more fully solved NMPC. The torque rate remains within ±15 kN·m/s in all scenarios. It is noted that the DC-link hard bounds apply to the predicted trajectory within the OCP, while the actual plant voltage is subject to the simplified model dynamics. The results demonstrate the computational feasibility and engineering potential of the electrically constrained, lidar-free RTI-NMPC for coordinated power-tracking and electrical constraint management of offshore FPC wind turbines. Full article

Transient pore-air pressure in gas-loaded cover soils is difficult to observe between monitoring depths. We develop a coupled water–air inverse physics-informed neural network as a numerical proof of concept. Under a pre-calibrated constitutive model, three numerically sampled depths supply pore-air pressure, pore-water pressure, and saturation targets. An independent air-pressure field represents overpressure; the coupled balances constrain the joint state. On a one-dimensional same-equation benchmark, the full-window gas-pressure error is 0.0463. The data-only ablation reaches 0.0043, while the coupled residuals improve water pressure, saturation, and front monotonicity. Before the 495 s reference injection-base zero-stress crossing, gas-pressure error is 0.0448; on cells with positive reference effective stress, the stress-reduction error is 0.0450. Later states test numerical tracking under overload. The plane-model case fits two-dimensional FLAC2D training series with a one-dimensional residual omitting lateral transport, giving a responding-depth excess-overpressure error of 0.186. A residual-free line-M test evaluates architecture-only interpolation at 11 held-out depths. The configuration-specific Bishop post-process yields zero-to-seven-minute onset times and a 43% minimum-stress spread across three effective-stress parameter forms. These numerical fields support subsequent mechanical interpretation; physical validation requires measured interior states, and stability assessment requires a mechanical model. Full article

Downhole dynamic compaction (DDC) improves soft ground by delivering high-energy impacts inside boreholes, where repeated hammer blows compact soil–rock mixture backfills and exert dynamic loading on surrounding soil. However, conventional penetration records cannot reveal the internal stress evolution, deformation, and energy transfer mechanisms that govern pile formation. This study established a three-dimensional coupled continuum–discrete numerical model, in which the native ground was simulated by continuum zones, and the in-hole soil–rock mixture backfill, together with the hammer, was simulated by discrete particles. The model was validated against a field test with three staged backfilling operations and nine field blows, reproducing the measured penetration attenuation with a maximum single-blow deviation of 6 cm and a cumulative relative error of 5.8%. The results indicated that the first blow after each filling mainly caused backfill rearrangement and borehole bottom compression, while later filling and tamping increased lateral deformation around the upper backfill, with limited additional movement at depth. Vertical compression predominated beneath the borehole bottom, whereas lateral compression predominated at the off-axis monitoring point. The displacement and plastic-work distributions supported a conceptual four-zone interpretation of variable-section pile formation, including upper contact expansion, intermediate radial compression, lower backfill expansion and shallow disturbance. Full article

In a global landscape characterized by interconnected risks and the deepening of digital transformation, understanding the impact and mechanisms of artificial intelligence on social–ecological resilience is crucial for advancing resilience governance and the green transition. This paper utilizes panel data from 30 provinces in China (2013–2023) to construct a comprehensive index of social–ecological resilience and artificial intelligence. It employs two-stage fixed-effects models, mediation effect models, and robustness and endogeneity tests to systematically examine the impact of AI and regional heterogeneity. The results demonstrate that: first, AI significantly enhances social–ecological resilience, and this conclusion remains robust after controlling for measurement errors, trimming, and instrumental variable estimation; second, employment structure and innovation levels play a partial mediating role in the influence of AI on social–ecological resilience; and third, the promoting effect of artificial intelligence on social–ecological system resilience exhibits a regional heterogeneity pattern: strongest in the central region, followed by the western region, and weakest in the eastern region. The study highlights that AI not only directly enhances resilience by improving resource allocation and governance responsiveness, but also indirectly optimizes the social–ecological coupling through employment restructuring and innovation diffusion. Therefore, efforts should be directed towards promoting AI-enabled resilience governance through digital capacity building, industrial-employment synergy, institutional innovation, and infrastructure improvements. Full article

