Search Results (15,939)

Hybrid renewable microgrids are increasingly considered a practical solution for supplying reliable and sustainable electricity to remote regions. However, many planning studies do not consider the longer-term effects of battery degradation and the effect of changing operating conditions on system performance. This study develops an integrated techno-economic planning mathematical model for a self-sufficient PV/WT/DG/BESS microgrid using actual hourly meteorological and load-demand data from New Minia, Egypt. This framework integrates variability in renewable resources, battery degradation, reliability constraints, environmental factors, and uncertainty evaluation as part of a comprehensive assessment. The proposed approach identifies a system configuration that balances supply reliability, economic performance, renewable energy (RE) penetration, and long-term storage sustainability. The optimal configuration achieved a cost of energy (COE) of 0.1545 $/kWh and a net present cost of approximately 4.5 M$, while maintaining the RE contribution of 79.43%. The configuration achieved a low loss of power supply probability (LPSP) of 0.00593 and annual expected energy not served of 13,516.52 kWh. The resultant configuration reduced dependence on diesel generation and decreased yearly CO₂ emissions to 483.05 ton/year. A degradation-aware battery model was incorporated to represent long-term storage behavior, yielding an estimated battery service life of approximately 12.16 years under the adopted operating assumptions. Furthermore, deterministic sensitivity analysis was conducted to evaluate the influence of key economic and system parameters on the techno-economic performance of the proposed microgrid. Overall, the findings demonstrate the effectiveness and practical potential of the proposed degradation-aware and uncertainty-driven framework in supporting reliable and economically sustainable hybrid-microgrid planning. Full article

Light Detection and Ranging (LiDAR)-based forest inventory increasingly relies on diverse platforms, ranging from proximal systems including BackPack, All-Terrain Vehicle (ATV), and terrestrial laser scanning (TLS) to near-proximal systems such as uncrewed aerial vehicles (UAVs). However, differences in point density, viewing geometry, and occlusions among these acquisition systems pose challenges for processing heterogeneous LiDAR datasets using a common workflow. Traditional geometric approaches often rely on parameter tuning. On the other hand, deep learning (DL) approaches can be constrained by domain shift when applied to different sensors or forest environments. This study proposes a forest inventory pipeline for individual tree segmentation and the derivation of key forest biometrics including tree location and diameter at breast height (DBH) across heterogeneous LiDAR datasets. The pipeline uses a confidence-guided, multi-stage quality control framework that evaluates agreement between complementary tree location estimates to reduce common segmentation errors. In addition, a semi-automated procedure is developed to generate reference data for datasets lacking field measurements. The proposed workflow was evaluated using eight diverse datasets representing different platforms, sensors, acquisition patterns, and forest environments and was compared with 3DFIN, TreeLearn, and ForestFormer3D. Field reference measurements were available for a natural forest site, while the remaining datasets were evaluated using semi-automatically generated and manually refined reference data. The proposed tree detection pipeline achieved Precision ranging from 86.44% to 100%, Recall from 74.17% to 100%, and F1-scores from 81.82% to 100% across the evaluated datasets. For the Martell–BackPack dataset with independent field reference measurements, Precision, Recall, and F1-score were 97.55%, 96.95%, and 97.25%, respectively. For correctly detected trees by the proposed approach in the natural forest dataset with field measurements, DBH estimates achieved an RMSE of 2.5 cm with the total basal area underestimated by 1.88%, compared with DBH RMSE and reduction in basal area of 4.0 cm and 3.67%, respectively, for 3DFIN. Although the proposed pipeline did not achieve the highest performance in every test case, it maintained strong and generally consistent tree detection performance for the evaluated datasets. The main limitation of the proposed pipeline is its dependence on sufficient lower-stem visibility, which reduced tree detection accuracy in sparsely sampled areas. The proposed framework provides a practical workflow for LiDAR-based individual tree segmentation and DBH estimation using a fixed parameter configuration for all datasets captured by a given acquisition system. Full article

