Search Results (16,917)

Multi-sensor fire detection technology has been widely adopted in practical applications; however, existing methods still suffer from high false alarm rates and inadequate adaptability in complex environments due to their limited capacity to capture deep time-series dependencies in sensor data. To enhance robustness and accuracy, this paper proposes a novel model named BiLSTM-LN-SA, which integrates a Bidirectional Long Short-Term Memory (BiLSTM) network with Layer Normalization (LN) and a Self-Attention (SA) mechanism. The BiLSTM module extracts intricate time-series features and long-term dependencies. The incorporation of Layer Normalization mitigates feature distribution shifts across different environments, thereby improving the model’s adaptability to cross-scenario data and its generalization capability. Simultaneously, the Self-Attention mechanism dynamically recalibrates the importance of features at different time steps, adaptively enhancing fire-critical information and enabling deeper, process-aware feature fusion. Extensive evaluation on a real-world dataset demonstrates the superiority of the BiLSTM-LN-SA model, which achieves a test accuracy of 98.38%, an F1-score of 0.98, and an AUC of 0.99, significantly outperforming existing methods including EIF-LSTM, rTPNN, and MLP. Notably, the model also maintains low false positive and false negative rates of 1.50% and 1.85%, respectively. Ablation studies further elucidate the complementary roles of each component: the self-attention mechanism is pivotal for dynamic feature weighting, while layer normalization is key to stabilizing the learning process. This validated design confirms the model’s strong generalization capability and practical reliability across varied environmental scenarios. Full article

This manuscript presents a robust formation and collision-free containment control system designed for a quadrotor fleet operating under turbulent wind conditions. Emphasizing collision avoidance, we introduce a two-layer strategy in which a virtual leader defines a trajectory, and leaders and followers maintain their positions while avoiding collisions among them. A graph convention is used to illustrate the roles of leaders and followers, as well as their interactions. Inter-agent collision avoidance is proposed by expanding the desired distance relative to all neighboring agents, thereby guaranteeing the convergence stage. Moreover, the approach employs a class of adaptive sliding mode strategies to ensure finite-time convergence, as well as non-overestimation of the control gain in the presence of uncertainties and perturbations. A stability analysis demonstrates the practical finite-time stability of the system using the Lyapunov methodology. Results from the simulation underscore the effectiveness of our proposal in adhering to the desired time-varying trajectories and ensuring sensor-less inter-agent collision avoidance for the followers, even in the presence of turbulent wind conditions. Full article

This paper presents an electromagnetic levitation system that stabilizes a magnetic body using an array of electromagnets controlled by a Hall-effect sensor array and TinyML-based position detection. Departing from conventional optical tracking methods, the proposed design combines finite-element-optimized electromagnets with a microcontroller-optimized neural network that processes sensor data to predict the levitated object’s position with 0.0263–0.0381 mm mean absolute error. The system employs both quantized and full-precision implementations of a supervised multi-output regression model trained on spatially sampled data (40 × 40 × 15 mm volume at 5 mm intervals). Comprehensive benchmarking demonstrates stable operation at 850–1000 Hz control frequencies, matching optical systems’ performance while eliminating their cost and complexity. The integrated solution performs real-time position detection and current calculation entirely on-board, requiring no external tracking devices or high-performance computing. By achieving sub 30 μm accuracy with standard microcontrollers and minimal hardware, this work validates machine learning as a viable alternative to optical position detection in magnetic levitation systems, reducing implementation barriers for research and industrial applications. The complete system design, including electromagnetic array characterization, neural network architecture selection, and real-time implementation challenges, is presented alongside performance comparisons with conventional approaches. Full article

