Journal bearings are critical components in a wide range of rotating machinery, from internal combustion engines and industrial turbines to wind turbine drivetrains, where accurate monitoring of the lubrication regime is essential for predictive maintenance and efficiency optimisation. This study investigates the classification of lubrication regimes (boundary, mixed, and hydrodynamic) in a journal bearing test system using acoustic emission (AE) and knock sensors, processed by an artificial neural network (ANN) supported by a genetic algorithm (GA) for feature selection. Sensor signals were segmented and characterised using power spectral, hit rate, and wavelet-based features, with lubrication regime labels derived from the lambda ratio. The AE-sensor achieved a classification accuracy of 95.02%, with predicted regime transitions closely matching those of the lambda ratio, while the knock sensor performed substantially worse at 67.18%, attributed to its lower operating frequency range. The trained ANN generalised successfully to previously unseen operating conditions, including a different oil temperature and sensor configuration, without retraining. GA-based feature selection provided a measurable improvement in classification accuracy over random selection at reduced feature set sizes, supporting its use in computationally constrained applications. These findings demonstrate that AE sensing combined with machine learning offers a robust and adaptable method for real-time lubrication regime monitoring in journal bearings.
In precision manufacturing, accurate prediction of tool wear is pivotal for maintaining production quality and efficiency. This paper introduces a novel predictive model that employs a multi-branch convolutional neural network (multi-CNN) enhanced by an attention mechanism and Bayesian optimisation to forecast tool wear with high accuracy. The proposed approach is evaluated using the publicly available dataset, which comprises force and torque measurements acquired from machining processes. By incorporating time-domain (TD), frequency-domain (FD), and discrete wavelet transform (DWT) features, the model comprehensively analyses signals from machining processes, enabling robust predictions. The attention mechanism strategically emphasises significant features, augmenting the model’s predictive capability. Demonstrated results show superior performance with a mean absolute error (MAE) of 2.8, root mean squared error (RMSE) of 3.4, and an R2 score of 0.986, significantly outperforming conventional models. Additionally, post-training quantisation optimises the model for deployment on edge computing devices, maintaining effectiveness while being lightweight, ideal for real-time applications in resource-constrained environments. This study exemplifies how advanced machine learning techniques can improve predictive maintenance in manufacturing which not only enhances operational efficiency but also reduces downtime.
This study evaluates the metrological performance and statistical reliability of a laser distance measurement system developed for convergence monitoring in the Kłodawa Salt Mine S.A. The standard deviation of single distance measurements was determined for sighting lengths from a few to nearly 50 m under controlled laboratory conditions and during underground field tests. Measurement uncertainty propagation was applied to determine the minimum detectable annual change in baseline length at confidence levels of 95–99%. Laboratory measurements demonstrated high stability, yielding standard deviations between 0.18 and 0.38 mm. Despite the challenging underground environment, field measurements maintained standard deviations below 0.7 mm for baselines up to several tens of meters. The analysis shows that for baselines ≤30 m, the system enables detection of convergence as small as 2 mm/year, with detectability increasing nearly linearly to 5 mm/year at 43 m and 8 mm/year at 49 m. Comparison with electronic total stations confirmed comparable performance on short baselines, with the laser system offering advantages in operational efficiency and setup simplicity. The results represent the first comprehensive metrological validation of a laser distance measurement system used in deep underground mining conditions and demonstrate its suitability for long-term, high-precision displacement monitoring in deforming rock masses.