A pipe-in-pipe structure with a buoyancy damping layer is designed to reduce pipeline weight and suppress vortex-induced vibration. This structure avoids structural pre-stress problems caused by external buoys and mitigates severe local vibration. The four-degree-of-freedom motion equation of the novel pipe-in-pipe system is established, and a corresponding numerical program is compiled. Bidirectional fluid–structure interaction (FSI) simulations are performed to verify the vibration reduction performance of the proposed structure. The auxiliary outer pipe alters surrounding flow field characteristics, generates distinct hydrodynamic forces and modifies structural vibration behaviors. Massive simulation data demonstrate that the buoyancy damping layer of the pipe-in-pipe structure can adapt to hydrodynamic forces with various characteristics and achieves favorable broadband vibration suppression and energy absorption. For all layout parameters, the inner damping layer exhibits an outstanding suppression effect on cross-flow vibration of the cylinder-auxiliary pipe system. The inline vibration amplitude of the pipe-in-pipe-auxiliary pipe system is consistently less than 0.08D and can be neglected. Its vibration frequency is close to that of the cross-flow direction and only half that of the single-layer pipe. The inner damping layer also affects the fluid–structure interaction between the pipeline system and external flow field, resulting in variations of the vorticity field. Full article

To meet the demands for high-density integration and broadband RF performance in next-generation electronic equipment, this paper investigates the structural design and process sensitivity of vertical interconnects in three-dimensional stacked system-in-package (3D-SiP). Based on quasi-coaxial matching and low-impedance compensation techniques, three typical vertical interconnect structures are designed: the PCB-BGA-SiP microstrip structure achieves S₁₁ better than −15 dB and S₂₁ < 0.3 dB within 2–20 GHz; the lower stripline–BGA–upper microstrip structure achieves S₁₁ better than −27 dB; the lower microstrip–BGA–upper microstrip structure achieves S₁₁ better than −18 dB within 1–23 GHz. Isolation simulation shows that the isolation between adjacent channels exceeds 55 dB within 25 GHz. For process sensitivity evaluation, physical samples of gold wire bonding parameters (length, diameter, number) were fabricated and tested for S₁₁, confirming that the dual-wire topology extends the effective bandwidth to 2–20 GHz (a 54% improvement over single wire), the third-wire marginal gain is only ~5%, and 25 μm diameter offers the best overall performance. For BGA ball radius and pad pitch, parametric sensitivity analysis via Ansys HFSS was performed (not experimentally validated process variation results), identifying the optimal BGA radius as 0.245 mm with an allowable variation of ±0.015 mm, and the pad center-to-center distance should be controlled near 0.8 mm. Based on these findings, process control strategies are proposed: low-loop wire bonding (loop height < 50–80 μm) with statistical process control; substrate warpage controlled through symmetric copper filling, a thick-middle dielectric stack-up, and distributed symmetric cavity layout, combined with eutectic pressure of 1.5 kPa, validated on 10 fabricated substrates with peak warpage consistently < 80 μm, void rate 4.75%, and solder overflow 94%; and BGA soldering using high-precision vision alignment and controlled collapse (20–35%). This work provides a theoretical basis and practical process pathway for transitioning 3D-SiP vertical interconnects from ideal design to mass production. Full article

Smart-city energy platforms—peer-to-peer markets and tokenized renewable energy certificates—settle against metered generation, yet never check the physical plausibility of those claims. For wind, attainable energy is not measured but derived from anemometry through a vertical extrapolation whose exponent—the wind-shear coefficient—is a discretionary modeling choice. Using a five-height, 52,192-record campaign from Phangan Island, Thailand—reproduced here as a statistically anchored reconstruction, the raw archive not being redistributable—we show that the conventional $1/7$ rule understates attainable energy by $29.8\%$, and that an adversary asserting the exponent could inflate a resource claim by $87.4\%$. WindChain sits beneath the smart-city transactive layer rather than beside it: it does not mint certificates from wind data but bounds what a revenue meter may claim. It enforces boundary-layer, kinematic, and thermodynamic admissibility as a consensus predicate and commits wind statistics to a hierarchical Merkle–Weibull accumulator whose 104-byte root lets any verifier recompute the Weibull parameters, power density, and shear exponent in constant time. We prove that an epoch-level admissibility gate bounds over-issuance, and that attestation windows must be thirty-six times longer than independence assumes. On the reconstruction, WindChain detects six of eight manipulation classes—four of them with recall $\ge 0.99$—at a $1.75\%$ false-positive rate; we also report a camouflage regime defeating every per-record test, and a sustained bias at or below $2.6\%$ that the epoch detector does not see. Consensus performance is modeled, not deployed. WindChain narrows the trust boundary rather than removing it: the guarantee is conditional on an independently certified site reference and on physical calibration of the mast and is best read as an auditable plausibility layer beneath settlement rather than as a trustless one. Full article