Self-oscillating systems are capable of transforming ambient energy directly into mechanical output, and exploring novel designs is of great value for energy harvesters, actuators, and engine applications. The inspiration for this study is drawn from the four-stroke engine; we designed a new self-rotating engine formed by a turnplate, a hinge, and an LCE fiber, operating with steady illumination applied. To analyze its rotation dynamics, a nonlinear theoretical framework was formulated constructed with the dynamic LCE model as a framework. The central discovery is that the light-driven LCE engine can operate in three distinct states under steady illumination—static equilibrium, pendulum-like oscillation and sustained self-rotation—switching between them through a supercritical Hopf bifurcation. The persistence of both the pendulum and rotary motions stems from an energy balance in which the positive work produced by photo-induced contraction of the LCE fiber is exactly offset by damping dissipation, while oscillation amplitude and rotation frequency are strongly governed by light intensity, contraction coefficient, damping coefficient, spring constant and turntable radius. Compared with many previously reported self-oscillating designs, the present self-rotating engine is distinctive for its lightweight and simple configuration, tunable size, and rapid operation. These features enable compact integration and broaden its potential applications in micro-scale systems and devices. The advancement in artificial muscles, medical instruments and micro sensors is strongly promoted by this, making it possible to create devices that are both smaller in size and superior in functionality. Full article

Children with intoeing gait are at increased risk of tripping and consequent injury, reduced mobility, and psychological issues. Quantification of tripping is needed outside the gait lab during daily life for improved clinical assessment and treatment evaluation and to enrich the database for artificial intelligence (AI) learning. This paper presents the development of a low-cost, wearable tripping monitor to log a child’s Tripping Hazard Events (THEs) and steps taken during two weeks of everyday activity. A combination of sensors results in a high probability of THE detection, even during rapid gait, while guarding against false positives and minimizing power and therefore monitor size. A THE is logged when the feet come closer than a predefined threshold during the intoeing foot swing phase. Foot proximity is determined by a Radio Frequency Identification (RFID) reader in “sniffer” mode on the intoeing foot and a target of passive Near-Field Communication (NFC) tags on the contralateral foot. A Force Sensitive Resistor (FSR) in the intoeing shoe sets a time window for sniffing during gait and enables step counting. Data are stored in 15 min epochs. Laboratory testing and an IRB-approved human participant study validated system performance and identified the need for improved mechanical robustness, prompting a redesign of the monitor. A custom Python (version 3.10.13)-based Graphical User Interface (GUI) lets clinicians initiate recording sessions and view time records of THEs and steps. The monitor’s flexible design supports broader applications to real-world activity detection. Full article

This paper addresses the suppression of multi-frequency line spectrum vibrations during the operation of indoor substations through the development of an active vibration isolation scheme based on a piezoelectric stack inertial actuator. Based on the finite element modal analysis, the excitation frequencies that strongly influence structural response were identified, and the excitation points and sensor layout strategies were determined under a collocated control configuration. Subsequently, the actuator’s structural design and system integration were carried out. Experimental results demonstrate vibration amplitude reductions of 18.75 dB at 100 Hz, 46.02 dB at 200 Hz, and 32.04 dB at 300 Hz, respectively, validating the effectiveness of the proposed method in controlling line spectrum vibrations at multiple frequencies. The study shows that the coordinated optimization of modal matching and the dynamic response capability of inertial actuators provides experimental evidence and technical guidelines for active vibration isolation in large plate-shell structures. Full article

Semiconductor wafer manufacturing is one of the most complex and data-intensive industrial processes, comprising 500–1000 tightly interdependent steps, each requiring nanometer-level precision. As device nodes approach 3 nm and beyond, even minor deviations in parameters such as oxide thickness or critical dimensions can lead to catastrophic yield loss, challenging traditional physics-based control methods. In response, the industry has increasingly adopted regression analysis and predictive modeling as essential analytical frameworks. Classical regression, long used to support design of experiments (DOE), process optimization, and yield analysis, has evolved to enable multivariate modeling, virtual metrology, and fault detection. Predictive modeling extends these capabilities through machine learning and AI, leveraging massive sensor and metrology data streams for real-time process monitoring, yield forecasting, and predictive maintenance. These data-driven tools are now tightly integrated into advanced process control (APC), digital twins, and automated decision-making systems, transforming fabs into agile, intelligent manufacturing environments. This review synthesizes foundational and emerging methods, industry applications, and case studies, emphasizing their role in advancing Industry 4.0 initiatives. Future directions include hybrid physics–ML models, explainable AI, and autonomous manufacturing. Together, regression and predictive modeling provide semiconductor fabs with a robust ecosystem for optimizing performance, minimizing costs, and accelerating innovation in an increasingly competitive, high-stakes industry. Full article

