Search Results (2,451)

Background/Objectives: Tetralogy of Fallot (TOF) is the most common cyanotic congenital heart defect, and cardiac computed tomography (CT) is increasingly central to its anatomical and pre-procedural assessment. Artificial-intelligence research in TOF is dominated by MRI; deep learning on cardiac CT in congenital heart disease exists but addresses multi-class diagnosis and segmentation, and the one binary TOF-versus-control CT study used slice-level validation without confounder control, and, to our knowledge, no CT study reports controlling the confounding intrinsic to a TOF-versus-control comparison. This confounding is structural: TOF is imaged predominantly in infancy, so a naive classifier can learn age, body size, and acquisition protocol rather than pathology. We develop and internally evaluate a confounder-matched, anatomy-guided deep-learning pipeline for TOF on cardiac CT. Methods: Contrast-enhanced cardiac CT from a single scanner was de-identified and restricted to one reconstruction (FC15 kernel, 0.5 mm), then matched 1:1 on age and sex, yielding 42 TOF and 42 controls (n = 84); controls were children imaged for suspected but excluded cardiovascular disease, so scan indication, unlike age and sex, was not matched. Standardized volumes were decomposed into four fixed sub-volumes positioned to approximate the components of the diagnostic tetrad: malalignment ventricular septal defect (VSD), overriding aorta, right-ventricular outflow tract (RVOT), and right-ventricular hypertrophy (RVH). Whether each sub-volume contains its named target was audited against independent physician region-of-interest annotations. Per region, a 2.5D transfer-learning classifier (ImageNet ResNet18) and a 3D CNN (DenseNet121) were trained with leak-free patient-level five-fold cross-validation and the branches fused. Optimism was assessed by repeated cross-validation and, for model selection, by nested cross-validation with the component subset and operating point chosen inside an inner loop. Discrimination was reported with bootstrap 95% confidence intervals (CIs); AUROCs were compared by DeLong test, with Benjamini–Hochberg correction applied to a seven-member family (the four within-component comparisons, two hybrid-versus-VSD contrasts, and hybrid versus whole-heart) and other comparisons reported uncorrected. Results: Matching removed the age difference (median 0.33 years, IQR 0.17–0.92 vs. 0.33, IQR 0.27–0.73; p = 0.86) with balanced sex (p = 1.00). The pre-specified four-component hybrid reached AUROC 0.829 (95% CI 0.74–0.91); the VSD region alone reached 0.828 (0.74–0.91), so the tetrad decomposition did not improve accuracy, and the containment audit shows it does not deliver the intended anatomical interpretability either. The 2.5D model exceeded the 3D CNN for every component (0.769–0.828 vs. 0.573–0.656; raw DeLong p = 0.007–0.037, Benjamini–Hochberg q up to 0.065 under a seven-member family, the weakest comparison (RVH) not surviving correction). Repeated cross-validation gave 0.811 ± 0.026 and nested cross-validation 0.787 ± 0.029; a stronger backbone with multi-phase data, handcrafted radiomics, and a large CT foundation model did not significantly improve on the matched pipeline. Grad-CAM maps were sensitive to both model weights and labels and superior to a centred-blob null in all eight comparisons and significantly so in seven, but not consistently superior to a resolution-matched random attribution, so no localization claim is made. Calibration was imperfect (slope 0.67) and recalibration gave no net gain; at an in-sample Youden threshold sensitivity was 0.93 and specificity 0.64. Occlusion sensitivity on the whole-heart baseline model showed it relies on the physician-marked septal, aortic and right-ventricular sites 1.8–4.1 times more than distance-matched surrounding tissue, while gross morphometry alone reached 0.651–0.663. Two of the four sub-volumes did not contain their target: the RVOT prior contained the physician annotation in 48.1% of cases and, because the model samples only the central band, excluded it in 99.4%; the RVH box was offset toward the midline, containing the marked target in 43.4% of annotations. Repositioning the priors, leak-free and derived from controls only, did not change discrimination (all p≥ 0.10), and boxes placed at random positions inside the standardized heart reached 0.765 on average against 0.796 for the published priors, a difference this cohort cannot resolve. Conclusions: As a proof of concept, confounder-matched deep learning can recognize TOF on cardiac CT. Increasing model capacity did not significantly improve on the matched pipeline; the separate contribution of matching itself was not isolated against an unmatched comparator. Given the small, single-centre sample and the absence of external validation, these findings are hypothesis-generating and require external, multi-centre confirmation before any clinical use. Full article