We investigate scalar perturbations of a rotating asymptotically anti-de Sitter (AdS)-Teo traversable wormhole with a controlled non-separable angular deformation. The geometry retains a regular wormhole throat and the required AdS asymptotics, while an explicit quadrupolar deformation generates angular-channel coupling in the intermediate region. A sufficient condition is derived for an ergoregion-free parameter regime in which the spectral analysis is performed. Projecting the geometry-derived Klein–Gordon equation onto spherical harmonics yields a matrix-valued Sturm–Liouville system and an associated quadratic operator pencil. Throat regularity together with normalizable AdS boundary conditions leads to a determinant quantization condition for the discrete normal-mode spectrum. Two-, four-, and six-channel calculations demonstrate systematic numerical convergence of the retained low-lying normal-mode frequencies under enlargement of the angular basis. The complete finite generalized eigenspectrum is also examined without imposing a near-reality selection criterion, and no growing scalar mode is found within the ergoregion-free parameter range and numerical resolutions studied. The six-channel rotation continuation exhibits a finite interior level-repulsion feature accompanied by collective redistribution of the multichannel eigenvectors. The same coupled spectral framework formally defines a matrix-valued two-boundary response whose poles are selected by the normal-mode matching condition; the response plots presented here illustrate this structure using the reduced two-channel effective model rather than a numerical reconstruction of the full six-channel response. We further derive the geometry-dependent short-distance Hadamard subtraction and identify the finite inter-channel structure entering a truncated local quantum mode sum, without claiming a complete numerical evaluation of ${⟨{\Phi }^2⟩}_{\mathrm{ren}}$. Together, these results establish a geometry-to-spectrum framework in which the spacetime geometry determines the coupled operator, the global boundary conditions determine its normal-mode spectrum, and the same coupled spectral framework organizes the associated asymptotic response and local quantum mode-sum structure. Full article

Generalized category discovery (GCD) aims to classify known categories while discovering novel ones in unlabeled data, yet existing methods lack mechanisms to correct boundary-ambiguous samples that receive noisy pseudo-labels, as they primarily rely on visual feature learning without external semantic guidance. Vision-language models (VLMs) offer a natural source of cross-modal semantic correction. However, applying VLM-guided contrastive signals directly within the GCD training loop proves counterproductive because the locally-oriented InfoNCE loss conflicts geometrically with the globally oriented K-means objective in the shared backbone space. We identify the root cause as a dual resource allocation problem: the VLM-derived signal must be allocated to the correct feature subspace to avoid geometric conflict with K-means clustering (space allocation), and the limited VLM inference budget must be allocated to the correct samples to maximize discriminative return (budget allocation). These two decisions are coupled; failure on either renders the other ineffective. To resolve this, we propose DualGLEAN, a framework that addresses the dual allocation challenge through two coupled mechanisms: decoupled contrastive alignment (DCA), which routes the VLM-guided neighbor contrastive loss to a dedicated projector space while preserving the backbone space for global clustering, and compound uncertainty querying (CUQ), a three-stage filtering metric that jointly evaluates predictive entropy, boundary proximity, and local label inconsistency to direct VLM queries exclusively to truly boundary-critical samples. Extensive experiments on the AID and RSSDIVCS datasets demonstrate that DualGLEAN achieves strong performance, improves four diverse GCD baselines as a plug-in module, generalizes across seven VLM backbones, introduces zero additional trainable parameters to the base GCD network, and incurs a total VLM API cost of only CNY 2.45 per full training run on the AID dataset under the default search-scope configuration, with the cost scaling linearly with the query budget. Full article

The accelerating contamination of global ecosystems by heavy metal ions, per- and polyfluoroalkyl substances (PFASs), microplastics and nanoplastics (MNPs), and emerging contaminants demands sensing technologies that are rapid, sensitive, selective, and field-deployable. Quantum dots (QDs) have emerged as leading candidates for environmental sensing; however, their performance is often interpreted empirically rather than through a unified understanding of the underlying excitonic physics. This narrative review presents a mechanistically integrated framework for QD-based environmental sensing, establishing the exciton, the spatially confined electron–hole quasiparticle, as the primary signal carrier in the most analytically powerful QD sensing modalities. A critical distinction is drawn between three categories of signal-generating processes: genuine excitonic transduction (photoinduced electron transfer, trap-state modulation, FRET, charge-transfer exciton formation, and binding energy modulation); non-excitonic optical phenomena, including the inner filter effect and light scattering, which are frequently misattributed as excitonic responses; and partially excitonic processes such as certain electrochemiluminescence pathways. Exciton fundamentals, confinement effects, and the influence of defects, dopants, and surface states are examined across carbon, chalcogenide, perovskite, and III–V QD families. A Defect–Exciton Energy Map is introduced as a rational design tool linking defect characteristics to excitonic response regime and sensing modality. Application of the mechanistic framework to heavy metal ions, PFASs, microplastics, and emerging contaminants demonstrates that sensing performance differences are mechanistically predictable from excitonic parameters rather than being arbitrary outcomes of materials choice. Benchmarking against competing platforms identifies conditions under which QD sensors offer genuine advantages. The roles of density functional theory, molecular dynamics, and machine learning in enabling rational sensor design are assessed. Key challenges, including stability, real-sample validation, standardisation, and toxicity, and future directions, including QD/two-dimensional material heterostructures and circular economy carbon QD platforms, are identified. Full article