Achieving high-precision and highly adaptive leg joint control is a formidable challenge for hydraulic quadruped robots, owing to their dynamic uncertainties, nonlinearities, and frequent contact events. This paper proposes an event-based active compliance control strategy for hydraulic quadruped robots, which integrates the extended state observer (ESO) based robust integral of the sign of the error (RISE) inner-loop joint trajectory tracking controller and the event-based admittance outer-loop force controller with no contact force measurement. In the inner loop, ESOs are employed to estimate and compensate for the difficult-to-model dynamic coupling characteristics of the joints, while the RISE controller suppresses the residual disturbance compensation errors, thereby achieving asymptotic tracking performance. In the outer loop, a generalized momentum observer (GMO) compensated by neural networks is adopted to estimate contact forces, avoiding the installation of end-effector force sensors. Meanwhile, a contact detection mechanism is introduced to trigger admittance control upon unintended contact, thereby buffering the contact forces. Demonstrated through simulations, the proposed algorithm enhances joint position tracking accuracy, effectively mitigates impact forces from unintended contact, and improves the stability of robot locomotion. Full article

This paper presents a hybrid flux-frame high-frequency (HF) injection and model reference adaptive control (MRAC)-based sensorless rotor position and speed estimation scheme for advanced direct torque-controlled interior permanent magnet synchronous motor (IPMSM) drives. Unlike conventional stationary-frame (α-β) HF-injection methods, the proposed technique demodulates the injected high-frequency current directly in the stator-flux reference frame, yielding a saliency signal with reduced cross-coupling and improved estimation fidelity at zero and low speed. At medium-to-high speed, an adaptive MRAC estimator, driven by the stator-current error, provides back-EMF-based speed and angle estimation. A speed-dependent gating mechanism blends the two estimators, avoiding the discontinuous angle transitions typically introduced by hard-switched hybrid observers. The complete estimation and control algorithm, including a constant-switching-frequency torque regulator, is implemented and validated through processor-in-the-loop (PIL) co-simulation on a Texas Instruments LAUNCHXL-F280039C DSP-based microcontroller. Simulation and PIL results confirm the practical feasibility of the proposed hybrid estimation scheme, which offers reduced implementation complexity compared to Field-Oriented Control (FOC)-based sensorless techniques predominantly reported in the literature, since direct torque control (DTC) inherently avoids continuous coordinate rotation and current–vector modulation. Successful validation on the single-core, cost-effective LAUNCHXL-F280039C microcontroller, compared to higher-end dual-core DSP platforms (e.g., the F2837xD series) typically used in sensorless drive research, further demonstrates the suitability of the proposed approach for low-cost, resource-constrained embedded motor drive implementations. Full article

We formulate a chain of damped harmonic oscillators as a mechanical analogy for a single-lane vehicle platoon and show that it reproduces, and yields analytical insight into, the congestion patterns of car-following traffic. Each follower is treated as a point mass m coupled to its leader by a spring of stiffness $k_2$ (spacing-error feedback) and a dashpot of coefficient c (relative-speed feedback), with an additional optimal-velocity (OV) force $k_1[V\left(\Delta x\right)-v]$ acting as a non-conservative driving input. The resulting second-order equation is a damped-oscillator chain with headway-tunable effective stiffness $K\left(h\right)=k_1{V}^{\prime }\left(h\right)-k_2$, for which the local damping ratio partitions the parameter plane into an active (negatively damped) regime, an overdamped regime, and an underdamped window that supports ringing oscillations. Long-wavelength string-stability analysis yields a closed-form stable band for the OV slope, whose boundaries trace a tongue in the parameter plane. A weakly nonlinear multiple-scale analysis derives amplitude-dependent frequency shifts, centre drift, and nonlinear envelope decay, and recovers a double-well potential that saturates density-wave growth at a finite amplitude. Ring road simulations validate the predictions on two levels: a systematic numerical sweep shows that the zero-growth contour coincides with the analytic band along the whole tongue, and unbiased cycle-resolved measurements follow the analytic amplitude–frequency curve. Although developed for traffic, the construction is generic for dissipative coupled-oscillator chains with non-monotone driving laws. Full article