Multispectral imaging systems often exhibit wavelength-dependent focus responses because of chromatic aberration, nonuniform spectral sensitivity, and scene-dependent reflectance, making single-band autofocus inadequate for optimizing the complete spectral image cube. This study proposes a nonlinear band-correlation guided deep reinforcement learning framework for autofocus optimization in a self-developed 31-band multispectral imaging system covering 360–980 nm. Through-focus response curves are first extracted from all spectral channels, and a nonlinear inter-band dependency matrix is constructed to characterize complementary and redundant focusing information across wavelengths. A compact subset of representative bands is then selected and embedded into the reinforcement-learning state to guide closed-loop motor control. The agent jointly determines focusing direction, displacement, and stopping time while maximizing a global multispectral focus objective that considers mean sharpness, inter-band consistency, and worst-band degradation. During online autofocus, only the representative bands are acquired at intermediate positions, whereas the complete 31-band cube is captured after convergence. The framework is evaluated using standardized optical targets and spectrally heterogeneous scenes through focal-position error, MTF, full-band focus retention, acquisition time, and robustness. The proposed design provides an efficient computational-photonics solution for broadband multispectral autofocus with reduced band-acquisition and motor-search overhead. Full article

Power-electronic converter control is a key issue in the weak-grid integration of doubly fed induction generator (DFIG)-based offshore wind farms. In conventional fixed-reference DFIG control, the rotor-side power command is usually treated as a tracking target, although its suitability may change with the present point of common coupling (PCC) voltage and phase-angle condition. This paper proposes a weak-grid dynamic sensitivity-based model reference governor (WG-DSMRG) for DFIG offshore wind-farm converter control. The governor is inserted upstream of the RSC power-reference path, while the conventional RSC/GSC current controllers, phase-locked loop (PLL), coordinate transformations, and modulation structure are retained. The short-horizon relation between wind-farm power variation and PCC voltage-angle response is estimated from measured electrical signals. The RSC power command is then corrected through weak-grid scheduling and converter-capability projection. A 60-MW offshore DFIG wind farm connected to a weak AC grid is tested under an upstream voltage sag and a PCC single-line-to-ground fault. In the tested cases, the PCC reactive-power peak decreases from about 3.5 Mvar to 2.5 Mvar, and the DC-link voltage peak is reduced under both disturbances. The evaluated PCC voltage and current THD values are also lower with the proposed controller. These results show that reference-layer correction can improve weak-grid integration and power quality without replacing the established DFIG converter-control platform. Full article