In this study, two types of gas sensors—silicone-based interdigital electrode and silicon micropillar sensors based on rGO and rGO/SnO₂—were fabricated. Their gas-sensing performance was investigated at room temperature. First, interdigital electrodes of different channel widths were fabricated to investigate the impact of the channel width parameter. Subsequently, the rGO/SnO₂ doping ratio in the composite material was varied to identify the optimal composition for gas sensitivity. Additionally, triangular and square-arrayed silicon micropillar substrates were fabricated via photolithography and inductively coupled plasma etching. The rGO/SnO₂-based gas sensor on a silicon micropillar substrate exhibited an ultra-high specific surface area. The triangular micropillar arrangement of rGO/SnO₂-160 demonstrates the best performance, showing approximately 14% higher response and a 106 s reduction in response time compared with interdigital electrode sensors spray-coated with the same concentration of rGO/SnO₂ when tested at room temperature under 250 ppm NO₂. The optimized sensor achieves a detection limit as low as 5 ppm and maintains high responsiveness, even in conditions of 60% relative humidity (RH). Additionally, the repeatability, selectivity, and stability of the sensor were evaluated. Finally, structural and morphological characterization was conducted using XRD, SEM, TEM, and Raman spectroscopy, which confirmed the successful modification of rGO with SnO₂. Full article

Ever-increasing demands to improve fuel burn efficiency of aero gas turbines lead to rises in fuel system pressures and temperatures, posing challenges for the structural integrity of the pump housing and creating internal deflections that can adversely affect volumetric efficiency. Non-invasive strain and vibration measurements could allow transient effects to be quantified and considered during the design process, leading to more robust fuel pumps. Fuel pumps used on a high bypass turbofan engine were instrumented with optical fibre Bragg grating (FBG) sensors, strain gauges and thermocouples. A hydraulic hand pump was used to facilitate measurements under static conditions, while dynamic measurements were performed on a dedicated fuel pump test rig. The experimental data were compared with the outputs from a finite element (FE) model and, in general, good agreement was observed. Where differences were observed, it was concluded that they arose from the sensitivity of the model to the selection of nodes that best matched the sensor location. Strain and vibration measurements were performed over the frequency range of 0 to 2.5 kHz and demonstrated the ability of surface-mounted FBGs to characterise vibrations originating within the internal sub-components of the pump, offering potential for condition monitoring. Full article

This paper introduces an autonomous vision-based tracking system for a quadrotor unmanned aerial vehicle (UAV) equipped with an onboard camera, designed to track a maneuvering target without external localization sensors or GPS. Accurate capture of dynamic aerial targets is essential to ensure real-time tracking and effective management. The system employs a robust and computationally efficient visual tracking method that combines HSV filter detection with a shape detection algorithm. Target states are estimated using an enhanced extended Kalman filter (EKF), providing precise state predictions. Furthermore, a closed-loop Proportional-Integral-Derivative (PID) controller, based on the estimated states, is implemented to enable the UAV to autonomously follow the moving target. Extensive simulation and experimental results validate the system’s ability to efficiently and reliably track a dynamic target, demonstrating robustness against noise, light reflections, or illumination interference, and ensure stable and rapid tracking using low-cost components. Full article

An incision point acts as the pivot point when a minimally invasive surgery procedure is applied. The assistive robot arms employed for such operation must have the capability to perform a remote center of motion (RCM) at this pivot point. Other than designing RCM mechanisms, a common practice is to use a readily available spatial serial robot arm and control it to impose this RCM constraint. When this assistive robot is required to be backdriven by the surgeon, the relation between the interaction forces/moments and the motion with RCM constraint becomes challenging. This paper carefully formulates a hybrid position/force control scheme for this relationship when any readily available robot arm that is coupled with a force/torque sensor is used for an RCM task. The verification of the formulation is carried out on a readily available robot arm by implementing the additional constraints that are derived from a surgical robot application. Full article