This study presents the development of a robust RF sensing system to address design as well as measurement challenges such as operating bandwidth, gain, near field coupling, and external factors during measurement in industrial ultra-wideband (UWB) applications. The proposed system integrates UWB radio sensors, specialized dielectric loads, and a 3D-printed refractive index lens to enhance overall sensing accuracy and performance. The system was designed, manufactured, and prototyped specifically for non-destructive testing, with a focus on accurately determining the complex refractive index (n + jk) of various industrial materials. The sensors with dielectric loads were experimentally tested to examine the near field response. The lens increases the antenna gain from 3.1 to 4.8 dBi in the azimuth plane and from 2.5 to 4.3 dBi in the elevation plane, resulting in higher received signal strength and an improved signal-to-noise ratio. The system was experimentally validated using solid wood samples with moisture contents ranging from 1% to 11% over the 1.5–2.5 GHz frequency band. The enhanced antenna gain enabled more reliable complex refractive index estimation despite the increased dielectric losses associated with higher moisture content. The lens suppresses near field and coupling effects, such that unbiased values for n and k are obtained. Using the lens gives a moderate decrease in relative error, Δn/n from 0.09 to 0.08, and a significant decrease in Δk/k from 0. 18 to 0.13. Polarimetric measurements were also performed to account for the anisotropic properties of wood. The results demonstrate that the proposed sensing platform provides accurate and repeatable refractive index characterization. For parallel polarization, n ranges from 1.45 to 1.82, while k varies from 0.23 to 0.46. For perpendicular polarization, n ranges from 1.28 to 1.55, and k from 0.10 to 0.36, demonstrating the capability of the proposed system to characterize the anisotropic dielectric properties of wood. The results were further compared with previously reported far-field measurements on solid wood (same samples), where the incident EM waves are approximately planar, demonstrating good agreement and validating the proposed near-field UWB sensing approach.
Dimensional metrology is moving from isolated, post-process inspection to integrated, intelligent measurement within digital manufacturing. In the context of Industry 4.0 and 5.0, smart metrology spans tactile, optical, and hybrid modalities embedded in cyber-physical production environments. Standardized information models such as STEP AP242, QIF, STEP-NC, and OPC UA connect design, manufacturing, and quality assurance to establish an interoperable digital-thread with lifecycle traceability. This review synthesizes recent developments in software and data integration, automation and robotics, and artificial intelligence for metrology, and introduces a structured taxonomy and conceptual framework that clarify how these technologies interact across technical, syntactic, semantic, and organizational layers. It highlights how model-based definition, architectures that combine edge and cloud, and robot-assisted inspection enable real-time, closed-loop quality control. A comparative analysis clarifies the shift from static verification to adaptive measurement networks. Persistent challenges include semantic interoperability across heterogeneous tools, real-time uncertainty evaluation, and trustworthy AI that remains explainable and auditable, alongside human-centric design requirements. The paper concludes with practical recommendations for industrial deployment and a research agenda focused on online uncertainty estimation, interoperable ontologies, digital-twin-driven control, and credible sustainability metrics. Smart metrology is emerging as a cornerstone of resilient and sustainable manufacturing by linking standardized data, automation, and human oversight.
Transimpedance amplifiers are current-to-voltage converters. They are found in a wide variety of applications that require current measurements at high sensitivity, low noise and/or high bandwidth. Inherent to the construction of such an amplifier are certain tradeoffs between bandwidth, amplification, and noise level, requiring a careful design and layout of a transimpedance amplifier for any given task. These design limitations become especially relevant for high-gain, high-bandwidth applications, such as photodiode readout or ion current sensing. In these applications, commercially available amplifiers are often not suited for the task, leading to a low signal-to-noise ratio. This paper discusses the most significant limiting factors of transimpedance amplifiers, and describes their effects, source and possible mitigation options. It presents a novel combination of feedback resistor configuration and shielding layout to improve performance beyond typical single-resistor shielded options. Based on this, a low-cost transimpedance amplifier made with readily available off-the-shelf components reaching 30 kHz bandwidth and 1 GΩ amplification with only 3.898 pA root-mean-square noise was developed, characterized and demonstrated for use in ion mobility spectrometry as one exemplary application requiring high gain, low noise and high bandwidth.
Vector vortex beams (VVBs), which simultaneously carry spin angular momentum (SAM) and orbital angular momentum (OAM), exhibit spatially varying polarization distributions dependent on the azimuthal angle. These properties make VVBs highly valuable in advanced optical applications, particularly in polarization-resolved sensing and Light Detection and Ranging systems, where they enable directional surface detection. However, conventional VVBs generated by superimposing vortex beams with differing topological charges suffer from beam radius dependence on the charge number and structural instability during propagation, limiting their practical utility. In this work, we demonstrate the generation of perfect vector vortex beams (PVVBs) using a fully dielectric Jones matrix metasurface platform based on phase-only modulation. Unlike conventional vortex beams, PVVBs feature a charge-independent beam radius, offering enhanced stability and uniformity. By designing a series of single-layer Jones matrix metasurfaces operating at 973 nm, we realize the emission of radially and azimuthally polarized PVVBs. Moreover, by integrating perfect vortex phase profiles with additional functional phase distributions, we generate advanced structured light fields, including PVVB arrays and holographic PVVBs. These metasurface-generated beams provide robust, compact, and tunable vector light sources, opening new opportunities for integration into miniaturized polarization-resolved sensing systems and high-precision optical manipulation platforms, expanding the potential application range of polarization-resolved surface orientation sensing systems and micro-manipulation systems.