Background: Ultra-high-frequency ultrasound (UHFUS), performed in the present study using 48 and 70 MHz transducers, enables high-resolution assessment of superficial tissues and may provide useful information on the morphology, internal architecture, vascularity, and anatomical relationships of oral cavity lesions. Methods: This retrospective single-center study included consecutive patients examined between January 2025 and June 2026. Of 137 eligible cases, five were excluded because of inadequate image or cine-loop quality, leaving 132 lesions for analysis. All examinations were performed by the same experienced radiologist using 48 and 70 MHz linear probes. Each lesion was assessed in B-mode in at least two orthogonal planes and with Color Doppler. Archived anonymized examinations were reviewed using a predefined structured form. Morphology, margins, echogenicity, echotexture, composition, posterior acoustic features, vascularity, dimensions, presumed plane of origin, and involvement of adjacent anatomic layers were evaluated. The qualitative synthesis was performed jointly by a radiologist and an oral pathologist, with disagreements resolved by consensus. Results: UHFUS allowed detailed visualization of mucosal, submucosal, muscular, and bone/periodontal interfaces and enabled qualitative characterization of lesion boundaries, internal structure, vascular patterns, and local extension. The combined use of 48 and 70 MHz probes provided complementary information according to lesion depth and tissue composition. Five representative histologically confirmed cases from the study cohort illustrated reactive/inflammatory, benign neoplastic, and malignant conditions. One additional clinically confirmed gingival fistula, not included in the analytical cohort, was presented solely as an illustrative example of a superficial fistulous tract detectable with UHFUS. Conclusions: UHFUS provides detailed, layer-based assessment of oral cavity lesions and may complement clinical examination and histopathology in lesion characterization and preoperative evaluation. Sonographic findings should be interpreted within a multimodal diagnostic framework rather than as stand-alone criteria. Full article

With the increasing demand for miniaturization, multi-channel RF transceiver capability, and high-speed digital processing in unmanned aerial vehicles, satellite remote sensing, and anti-jamming communication systems, conventional board-level discrete integration schemes face significant limitations in terms of interconnect length, parasitic effects, volume and weight, and electromagnetic compatibility. This paper proposes a high-density integrated RF microsystem packaging scheme based on a 12-layer ABF organic substrate. Within a package size of 37.5 mm × 37.5 mm, the microsystem integrates a digital processing chip, two DDR3 memories, two Flash memories, two broadband RF transceiver chips, and passive components, thereby realizing a 4-receiver/4-transmitter MIMO architecture. Compared with a conventional discrete PCB-based integration scheme, the proposed microsystem reduces the board-level occupied area by approximately 70% and decreases the weight by more than 30%. To address the non-reworkable nature of the in-package DDR3 address/command/clock links and their sensitivity to high-speed signal integrity, a field-circuit co-simulation method combining three-dimensional electromagnetic S-parameter extraction with IBIS models is adopted to optimize the transmission-line impedance and termination parameters under a fly-by topology. The results show that by optimizing the DDR3 clock-line impedance from the conventional 100 Ω differential impedance to 80 Ω differential impedance, and the address/control-line impedance from the conventional 50 Ω single-ended impedance to 40 Ω single-ended impedance, together with 40 Ω and 120 Ω terminations, respectively, overshoot and ringing can be effectively suppressed. The DDR3 eye width is improved from 0.89 ns to 0.92 ns. To address impedance discontinuities in the vertical interconnects of RF channels, a refined structure combining enlarged antipads and accompanying ground vias is proposed. As a result, the worst-case return loss of the RF channel is improved from below 16 dB to 19.69 dB, the maximum insertion loss is reduced from above 0.5 dB to 0.35 dB, and the worst-case inter-channel isolation is improved from below 60 dB to 73.42 dB. To mitigate thermal coupling and localized thermal isolation caused by thickness differences among multiple chips, a locally recessed copper heat spreader is designed, reducing the junction-to-case thermal resistance of the RF chip from 1.25 °C/W to 0.50 °C/W, corresponding to a reduction of approximately 60%. Preliminary hardware-in-the-loop frequency-hopping tests based on the proposed microsystem demonstrate that the system can achieve an analog frequency-hopping rate exceeding 4000 hops/s and a hybrid frequency-hopping rate exceeding 10,000 hops/s. The results indicate that the proposed packaging scheme provides a feasible engineering implementation path for high-density, multi-channel RF microsystem design. Full article