Purpose: To compare the clinical and functional outcomes of two successive generations of a trifocal intraocular lens (IOL), the original Liberty 677MY and the optimized Liberty 677CMY. To the best of our knowledge, this is the first head-to-head comparison between two successive generations of any IOL model reported. Methods: This retrospective study included 42 patients (84 eyes) who underwent bilateral phacoemulsification with implantation of either the Liberty 677MY (26 eyes) or the Liberty 677CMY (58 eyes). The primary outcome was postoperative binocular intermediate visual acuity (VA). Secondary outcomes included monocular intermediate VA, monocular and binocular distance and near VA, the proportion of patients or eyes achieving ≤0.1 LogMAR, and defocus curves. Exploratory outcomes included spectacle independence, patient-reported satisfaction (Quality of vision—QoV- and Catquest-9SF), photic phenomena, and intraoperative or postoperative complications. Results: At 3 months, binocular UIVA differed by approximately 0.10 LogMAR in favor of the 677CMY cohort, which was considered clinically significant, although the difference did not reach statistical significance (p = 0.052). Supporting this tendency, monocular UIVA was significantly better with the 677CMY IOL (p = 0.012). Also, a significantly higher percentage of patients and eyes achieved binocular and monocular UIVA ≤ 0.1 LogMAR with the 677CMY IOL (p = 0.025). Defocus curves additionally showed significant between-group differences at intermediate defocus levels (−1.00 and −1.50 D; p < 0.05), favoring the 677CMY cohort. No between-group differences were observed for any distance VA parameter. Binocular near VA and near-vision thresholds were also comparable between groups, although a greater proportion of eyes achieved monocular UNVA ≤ 0.1 LogMAR with the 677CMY IOL (p = 0.016). Spectacle use, overall satisfaction, photic phenomena profile, and postoperative complications were comparable between groups. Conclusions: The optimized Liberty 677CMY exhibited a favorable pattern of intermediate visual performance compared with the 677MY while maintaining comparable distance and near VA and a similar safety profile. These findings suggest potential clinical benefits of the design modifications, although larger prospective studies are warranted to confirm these observations. Full article

Objectives: To investigate the rate and predictors of recurrence at second-look ureterorenoscopy (URS) and to assess the adequate timeframe for second-look URS in upper tract urothelial carcinoma (UTUC). Methods: This multicenter retrospective study included 179 patients who underwent endoscopic kidney-sparing surgery (KSS) and second-look URS for non-invasive UTUC between 2004 and 2017. We performed logistic and Cox proportional hazard regression analyses to investigate the association between clinical parameters and second-look recurrence, recurrence-free survival (RFS), and cancer-specific mortality. The optimal duration to second-look URS was extrapolated using restricted cubic splines. Results: Overall, 66 (36.9%) patients experienced recurrence at second-look URS. After second-look URS, further recurrence was observed in 87 (48.6%) patients during a median follow-up of 12 months. Female gender (OR [odds ratio] 2.3, 95% CI [confidence interval] 1.1–4.7, p = 0.026) and tumor size > 1 cm (OR 3.3, 95% CI 1.2–9.5, p = 0.027) were independently associated with recurrence at second look. Second-look recurrence was a strong prognostic factor for RFS (HR 5.0, 95% CI 3.1–8.0, p < 0.001). Second-look URS between 5 and 12 weeks after initial endoscopic laser ablation was an independent favorable factor for RFS (HR 0.5, 95% CI 0.3–0.9, p = 0.014). Conclusions: Early recurrence at second-look URS appears to be one of the strongest predictive factors for RFS in patients undergoing endoscopic KSS. Our results suggest that second-look URS performed within 5 to 12 weeks is associated with a lower probability of further disease recurrence. Given that this interval was derived and tested in the same cohort, this finding should be regarded as hypothesis-generating and requires external validation before being adopted as a clinical standard. Full article