Robotic positioning plays a crucial role in enabling accurate navigation, control, and autonomy across diverse domains, including industrial automation, autonomous vehicles, unmanned aerial vehicles (UAVs), underwater exploration, swarm robotics, and biomedical systems. Despite notable advancements, challenges such as sensor drift, environmental variability, computational complexity, and energy constraints continue to affect positioning accuracy. To address these limitations, diverse measurement systems and sensor fusion techniques have been developed, integrating inertial measurement units (IMUs), Global Positioning System (GPS), Light Detection and Ranging (LIDAR), optical motion capture, ultrasonic sensors, radio-frequency identification (RFID), and infrared (IR) tracking. This review introduces a structured classification of positioning systems into on-board, out-board, and hybrid types, highlighting trade-offs and application-specific strengths. Measurement methods are categorized into relative and absolute approaches. Furthermore, artificial intelligence (AI)-driven fusion strategies such as Kalman filtering, particle filtering, visual–inertial odometry (VIO), and simultaneous localization and mapping (SLAM) are analyzed for their roles in enhancing robustness and mitigating drift. Evaluation is based on accuracy, latency, efficiency and adaptability. Cross-domain comparisons illustrate how sensor-algorithm integration impacts outcomes across aerial, mobile, underwater, industrial and biomedical platforms. Finally, emerging directions such as 6G-enabled ultra-precise localization, neuromorphic computing for low-power SLAM, and blockchain-based decentralized frameworks are proposed to improve trust, adaptability, and reliability in next-generation robotic autonomy.
- This study presents a novel, high-precision pipeline leak detection and localisation method that integrates the Transient Reflection Method (TRM) with Mel-Frequency Cepstral Coefficients (MFCC) and a lightweight Artificial Neural Network (ANN) for leak size estimation. Unlike conventional time-domain or FFT-based approaches, the proposed method uses MFCC to characterise the spectral signature of transient pressure wave reflections caused by leak-induced impedance discontinuities, rather than relying on leak-generated frequency shifts. Laboratory experiments on a 152-m Medium-Density Polyethene (MDPE) pipe system achieved an average localisation error of ±1.98 m and 96.5% leak detection sensitivity for leaks as small as 1 mm. The ANN regression model demonstrated high predictive reliability, producing a mean absolute error (MAE) of 0.15 mm for leak size estimation. Field validation of a 220-m buried MDPE pipeline yielded comparable performance, maintaining a localisation error of ±2.12 m and 94.1% detection sensitivity, confirming practical scalability under real municipal operating conditions. A structured preprocessing pipeline including signal normalisation and MFCC feature vectorisation proved essential for model stability, improving leak size MAE by more than 12% post-normalisation while preserving reflection-based spectral patterns. The system requires only single-point hydrant access, uses minimal portable hardware, and avoids the computational overhead typical of deep convolutional models. These results confirm a cost-effective, portable, and scalable solution for early leak detection and quantification in pressurised water pipeline systems, offering high spatial resolution and strong robustness for field deployment.
A microwave photonic chaotic radar with frequency up/down conversion capability is proposed and demonstrated to implement high-resolution ranging. The system generates broadband chaotic signals via a chaotic optoelectronic oscillator (OEO), offering superior range resolution and anti-jamming performance compared to deterministic waveforms. By leveraging a dual-polarization quadrature phase shift keying modulator (DP-QPSKM), the system integrates radar signal generation with photonic-assisted frequency conversion. The radar operating frequency can be adjusted to the Ku-band without increasing the workload of digital correlation reception. The microwave photonic frequency converter avoids the bandwidth limitations and remote transmission flexibility issues associated with electric mixers. A polarization division multiplexing (PDM) scheme is introduced to co-transmit the reference and down-converted echo signals between the remote and central stations, which ensures system coherence and avoids strict clock synchronization. Experimental results demonstrate a radar bandwidth of 6 GHz, achieving a range resolution of 2.55 cm with an error of less than 0.35 cm. The proposed method offers a flexible and robust solution for distributed sensing in autonomous driving and smart city applications.