Passive and fixed-damping transfemoral prostheses cannot adapt their resistance to the phase-dependent demands of human gait, and microprocessor-controlled commercial knees remain out of reach for most amputees. We present a fuzzy logic controller that modulates a magnetorheological (MR) damper directly from gait phase and knee joint angle, since linear state-feedback and discrete PI designs are valid only at a single linearization point and require retuning across the gait cycle. The controller is formalized as a fuzzy-basis-function expansion with established coverage and Lipschitz continuity; the universal-approximation property of Mamdani systems grounds fuzzy logic theoretically but does not certify this 8-rule controller’s performance, established empirically instead. A dissipativity-based Lyapunov argument and a numerical sweep establish local closed-loop stability and bounded, rate-limited actuation. The damper couples to the knee through a shaft–bearing–housing assembly sized by free-body and Goodman fatigue analysis and verified by finite-element analysis, with the control pipeline embedded on an ESP32 microcontroller in a 2 kg prototype, corresponding to Technology Readiness Level (TRL) 5–6. In a single-subject case study with one transfemoral amputee, the controller achieved the lowest mean RMSE (0.0557 over three trials) against a non-disabled gait reference among five compared conditions, improving on the best fixed voltage by 20.2% and a passive prosthesis by 6.8×; these are single-subject feasibility results, not a claim of generalizable performance. Full article

The strong coupling among guidance laws, control loops, aerodynamics, and mission constraints poses growing challenges to the design of modern tactical missile guidance systems for autonomous flight. Conventional manual tuning and simulation-based trial-and-error result in long iteration cycles, limited reuse of design knowledge, and poor adaptability to changing scenarios. To address these limitations, we propose GSMultiAgent, a multi-agent collaborative cascade framework built atop Hermes Agent, which transforms natural-language mission requirements into optimized guidance system models through structured agent cooperation with feedback-driven iterative refinement. Three innovations are introduced: (1) a three-layer correction pipeline covering syntactic checking, deterministic mathematical verification, and semantic reasoning; (2) a bimodal experience repository supporting similarity-guided retrieval with access-count decay and best-quality retrieval for PPO warm-start initialization; and (3) a self-adjudicating optimizer that autonomously decides between PPO-based systematic parameter search and heuristic LLM-tuning guided by a reflection agent. Across four engagement scenarios, GSMultiAgent consistently attains high feasibility at a small fraction of the simulation budget required by conventional optimizers and single-agent baselines, and its design paths escalate autonomously from parameter tuning to structural law modification as task difficulty increases. Ablation studies confirm that the reflection agent, the optimization agent, and structured memory each contribute essential and complementary gains. These results establish multi-agent coordination with structured memory and self-adjudicating optimization as an effective paradigm for intelligent, reusable guidance system design. Full article

This paper develops a distributed control framework for leader–follower formation tracking in multi-agent unmanned aerial vehicle (UAV) systems. Feedforward compensation converts the networked tracking task into local error stabilization problems, and an L_p zero-sum game is used to construct a finite-time robust feedback law. The disturbance-free closed loop is proven to converge in finite time, whereas under nonzero disturbances, the result is a certified L_p attenuation bound rather than exact finite-time convergence. A single-critic adaptive dynamic programming architecture approximates the value function, and an offline sampled data training procedure avoids injecting probing noise into the physical plant. In the reported planar outer-loop simulation, the local errors settle within 4.3 s, compared with 8.2 s for the quadratic L₂ baseline, and the reported cumulative disturbance attenuation indicator decreases from 2.74 to 0.48. The current validation uses a fully actuated translational outer-loop abstraction; extensions to underactuated six-degree-of-freedom dynamics, saturation, and hardware experiments are left for future work L_p. Full article

