Sensors最新文献

Unmanned aerial vehicles (UAVs) are increasingly used for crack inspection of civil infrastructure. However, crack interpretation from UAV imagery is constrained by trade-offs among imaging resolution, operational efficiency, and measurement uncertainty. Higher resolution generally requires reduced flight distance, increased image quantity, and greater data-processing effort, which can limit inspection efficiency. This study presents an exploratory analysis of UAV-based crack inspection from a measurement-oriented perspective. Empirical UAV flight experiments were conducted to examine the relationships among flight distance, ground sampling distance (GSD), image quantity, and photogrammetric processing effort under controlled acquisition conditions. In addition, a dataset-based segmentation analysis was performed to investigate pixel-level uncertainty associated with crack thickness representation near the resolution limit. This analysis does not aim to estimate physical crack width, but rather to identify intrinsic limitations of image-based crack interpretation. The results indicate that while flight distance and GSD follow expected geometric relationships, image quantity and processing effort are influenced by multiple interacting factors rather than resolution alone. Pixel-level analysis further reveals substantial segmentation uncertainty for thin cracks represented by only a few pixels. These findings highlight the importance of accounting for measurement uncertainty and operational trade-offs when planning efficient UAV-based crack inspections.

The detection of defects in key transmission-line equipment under complex environments often suffers from insufficient accuracy and reliability due to background interference and multi-scale feature variations. To address this issue, this paper proposes an improved defect detection model based on YOLOv11, named DCDW-YOLOv11. The model introduces deformable convolution C2f_DCNv3 in the backbone network to enhance adaptability to geometric deformations of targets, and incorporates the convolutional block attention module (CBAM) to highlight defect features while suppressing background interference. In the detection head, a dynamic head structure (DyHead) is adopted to achieve cross-layer multi-scale feature fusion and collaborative perception, along with the WIoU loss function to optimize bounding box regression and sample weight allocation. Experimental results demonstrate that on the transmission-line equipment defect dataset, DCDW-YOLOv11 achieves an accuracy, recall, and mAP of 94.4%, 92.8%, and 96.3%, respectively, representing improvements of 2.8%, 7.0%, and 4.4% over the original YOLOv11, and outperforming other mainstream detection models. The proposed method can provide high-precision and highly reliable defect detection support for intelligent inspection of transmission lines in complex scenarios.

Performance diagnostics in alpine skiing increasingly rely on integrated biomechanical and kinematic assessments to support technique optimization under real training conditions; however, many existing approaches address trajectory geometry or biomechanical variables separately, limiting their explanatory power. This study evaluates an integrated analysis framework combining OptiPath, an AI-assisted video-based trajectory analysis tool, with XSensDOT wearable inertial sensors to identify technical inefficiencies during giant slalom skiing. Thirty competitive youth athletes (n = 30; 14-16 years) performed controlled runs with predefined lateral offsets from the gates, enabling systematic examination of the relationship between spatial trajectory deviations, biomechanical execution, and performance outcomes. Skier trajectories were extracted using computer vision-based methods, while lower-limb kinematics, trunk motion, and tri-axial acceleration were recorded using inertial measurement units. Deviations from mathematically defined ideal trajectories were quantified through regression-based calibration and arc-based modeling. The results show that although OptiPath reliably detected trajectory variations, shorter skiing paths did not consistently produce faster run times. Instead, superior performance was associated with more efficient biomechanical execution, reflected by coordinated trunk-lower limb motion, controlled vertical loading, reduced lateral corrections, and higher forward acceleration, even when longer trajectories were followed. These findings indicate that trajectory geometry alone is insufficient to explain performance outcomes and support the integration of wearable biomechanics with trajectory modeling as a practical, low-cost, and field-deployable tool for alpine skiing performance diagnostics.