Natural fractures play a key role in controlling fluid flow and heat extraction in carbonate geothermal reservoirs. Due to the high degree of rock fragmentation and strong seismic wave absorption at the top of the carbonate formation, it is difficult to identify the distribution of natural fractures inside the carbonate formation using conventional seismic methods. Additionally, the heterogeneous distribution of permeability and porosity in the geothermal reservoir remains a challenge for the numerical simulation of the geothermal production process. To overcome these difficulties, this study uses the damage mechanics approach to characterize the distribution of natural fractures generated by orogeny in the Wumishan Formation of the JR block. Based on the derived damage field, a finite element model for the geothermal reservoir with spatially heterogeneous porosity was constructed, and empirical relationships between damage variable values and values of permeability of the fractured geothermal reservoir were established. A thermo-hydro-mechanical (THM)-coupled finite element model was then employed to evaluate well transmissibility and porous flow behavior under specified injection–production pressure conditions. The flow capacities of 18 wells were quantified and ranked, providing a basis for optimizing the injection–production well layout. Optimization of well locations and injection–production functions was conducted considering alternating well placement, total flow balance, and target well ratios (1:1 and 3:2). By adjusting well functions and locations, the total injection and production rates were balanced under the given pressure conditions, with the total injection and production capacities increased by 12.1% and 6.73%, respectively. The error tolerance used for convergence in numerical calculation is 2% for the THM model. Numerical results of temperature contours indicate the following: (1) There is no risk of thermal breakthrough with the planned design. (2) The temperature at the wellhead of production wells ranges from 92 °C to 105 °C. The results demonstrate that the combined application of damage mechanics and THM-coupled modeling provides an effective framework for accurately characterizing natural fractures and optimizing injection–production strategies in fractured carbonate geothermal reservoirs. Full article

To reduce the operational cost and risk of conventional downhole testing, this study presents an integrated well-test interpretation workflow that uses wellhead shut-in pressure data to characterize formation flow conditions in real time. Fluid mass- and heat-transfer equations are established for the wellbore, and a heat-conduction equation is established for the formation. Flash calculations are used to capture condensate phase changes inside the wellbore, while wave propagation and seepage mechanics are integrated into the formation flow model. This yields a coupled wellbore–formation flow system, and a corresponding numerical solution method is proposed according to the mathematical characteristics of the governing equations. Using this model, an interpretation method for wellhead-pressure testing in fracture-vuggy condensate-gas wells is established to determine critical reservoir parameters, including permeability, skin factor, cave location, and cave volume. The results show that converting measured wellhead pressure to bottomhole pressure after shut-in introduces an error during the initial shut-in period; this error gradually decreases with increasing shut-in time. Validation against a field well shows that the overall relative pressure error is less than 2%. Interpreted wellbore storage and near-wellbore cave volume exhibit errors within 5%, while formation permeability, distant cave volume, and cave location are essentially unaffected. The skin-factor difference is 0.55 (dimensionless; relative difference 20.1%), which should be interpreted with caution because skin factor is strongly controlled by early-time data. The model is adaptable to various wellbore types in fracture-vuggy condensate-gas reservoirs. Compared with traditional downhole testing, the proposed method substantially reduces operational hazards and expenses while providing accurate results, offering a low-risk, low-cost, and efficient technical solution for optimizing deep gas-field development. Full article

The design of stiffened thin-walled structures, commonly employed in die-cast automotive and aerospace components, is constrained by limited high-fidelity engineering data and the inherent difficulty of decoding mechanical mechanisms within complex, non-periodic topologies. Data are scarce and complex topologies are hard to interpret. We propose a framework that combines a VAE-GAN with an adaptive graph attention network (GAT). In this framework, the VAE acts as a structural prior. It maps irregular rib layouts into a smooth continuous latent manifold defined by geometric and topological features. This reduced mode collapse. Generated layouts remained valid and coherent, capturing implicit design rules without explicit rules. The augmented dataset also helped the GAT predictor capture spatial connectivity among rib nodes. The adaptive attention mechanism assigned weights to critical regions, improving interpretability across eight tasks. Experimental results showed that the GraphVAE-GAN framework generated novel stiffening rib layouts with 100% validity and structural uniqueness, achieving high-level coupling reconstruction among design variables. For maximum stress prediction under six loading conditions, the baseline model trained on 1184 real samples achieved R² = 0.74 on a fixed test set. After expanding the training set to 3000 samples, R² reached 0.91. Results were averaged over 5 runs with fixed seeds. Furthermore, the predictive model surpassed seven mainstream machine learning and deep learning benchmarks across all evaluated mechanical tasks, achieving a mean R² value above 0.93 across the eight distinct prediction tasks while yielding substantially lower maximum absolute errors compared to the best-performing baseline configurations. These findings suggest a viable strategy for navigating high-dimensional topological design spaces in lightweight structural components and offer a physically grounded trajectory for automating the design exploration of complex engineering systems under multi-objective performance constraints. Full article