Sacubitril/valsartan is an angiotensin receptor–neprilysin inhibitor (ARNI) widely used in human heart failure and has demonstrated potential cardiorenal and neurohormonal effects in experimental canine studies. Hydrochlorothiazide is a thiazide diuretic that acts at a more distal segment of the nephron than furosemide. Its addition to loop diuretic therapy may enhance sodium and fluid excretion through sequential nephron blockade and thereby improve the diuretic response in cases of refractory congestion. However, information regarding the combination of these two drugs in cats with congestive heart failure is lacking. This report describes the long-term clinical outcome of a cat with recurrent cardiogenic pleural effusion treated with sacubitril/valsartan and hydrochlorothiazide. A 14-year-old spayed female Persian-cross cat was referred for severe respiratory distress caused by recurrent pleural effusion. Echocardiography identified advanced cardiomyopathy with a nonspecific phenotype and overlapping hypertrophic and restrictive features, including focal basal septal hypertrophy, biatrial enlargement, atrial fibrillation and spontaneous echocardiographic contrast. Despite conventional treatment, which included pimobendan, torasemide, and antithrombotic medications, the cat experienced multiple episodes of pleural effusion over the following months, requiring repeated thoracocentesis. Consequently, the treatment regimen was expanded to include sacubitril/valsartan and hydrochlorothiazide. According to the owner, all other cardiac medications were discontinued approximately one week later without veterinary consultation. At long-term follow-up approximately two years after sacubitril/valsartan and hydrochlorothiazide therapy initiation, the cat remained clinically stable, with no further episodes of respiratory distress. Echocardiography revealed persistent severe structural heart disease, although improved left ventricular systolic indices were observed. This case report describes prolonged clinical stabilization in a cat with advanced cardiomyopathy and recurrent congestive heart failure following the administration of sacubitril/valsartan and hydrochlorothiazide over a two-year period. Whether this outcome reflects a specific effect of sacubitril/valsartan, the addition of hydrochlorothiazide, or a combination of both cannot be determined from a single observation. Nonetheless, the duration and completeness of the response observed here warrant prospective evaluation of combination of ARNI therapy and hydrochlorothiazide in cats with refractory congestive heart failure. Further studies are needed to assess the safety and effectiveness of this approach in feline cardiomyopathy. Full article

Background/Objectives: Diabetes management technologies—including continuous glucose monitors (CGM), insulin pumps (CSII), advanced hybrid closed-loop (AHCL) systems, and mobile health applications—have transformed diabetes care, yet gender-specific differences in ad-herence, clinical outcomes, and patient-reported outcomes remain inadequately char-acterised. Methods: We conducted a systematic literature review following PRISMA 2020 guide-lines, searching SciSpace, Google Scholar, and PubMed databases. From 852 identified records, 500 underwent title/abstract screening after duplicate removal; 16 studies were ultimately included in the qualitative synthesis. Results: Included studies examined insu-lin pumps (n = 7), CGM (n = 6), AHCL systems (n = 4), remote monitoring programmes (n = 2), and mobile health applications (n = 1); sample sizes ranged from 72 to 22,697 participants. In predominantly paediatric evidence, females demonstrated higher insu-lin pump discontinuation rates (discontinuation groups: 75% vs. 46%, p = 0.001), driv-en by body image concerns and device visibility (data from a single small predomi-nantly paediatric cohort; group sex compositions, not absolute discontinuation rates), whereas in a single small adult cohort (n = 72), males showed poorer adherence to in-termittently scanned CGM (approximately four fewer scans/day, p = 0.011). Glycaemic outcomes were modestly but consistently different: males achieved slightly higher time-in-range while females exhibited lower glycaemic variability; menstrual cycle effects on glycaemic control were partially mitigated by AHCL systems. Females con-sistently reported greater diabetes-related distress and a more negative perception of glycaemic control despite similar or better objective metrics. Conclusions: In conclusion, sex-based and gender-related differences are apparent across technology type, adherence, gly-caemic control, and psychosocial outcomes—albeit from a limited evidence base re-quiring cautious interpretation; sex-sensitive prescription, education, and technology design are warranted. Note: most included studies reported biological sex (male/female) rather than self-identified gender; conclusions should be interpreted primarily as sex-based differences. Full article