For researchers in medical or social psychology, facial emotion knowledge is crucial as it serves as valuable interpretive data. Facial emotion information not only helps in understanding a person's inner thoughts but also enables observation of their likes and dislikes, which is essential for further exploration in medical and social psychology research. This is particularly important when patients are unable to express their thoughts through words or gestures, making facial emotions a valuable tool for doctors to assess and inquire about the condition of patients. In this study, we utilized the facial emotion dataset provided by AffectNet, which comprises over 450,000 photos categorized into 11 different emotions, as the sample data for neural network learning. We proposed a model called AlexNet_Plus_LSTM, which combines convolutional and recurrent layers. Specifically, we incorporated long and short-term memory networks (LSTM), commonly used in time-series situations, to replace the fully connected layer in our proposed model. The model utilizes convolutional neural network (CNN) to learn facial emotion features from static images, and these features are then fed into LSTM to learn the relationships between different facial emotion characteristics, thus improving the accuracy of facial emotion recognition. The results obtained from our research demonstrate that the AlexNet_Plus_LSTM model achieves an accuracy of 76.83%. This research contributes to the field of facial emotion recognition by proposing a novel model that combines convolutional and recurrent layers, and incorporating LSTM to improve accuracy. The findings have potential applications in medical and social psychology research, where facial emotion knowledge can provide valuable insights into understanding human emotions and behavior in various contexts.
Accurate real-time indoor environmental monitoring and prediction is essential for proactive control to protect occupant health and comfort. While prior studies have demonstrated real-time monitoring and AI-based prediction, real-time operation is often treated as near-real-time and evaluated using stored datasets rather than as a continuously operating, closed-loop streaming inference process. To address this practical and methodological gap, this study presents the initial phase of the Information and Communication Technology-enabled indoor air (ICTAir) framework, a real-time approach that integrates lightweight Internet of Things (IoT) technologies with hybrid deep learning (DL) model for continuous monitoring and prediction under a live closed loop deployment, while maintaining superior prediction accuracy with moderate inference-time and memory requirements compared to statistical and traditional machine learning approaches. ICTAir integrates low-cost sensors, a microcontroller, and a cloud back end, and employs an attention-enhanced gated recurrent unit with a simplified attention mechanism (GRU-SAM). The proposed GRU-SAM model is benchmarked against statistical baselines, traditional machine-learning models, and deep-learning alternatives, demonstrating consistently superior predictive accuracy. With a 1-h training window and a 5-min prediction horizon, GRU-SAM achieves low error (e.g., temperature RMSE = 0.22). Calibration experiments confirm sensor reliability and indicate a marked improvement in PM2.5 measurements after calibration. During deployment, temperature frequently exceeded 30 °C and relative humidity surpassed 60%, while PM2.5 and PM10 often crossed 35 and 50 μg/m3, respectively. Continuous operation achieved end-to-end system latency within 1 min while maintaining 97.7% streaming inference reliability. The proposed framework provides a practical, deployment-oriented foundation for future autonomous indoor environmental control systems.
Optical gas-sensing technologies offer distinct advantages and limitations, making them essential for applications in environmental monitoring, industrial safety, and healthcare. This review presents a comparative analysis of optical gas sensors classified according to their operating mechanisms (spectroscopic, chemical, and electronic) and technological platforms (fiber-optic, dye-based, carbon-based, and waveguide structures). Sensor performance is evaluated using key metrics, including sensitivity, selectivity, response time, environmental stability, durability, fabrication cost, and calibration requirements. Spectroscopic techniques such as tunable diode laser absorption spectroscopy (TDLAS) and Fourier-transform infrared (FTIR) spectroscopy achieve exceptional sensitivity (<1 ppm) and molecular selectivity, but often require complex calibration procedures and costly instrumentation. Fiber-optic and waveguide-based sensors demonstrate high stability and rapid response times (often <1 s), making them well suited for real-time and remote monitoring. Dye-based and chemical sensors offer cost-effective and selective solutions, although their long-term stability may be limited by material degradation. Electronic and carbon-based sensors, including metal-oxide semiconductors, carbon nanotubes, and graphene, provide high sensitivity and robustness, yet face challenges related to cross-sensitivity and large-scale fabrication. Overall, this comparative assessment highlights the performance trade-offs among different sensing mechanisms and platforms, clarifying their suitability for specific application scenarios. Future research should focus on material engineering, hybrid optical–electrical sensing architectures, scalable fabrication strategies, and advanced signal-processing approaches to enhance long-term stability, reduce system complexity, and accelerate the deployment of next-generation optical gas-sensing technologies.