As IoT environments continue to grow in scale and complexity, the increasing number of interconnected sensors and devices makes end-to-end system behaviour progressively harder to understand. Process mining offers strong potential to address this challenge by transforming sensor-driven event data into interpretable insights at the process level. Yet, current tools are typically designed for business processes, not sensor-driven IoT workflows, which raises questions about their suitability in the IoT context. This discrepancy is evident in existing comparative studies, which often rely on feature checklists, rarely consider usability and interaction effort, or fail to evaluate support for domain-specific analytical tasks. This study introduces a structured evaluation methodology that combines a functional capability assessment derived from vendor materials with a task-based evaluation grounded in 13 representative questions from an IoT-oriented smart factory scenario, focusing on clarity, ease of use, and the ability to address context-specific analytical needs. The results highlight notable strengths and trade-offs among the investigated tools, demonstrating substantial variation in usability, effort, and analytical coverage, and showing that no single tool fully supports the breadth of process-intelligence needs in IoT contexts. The proposed methodology provides a replicable foundation for evaluating process mining tools in domain-specific settings and supports more informed tool selection for IoT-driven analytical workflows.

Leaf area index (LAI) is a key indicator of crop growth and development and is widely used in both agricultural research and precision farming applications. PlanetScope imagery is generally used for monitoring crop growth due to its high revisit frequency, broad spatial coverage, and cost-effective access to consistent high-resolution multispectral data. Therefore, we developed regression models to estimate peanut LAI, combining PlanetScope spectral bands and vegetation indices (VIs). Specifically, we compared the performance of random forest (RF), eXtreme Gradient Boosting (XGBoost), and Partial Least Squares Regression (PLSR) regression algorithms for peanut LAI estimation. Our results showed that most of the VIs exhibited strong relationships with LAI. Thirteen VIs were individually evaluated for estimating LAI using the aforementioned algorithms, and our results showed that the best single predictors of LAI are: TSAVI (RF: R² = 0.87, RMSE = 0.83 m²/m², RRMSE = 24.20%; XGBoost: R² = 0.77, RMSE = 0.95 m²/m², RRMSE = 27.96%); and RTVIcore (PLSR: R² = 0.68, RMSE = 1.12 m²/m², RRMSE = 32.88%). The top six ranked VIs were used to calibrate the RF, XGBoost, and PLSR algorithms. Model validation indicated that RF achieved the highest accuracy (R² = 0.844, RMSE = 0.858 m²/m², RRMSE = 25.17%), followed by XGBoost (R² = 0.808, RMSE = 0.92 m²/m², RRMSE = 26.99%), whereas PLSR showed comparatively lower performance (R² = 0.76, RMSE = 0.983 m²/m², RRMSE = 28.85%). Further results showed that PlanetScope VIs provided superior model accuracy in estimating peanut LAI compared to the use of spectral bands alone. Additionally, integrating spectral bands with VIs reduced LAI estimation accuracy, underscoring the importance of selecting predictor variables in ensuring optimal model performance. Overall, the presented results are significant for future crop monitoring using RF to reduce overreliance on multiple models for peanut LAI estimation.

Passive radio frequency identification (RFID) sensing systems integrate wireless energy transfer with information identification. However, conventional passive RFID systems still face three key challenges in practical applications: low RF energy harvesting efficiency, high power consumption of sensor loads, and high complexity of tag anti-collision algorithms. To address these issues, this paper proposes a hardware-software co-optimized RFID sensor system. For hardware, low threshold RF Schottky diodes are selected, and an input inductor is introduced into the voltage multiplier rectifier to boost the signal amplitude, thereby enhancing the radio frequency to direct current (RF-DC) energy conversion efficiency. In terms of loading, a low-power management strategy is implemented for the power supply and control logic of the sensor node to minimize the overall system energy consumption. For algorithmic implementation, a Dual-Threshold Stepped Dynamic Frame Slotted ALOHA (DTS-DFSA) anti-collision algorithm is proposed, which adaptively adjusts the frame length based on the observed collision ratio, eliminating the need for complex tag population estimation. The algorithm features low computational complexity and is well suited for resource constrained embedded platforms. Through simulation validation, we compare the conversion efficiency of the RF energy harvesting circuit before and after improvement, the current of the sensor load in active and idle states, and the performance of the proposed algorithm against the low-complexity DFSA (LC-DFSA). The results show that the maximum conversion efficiency of the improved RF energy harvesting circuit has increased from 60.56% to 68.69%; specifically, the sensor load current drastically drops from approximately 2.0 mA in the active state to around 74 μA in the idle state, validating the efficacy of the proposed power gating strategy, and the proposed DTS-DFSA algorithm outperforms existing low-complexity schemes in both identification efficiency and computational complexity.