Large language models (LLMs) incur prohibitive computational costs when deployed as monolithic systems for multi-domain query processing. This paper proposes Accelerated LLM, a modular architecture that replaces a single general-purpose LLM with an ensemble of task-specialised small language models (SLMs) governed by a neural query router and a Mamdani fuzzy inference system. The router embeds each user query using a frozen sentence encoder and classifies it across four task domains—summarisation, translation, question answering, and text generation—routing confident queries directly to the corresponding SLM. Ambiguous queries are escalated to a three-input fuzzy logic system operating on Query Length, inter-Domain Overlap Score, and Classifier Confidence, enabling principled handling of imprecise inputs. A reinforcement-learning feedback loop, validated through a controlled pilot deployment, continuously refines the routing policy. The complete pipeline, including the sentence encoder, totals approximately 2.14 billion parameters—a 98.8% reduction relative to GPT-3.5 (175 B). The integration of fuzzy logic into the routing stage raises classification accuracy from 91.5% to 94.3% and reduces the hallucination rate to 9.8% (minor) and 6.4% (major). Evaluated on healthcare-augmented benchmarks against ChatGPT-3.5, Claude, Mistral 70B, and two contemporary compact models (GPT-4o-mini and Llama 3.1-8B-Instruct), Accelerated LLM achieves competitive or superior task-specific performance at a fraction of the parameter count. A small-scale pilot evaluation in the legal domain indicates that the routing and fuzzy logic components retain partial effectiveness beyond the primary healthcare setting, though full multi-domain validation remains future work. Full article

Reliable human target detection in smoke-filled environments is essential for firefighting robots and rescue perception systems. However, conventional RGB cameras are severely degraded by dense smoke, while a single infrared or thermal imaging sensor cannot simultaneously provide sufficient structural details and reliable target-related thermal information. To address these challenges, this paper proposes a local variance-guided adaptive infrared–thermal sensor fusion framework for human target detection in smoke-filled environments, aiming to alleviate smoke-induced degradation in multimodal perception through improved infrared representation and adaptive cross-modal information utilization. An improved dark channel prior-based infrared desmoking algorithm is designed, where guided filtering is employed to refine the transmission map, suppress halo artifacts, and enhance infrared image quality. Furthermore, a local variance-guided adaptive fusion strategy is proposed, which utilizes local variance as an information saliency metric to generate pixel-level adaptive modality weights for fusing desmoked infrared and thermal images. In addition, a lightweight YOLO11n detector is adopted to achieve efficient human target recognition while maintaining a favorable balance among detection accuracy, computational cost, and inference efficiency. Experimental results on the self-built dense-smoke dual-modal dataset demonstrate that the proposed framework achieves high detection performance with low model complexity and efficient detector-stage inference. The ablation results demonstrate the contribution of infrared–thermal multimodal fusion to reliable smoke perception and indicate that the proposed local variance-guided adaptive fusion strategy maintains comparable detection accuracy while providing a better detector-stage speed–accuracy balance than fixed-weight fusion. With a low parameter count, the adopted YOLO11n detector shows potential for future deployment on resource-constrained firefighting robotic platforms. Full article

Detecting small person and bicycle instances with lightweight models is relevant to resource-aware visual sensing, but evidence from a category-filtered general-purpose dataset cannot establish performance in dense surveillance, traffic monitoring, or aerial environments. This study therefore examines a narrower question: whether replacement-style Coordinate Attention (CA) integration can improve coordinate-sensitive representation in YOLOv11m under a controlled VOC-based person and bicycle setting without increasing model complexity. In the official Ultralytics YOLO11m architecture, the Spatial Pyramid Pooling–Fast (SPPF) layer is followed by a C2PSA block. The proposed configuration replaces this post-SPPF C2PSA block with CA, while retaining the remaining backbone, neck, and detection head. CA encodes directional positional information along the horizontal and vertical axes. Experiments are conducted on a filtered subset of PASCAL Visual Object Classes (VOC) 2012 that retains only the person and bicycle categories; this subset is not a dedicated small-object or surveillance benchmark. Within this setting, the proposed model improves mean average precision at an intersection-over-union threshold of 0.50 (mAP@50) from 79.2% to 81.5%. It also improves mean average precision averaged over thresholds from 0.50 to 0.95 (mAP@50–95) from 53.9% to 54.8% and small-instance average precision (${\mathit{AP}}_{\mathrm{s}}$) from 68.1% to 72.4%. The parameter count decreases from 20.03M to 19.07M, and the model achieves 90 frames per second (FPS) on the tested NVIDIA GeForce RTX 4060 Laptop GPU (NVIDIA Corporation, Santa Clara, CA, USA). These results provide incremental evidence for a complexity-aware CA replacement strategy within the evaluated VOC distribution; they do not demonstrate cross-domain generalization or deployment performance in real surveillance or aerial scenarios. No embedded platform was evaluated, and the reported RTX 4060 throughput should not be interpreted as evidence of edge-device latency, energy efficiency, or deployment readiness. Full article