This paper presents simulation models developed for investigating singlemode-multimode-singlemode (SMS) structures, with the intention of utilising precise nano-scale 3D printing to print the center-section of the structures modelled. The models allow the design of structures for use in SMS fiber sensors, replacing the silica fiber center-section structures commonly used, with 3D printed elements. The paper discusses the process of evaluating and validating COMSOL propagation models by comparison to existing reported results and proposes the use of SMS self-imaging length and the transmission spectra as useful metrics for model comparisons. Both 2D and 3D COMSOL models are developed, and both show good agreement in calculating self-imaging length with the other referenced models. In particular, the 3D model not only simulates the self-imaging length with high accuracy, but also shows good spectral agreement with the referenced models and an analytical calculation. In addition, results from previously published work by the group are used for comparison with the 3D COMSOL model.
The advancement of agricultural automation intensifies demands for highly reliable visual recognition systems in orchard harvesting robots. However, it is difficult to achieve robust fruit detection for balancing high accuracy and real-time performance under complex orchard conditions with variable illumination, occlusions, and phenotypic diversity. This study proposes a collaborative framework integrating adaptive image processing with heterogeneous model inference. The methodology begins with Lab color space conversion to ensure illumination invariance. It further utilizes dual-threshold HSV segmentation for handling both red and green apples, alongside morphological optimization with elliptical structuring elements to address occlusion. A novel architecture allocates real-time screening to an embedded Random Forest (RF) classifier and precise localization to a host-based lightweight YOLOv5 model through fused color-morphological features. Experimental results demonstrate that morphological feature enhancement consistently outperforms color-based approaches across both models. The models with dual-feature input achieves optimal performance that the RF classifier attains accuracy of 83.35%, while lightweight YOLOv5 reaches 98.90% accuracy. Quantitative analysis reveals dual-feature fusion with color and morphology improving all metrics by over 3% compared to non-enhanced baselines. Notably, the observed accuracy improvement exceeded the sum of gains from individual features, confirming a synergistic effect and proving the necessity of feature fusion. This work provides a computationally viable solution for reliable apple recognition in unstructured environments.
The transition to truly smart cities demands more than layered technologies; it requires convergent intelligence that unifies physical sensing, adaptive control, and ethical security. This editorial paper brief about the special issue entitled “Measurement, Control and Security of Systems for Smart Cities”. As smart cities are among the most active research areas, the call for papers for this special issue, “VSI: Systems for Smart Cities” attracted a wide range of manuscripts spanning multiple disciplines, including Computer Science, Electrical and Electronics Engineering, Communication Engineering, Civil Engineering, Mechanical Engineering, Urban Construction, Artificial Intelligence and Machine Learning, Cybersecurity, and Renewable Energy etc. Notably, authors from 16 different countries, including Algeria, Chile, China, Egypt, England, Ethiopia, India, Iran, Jordan, Kosovo, Nigeria, Portugal, Saudi Arabia and United Arab Emirates have contributed their research articles. In this editorial, we synthesize insights from all 30 published articles to present a holistic vision of next-generation urban ecosystems. Together, these works span renewable energy, structural health, healthcare, transportation, noise pollution, public space, and data security, these works collectively redefine what it means to build intelligent infrastructure that is sustainable, equitable, resilient, and trustworthy. We organize these advances around three interwoven pillars: (1) Measurement for Awareness, (2) Control for Adaptation, and (3) Security and Ethics for Trust-and demonstrate how fractional-order controllers, lightweight crack detectors, edge-based triage systems, multimodal transport models, and secure data-mining frameworks all contribute to a human-centered urban future. This synthesis serves as both a technical roadmap and a philosophical compass for the responsible evolution of smart cities.