This study examines the efficacy of phasor analysis in distinguishing between healthy and diseased retinas using multispectral imaging data together with machine learning approaches. Our results demonstrate that phasor analysis of multispectral images surpasses average reflectance values in classification performance, serving as an effective dimensionality reduction technique to extract essential features, with the first harmonic yielding optimal results when paired with Z-score normalization. To compare the effectiveness of multispectral images with that of a conventional color fundus camera, we extracted three spectral bands corresponding to the red, green, and blue regions and combined them to create RGB-like images, which were then subjected to the same analysis. Our study found that phasor analysis of multispectral images provided more accurate classification results than phasor analysis of RGB-like images. An examination of different regions of interest showed that using the entire retina yields the best classification performance, likely due to the advanced stage of the diseases, which had progressed to affect the entire fundus. Our findings suggest that phasor analysis of multispectral images and machine learning are a powerful tools for retinal disease classification.

Paper-based analytical devices have emerged as a versatile and cost-effective platform for on-site chemical and biological analysis. The integration of fluorescence detection with smartphone imaging has significantly enhanced the analytical performance and portability of these systems, enabling sensitive, rapid, and user-friendly detection of diverse analytes. This review highlights recent advancements in paper-based fluorescence sensing technologies, focusing on their design principles, materials, and detection strategies. Emphasis is placed on the use of nanomaterials, quantum dots, and carbon-based fluorophores that improve sensitivity and selectivity in food, bioanalytical, and environmental applications. The role of smartphones as optical detectors and data processing tools is explored, underscoring innovations in image analysis, calibration algorithms, and app-based quantification methods.

Accurate anomaly detection in rotating machinery under noisy conditions remains challenging in Prognostics and Health Management (PHM). Existing deep learning autoencoders and attention mechanisms rely primarily on data-driven similarity measures and fail to explicitly incorporate nonlinear dynamical characteristics of degradation. In this study, we propose a Recurrence Quantification Analysis-Aware Attention (RQAA) framework that systematically injects chaos-theoretic descriptors into the attention mechanism of LSTM-based autoencoders for unsupervised anomaly detection. Specifically, RQA metrics including recurrence rate, determinism, laminarity, entropy, and trapping time are computed at the window level and embedded into the query-key-value attention scoring to guide the model toward dynamically informative temporal patterns. Three attention variants are developed to investigate different fusion strategies between learned representations and RQA-driven structural cues. The proposed framework is evaluated on three widely used bearing vibration datasets, which are IMS, CWRU, and HUST. Experimental results demonstrate that RQAA consistently outperforms conventional LSTM autoencoders and classical attention-based models, achieving up to 99.85% F1-score and 99.00% AUC while exhibiting superior robustness in low signal-to-noise scenarios. Further analysis reveals that explicit dynamical guidance enhances anomaly separability and reduces false alarms, particularly in early-stage fault detection. These findings indicate that integrating nonlinear dynamical information directly into attention scoring offers a principled and effective pathway for advancing unsupervised anomaly detection in rotating machinery and safety-critical industrial systems.

The performance of a printed square split-ring resonator as a sensor for quantifying nanoparticle concentrations in PVDF-based nanocomposites was evaluated at UHF frequencies. The sensing mechanism was based on the frequency response of parameter S₂₁, observing the shift in the resonant frequency and a variation in S₂₁ level, while samples were placed on the ring split and compared to the sensor without a sample. Experiments with samples of PVDF-based nanocomposites combined with different concentrations of both MoS₂ and MXenes, ranging from 0.01% to 0.2%, were conducted. In general, considering both types of samples studied, it was observed that, as the concentration increases, S₂₁ (dB) increases from -6.35 to -6 dB. At the same time, the resonance frequency in the S₂₁ plot went from 500.4 to 498.25 MHz. Although the concentrations and their variations were relatively low, shifts in the resonance frequency of S₂₁ were evident, demonstrating the ability of the sensor to detect low concentrations and variations of MoS₂ and MXenes, being the detection of samples with higher concentrations feasible as future work, and concluding that the sensor had a relatively acceptable performance. In this study, MXenes were the concentrations that produced more noticeable shifts in the resonance frequency of S₂₁. Likewise, characterizations based on SEM and TEM were performed to corroborate the ones at UHF frequencies.