Accurate and low-cost fault diagnosis is essential for improving the reliability of cascaded neutral-point-clamped (NPC) inverters. This paper proposes a single-sensor fault diagnosis method for a single-phase three-module cascaded NPC inverter. Only one DC-side current sensor is required for the diagnostic algorithm, while the voltage sensor used in the outer voltage-control loop is not involved in fault-feature extraction. The measured DC-side current is decomposed via Fourier analysis, and a low-dimensional feature vector is constructed using the amplitudes of the zeroth, 2nd, 3rd, and 4th harmonics together with the phases of the 1st and 3rd harmonics. The six Fourier features are normalized using feature-wise Min–max parameters determined exclusively from the training data. A back-propagation (BP) neural network is then adopted to identify and locate 24 single-switch open-circuit faults in the three-module system. The investigated inverter produces 13 output-voltage levels under healthy operation, and the BP network converges after 5835 training iterations to an error threshold of 1 × 10⁻⁶. An adaptive confirmation criterion based on consecutive diagnosis-code consistency and inter-window feature convergence is introduced. For the nominal 25-class simulation test set, the accuracy, macro-precision, macro-recall, and macro-F1-score are all 100%. In addition, 134 of the 136 dynamic-condition simulation runs are correctly diagnosed, corresponding to an overall robustness-test accuracy of 98.53%. One confirmed, but incorrect final code occurs under the load disturbance applied at 90° of the output-voltage fundamental, and another occurs at an SNR of 20 dB, while no unconfirmed run is observed. Under the severe RL-load condition with τ/T₀ = 1, the mean and maximum diagnostic delays are 41.7 ms and 52 ms, respectively. Full article

Background/Objectives: Insulin clearance is a key determinant of circulating insulin availability, but its diurnal variation and relationship with circadian metabolic programs in metabolic dysfunction associated steatotic liver disease (MASLD) remain unclear. This study aimed to explore diurnal insulin clearance in humans and examine associated metabolic gene signatures in MASLD. Methods: A single-subject pilot assessment was performed to explore daytime-nighttime differences in insulin clearance rate (ICR) surrogate index, followed by evaluation using public hyperinsulinemic-euglycemic clamp datasets from healthy individuals and patients with MASLD. Public circadian transcriptomic datasets, spatial transcriptomic data, and a time course high-fat diet (HFD)-induced mouse dataset were integrated. A predefined panel of insulin clearance-related and circadian genes, including carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1), insulin receptor (INSR), insulin-degrading enzyme (IDE), clock circadian regulator (CLOCK), basic helix-loop-helix ARNT like 1 (BMAL1), nuclear receptor subfamily 1 group D member 1/2 (NR1D1/2), period circadian regulator 1/2 (PER1/2), and cryptochrome 1/2 (CRY1/2), was analyzed. Results: The pilot assessment showed higher nighttime than daytime ICR, and independent clamp datasets showed a similar pattern in healthy individuals. In MASLD, nighttime ICR remained relatively higher, whereas overall insulin clearance was reduced compared with controls. Human blood-based circadian transcriptomic datasets identified rhythmic expression patterns of selected genes involved in insulin clearance and circadian regulation, including CEACAM1, CLOCK, NR1D1, CRY1, PER1, and PER2. MASLD liver datasets showed reduced expression of insulin clearance-related and circadian genes, while spatial transcriptomics suggested altered lobular distribution of these signatures. In HFD mouse model, rhythmic expression of selected genes was attenuated. Conclusions: These integrative findings suggest that insulin clearance may exhibit diurnal variation and may be linked to circadian metabolic gene signatures across systemic and hepatic datasets in MASLD. Larger controlled human studies are needed to validate the temporal regulation of insulin clearance and its metabolic relevance. Full article

Robotic bin-picking of disordered, randomly stacked workpieces remains challenging because reliable grasping depends on an accurate estimate of object pose, yet many established solutions require high-precision 3D sensing, detailed object models, or large annotated datasets that raise the cost and effort of deployment on a new production line. This work presents a complete binocular vision framework that estimates workpiece pose by image matching and executes vision-guided grasping on a 6-DOF manipulator. A pose-annotated multi-view template library is constructed automatically through robot-driven image acquisition and compressed by a coarse-to-fine clustering scheme, and object pose is estimated by discriminative template matching with rigid refinement. To characterize the geometric reliability of the matched poses, an offline cross-modal analysis relates the 2D templates to a 3D reference model of the object and measures their agreement through region and contour reprojection metrics. Grasp configurations are then generated under orientation and collision constraints and corrected online by closed-loop visual feedback. Experiments on two representative workpieces show template-matching accuracy of 89–90% against classical and learned similarity measures, and grasp success between 81 and 87% across single-object and mixed scenes, outperforming the GraspNet baseline under the tested conditions. The framework offers an accurate and deployment-friendly route to robotic bin-picking. Full article