Background/Objectives: Hemorheological abnormalities, including increased blood viscosity and impaired red blood cell (RBC) deformability, contribute to poor vascular outcomes in patients with atherosclerotic diseases. Although sarpogrelate has shown vascular benefits, its additive effect with aspirin on blood viscosity remains unclear. This study aimed to evaluate whether adding sarpogrelate hydrochloride to aspirin improves blood viscosity and hemorheological parameters compared with aspirin monotherapy in patients with peripheral arterial disease (PAD) and coronary artery disease (CAD). Methods: This single-center, randomized, open-label trial evaluated sarpogrelate hydrochloride (300 mg/day) plus aspirin (100 mg/day) versus aspirin monotherapy over 12 weeks in patients with PAD and CAD. Patients were assigned using a computer-generated sequence. The primary analysis used the full analysis set. The primary outcome was change in systolic and diastolic blood viscosity. Secondary outcomes included RBC deformability, aggregation index, critical shear stress, flow-mediated dilation (FMD), metabolic parameters, and quality of life. FMD was performed using standardized protocols with blinded assessment. Results: Sixty-eight patients were randomized (34 per group), of whom 61 were included in the full analysis set (experimental group, n = 31; control group, n = 30). At week 12, systolic blood viscosity remained stable in the experimental group (4.51 → 4.49 cP) but increased in the control group (4.67 → 4.87 cP). The unadjusted mean between-group differences in change from baseline (experimental minus control) were −0.23 cP (95% CI, −0.66 to 0.20) for systolic blood viscosity and −0.55 cP (95% CI, −2.03 to 0.93) for diastolic blood viscosity; the corresponding longitudinal between-group p-values from the MMRM analyses were 0.9481 and 0.9630, respectively. FMD showed no significant between-group difference at week 12 (unadjusted mean difference, −0.60%; 95% CI, −2.05 to 0.85; MMRM p = 0.7727). Patient-reported outcomes (SF-36 and visual analog scale scores) did not differ significantly between groups. Conclusions: In patients with PAD and CAD, no statistically significant additional hemorheological benefit of sarpogrelate combined with aspirin was demonstrated compared with aspirin monotherapy. Full article

This work presents a compact square complementary split-ring resonator (CSRR) microwave sensor, combined with a machine-learning-based calibration strategy, to achieve superior texture-aware soil moisture quantification. Implemented on a Rogers RO3010 substrate with a 20 × 30 mm² footprint and operating near 1.3 GHz, the sensor exploits shifts in resonance/notch frequency and insertion loss (S21) to probe both the real and imaginary components of the soil’s complex permittivity. Full-wave 3D electromagnetic simulations guided optimisation of the CSRR topology and T-shaped microstrip feedline, yielding strong field confinement, high quality factor, and high Frequency Detection Resolution (FDR). Experiments on sand and loam across 0–30% and 0–40% moisture content ranges, respectively, demonstrate FDR values of 6.09 MHz (sand) and 6.86 MHz (loam), enabling discrimination of subtle permittivity changes. Several calibration strategies are developed and compared for complex permittivity extraction from measured S-parameters: linear and polynomial regression, a multivariable least-squares sensitivity-matrix model, and a delta-referenced multilayer perceptron (MLP) with z-score standardization. While polynomial and least-squares models significantly outperform linear regression (R² > 0.997), the MLP combined with the optimized CSRR architecture delivers the best performance, achieving near-ideal accuracy (R² ≈ 1, MAE < 0.001, RMSE < 0.001) for both soil types. These results demonstrate that the synergy between the novel CSRR sensor design and data-driven MLP calibration enables high-resolution, robust, and field-deployable soil moisture sensing, offering a compelling solution for next-generation agricultural and geotechnical monitoring systems. Full article