Manual road inspections lack sufficient 3D geometric hazard data because of limited coverage and sampling frequency. This paper develops a hazard-oriented safety assessment framework integrating UAV LiDAR reconstruction, PointNet++ point-cloud segmentation, geometric hazard indicators, and traffic-mechanism analysis. Instead of merely extracting road objects, the framework converts segmented scenes into measurable metrics (lane width compression ratio, encroachment ratio, and marking degradation) and couples them with artificial-potential-field trajectory analysis, Lighthill–Whitham–Richards (LWR) traffic-wave theory, and lateral trajectory entropy. Validation with ground control points yielded a planar root mean square error (RMSE) of 0.055 m and an elevation RMSE of 0.025 m for the processed survey scenes. The five displayed class-level intersection over union (IoU) values yield an arithmetic mean of 87.88%, reported as 87.9% after rounding. The geometric hazard cases and traffic outputs are presented as analytical scenarios and sensitivity results; they are not claims of newly observed vehicle trajectories or lane-level field counts. The framework provides an interpretable basis for road maintenance, construction-zone management, and hazard early warning, while future deployment requires independent multi-site and trajectory-based validation. Full article

Magnetic regulation of ferrofluid transport can modulate heat and fluid motion without mechanical actuation, but the combined effect of field strength and an orientable internal obstacle remains insufficiently resolved. This study numerically examines laminar natural convection of a 1.3 vol% Fe₃O₄/SWCNT (Single-Walled Carbon Nanotube)–water hybrid ferrofluid in a square cavity containing a rotatable elliptical barrier under a vertical magnetic field. Eight barrier orientations (0–315° in 45° increments) are evaluated at Ra = 10³ and Ha = 20 and 80 using the finite element method in COMSOL Multiphysics. The governing formulation was corrected to use a symmetric viscous operator and the Lorentz term associated with a vertical magnetic field; the base-fluid Prandtl number calculated from the tabulated properties is 6.07. An independent, differentially heated square-cavity benchmark reproduced the reference values for maximum horizontal and vertical velocities and the average Nusselt number with relative errors below 0.76%. The computed fields show that barrier orientation controls the locations and signs of the velocity extrema and the pressure range, whereas the reported temperature profiles change comparatively little between the two Hartmann numbers. The 90° orientation gives the minimum reported Bejan number, while 45° gives the maximum. Because Be is a ratio, these extrema identify changes in the relative contribution of thermal irreversibility and do not, by themselves, establish a minimum or maximum of total entropy generation. The results demonstrate a coupled geometric–magnetic redistribution of the local thermohydrodynamic fields and identify 90° as the preferred orientation only under the Bejan-number criterion adopted here. Full article

This paper presents an adaptive digital twin framework for intelligent fault diagnosis and predictive health monitoring of a coupled multi-machine electric drive system. The experimental platform consists of two synchronous motors and a synchronous generator arranged in an interconnected electromechanical configuration and instrumented through the Lucas-Nülle laboratory platform. A physics-based digital representation is integrated with experimental measurements to reproduce the electrical and mechanical behavior of the drive system under different operating conditions. Physical-to-digital residuals are subsequently used for health assessment, fault detection, fault isolation, and intelligent classification. Experimental validation under five load conditions resulted in mean root-mean-square errors of 0.0391 N·m for torque, 0.1079 A for motor current, and 0.4555 V for motor voltage, with an overall mean normalized RMSE of 4.16%. The experimentally validated healthy-state digital twin was subsequently evaluated using model-based current-sensor, voltage-sensor, and torque-degradation scenarios. Channel-specific residual analysis successfully detected and isolated the abnormalities under investigation. Using digital twin residual features, an Ensemble classifier achieved a test accuracy of 98.61%, outperforming Artificial Neural Network (ANN) and Support Vector Machine (SVM) classifiers. In addition, a prognostic indicator was developed to characterize progressive degradation, with threshold crossings observed at simulated severities of 1.26%, 12.44%, and 17.37% for voltage-sensor deviation, torque degradation, and current-sensor deviation, respectively. The results demonstrate that integrating experimentally validated digital twin modeling, residual-based diagnostics, intelligent classification, and degradation monitoring provides a unified framework for condition assessment of coupled multi-machine electric-drive systems. Full article