Conditional imitation learning (CIL) for CARLA depends on diverse expert data spanning map, weather, traffic, and control-perturbation axes, yet it remains unclear which axes actually drive closed-loop behavior, and whether ablation conclusions survive deployment guardrails. We train a matched baseline and four equal-budget leave-one-axis-out (LOO) variants of a fixed CIL architecture (v14) and evaluate each under three nested tiers: pure policy rollout, minimal traffic-rule shields, and a fully deployed stack with route blending and recovery. The factorial design comprises $18\times 5\times 3$ scenario-variant-tier cells, each repeated under n = 5 traffic-seed replicates (1350 closed-loop episodes) to estimate NPC-seed variance on every eval stack. No single withheld axis dominates pooled outcomes. LOO effects are tag- (scenario-category) and spawn- (vehicle starting location) specific: removing perturbation-labeled recovery data costs $-285$ m on geometry_stress but can gain distance on in-distribution spawns; removing multi-town data changes held-out Town05 mobility on some routes while depressing others. Axis-importance rankings reorder across tiers; Kendall $\tau$ between pure and full rankings is $0.0$, and guardrails compress or erase pure-tier gaps (e.g., drop_perturbation pooled distance $\Delta$ from $-93$ m to 0 m). Pure-tier seed replicates show that geometry-driven lane-tracking degradation under drop_perturbation is seed-stable; traffic-axis and full-tier drop_traffic loads are more seed-sensitive. We release the evaluation ledgers, parsing scripts, and analysis tooling with the paper. Training-data ablation claims should report per-scenario or tag-stratified LOO metrics under pure evaluation; guardrailed tiers are supplementary deployment checks, not substitutes for isolating what the policy learned. Full article

Assessing whether existing lattice towers can survive extreme wind when they are located on ridgelines or at saddle points requires three questions to be answered simultaneously: how the local wind field is modified by the surrounding topography, how the structural geometry recorded in legacy computer-aided design (CAD) drawings can be recovered accurately, and which member fails first and by what mechanism. This paper couples a Physics-Informed Neural Network (PINN) wind solver, jointly constrained by mass and momentum conservation, with a graph-theory-based Drawing Exchange Format (DXF) parser and a closed-loop vulnerability screening module, so that all three questions are answered in a single automated pass. The PINN reconstructs the three-dimensional steady-state wind field over irregular topography in approximately 0.12 s, holding the root-mean-square (RMS) velocity divergence below 2.1 × 10⁻³ (more than two orders of magnitude lower than that of linear interpolation) while recovering the pressure-gradient-driven acceleration that mass-consistent variational solvers cannot represent. On the CAD side, k-dimensional tree (KD-Tree) spatial indexing combined with breadth-first search (BFS) connected-component analysis resolves the pseudo-disconnections, spurious intersections, and multi-level nested block references that are common in production DXF files, achieving 100% node-merging accuracy across fifteen tower drawings. A unified Vulnerability Index (VI) that combines strength, member stability, and plate buckling into a single scalar, updated through Sherman–Morrison rank-one perturbation at a millisecond cost, closes the diagnose–strengthen–verify loop without re-solving the full stiffness system. Applied to a 220 kV line struck by Super Typhoon Meranti, the pipeline identified seven at-risk members that code-based checking had missed, a result consistent with the recorded field damage, and completed the full assessment in 34.4 s, over three orders of magnitude faster than conventional practice. Full article