With the increased demand for wearable sensors, image representations—such as persistence images and Gramian angular fields—transformed from time-series data have been investigated to address challenges in wearables arising from physiological variations, sensor noise, and limitations in capturing contextual information. To preserve the lightweight structural design of models, knowledge distillation (KD) has also been employed alongside image representations during training to distill smaller and more efficient models. Although image representations play a key role in providing richer and more informative features in training a model, their effectiveness within the KD framework has not been thoroughly explored. In this paper, we focus on image representation-driven KD to investigate whether these representations can provide useful knowledge leading to improved time-series interpretation in activity classification tasks. We explore the benefits of integrating image representations into KD, and we analyze the interplay between representation richness and model compactness with different combinations of teacher and student networks. We also introduce diverse KD strategies to utilize image representations, and we demonstrate the strategies with various perspectives, such as analysis of noises, generalizability, and compatibility, across datasets of varying scales to obtain comprehensive and insightful observations. These offer valuable insights for designing efficient and high-performance wearable sensor-based systems. Full article

In tightly coupled Industrial Control Systems (ICS), abnormal disturbances often propagate throughout the process, triggering a large number of time-correlated alarms that exceed the handling capacity of the operator. Consequently, a key challenge is how to leverage the directional and temporal characteristics of disturbance propagation to alleviate alarm overload. This paper proposes a delay-sensitive causal inference approach for industrial alarm analysis to address this problem. On the one hand, time delay estimation is introduced to precisely align the responses of two sensor sequences to disturbances, thereby improving the accuracy of causal relationship identification in the temporal domain. On the other hand, a multi-scale subgraph fusion strategy is designed to address the inconsistency in causal strength caused by disturbances of varying intensities. By integrating significant causal subgraphs from multiple scenarios into a unified graph, the method reveals the overall causal structure among alarm variables and provides guidance for alarm mitigation. To validate the proposed method, a case study is conducted on the Tennessee Eastman Process. The results demonstrate that the approach identifies causal relationships more accurately and reasonably and can effectively reduce the number of alarms by up to 51.6%. Full article

Offline reinforcement learning suffers from critical vulnerability to data corruption from sensor noise or adversarial attacks. Recent research has achieved a lot by downweighting corrupted samples and fixing the corrupted data, while data corruption induces feature entanglement that undermines policy robustness. Existing methods fail to identify causal features behind performance degradation caused by corruption. To analyze causal relationships in corrupted data, we propose a method, Robust Causal Feature Disentanglement(RCFD). Our method introduces a learnable causal feature disentanglement mechanism specifically designed for reinforcement learning scenarios, integrating the CausalVAE framework to disentangle causal features governing environmental dynamics from corruption-sensitive non-causal features. Theoretically, this disentanglement confers a robustness advantage under data corruption conditions. Concurrently, causality-preserving perturbation training injects Gaussian noise solely into non-causal features to generate counterfactual samples and is enhanced by dual-path feature alignment and contrastive learning for representation invariance. A dynamic graph diagnostic module further employs graph convolutional attention networks to model spatiotemporal relationships and identify corrupted edges through structural consistency analysis, enabling precise data repair. The results exhibit highly robust performance across D4rl benchmarks under diverse data corruption conditions. This confirms that causal feature invariance helps bridge distributional gaps, promoting reliable deployment in complex real-world settings. Full article

For large-scale measurement of microbubble parameters on the ocean surface beneath breaking waves, a buoy-type bubble sensor (BBS) is proposed. This sensor integrates a panoramic bubble imaging sub-sensor with a circular array fiber-optic sub-sensor. The sensitive unit of the latter sub-sensor is designed via theoretical modeling and experimental validation. Theoretical calculations indicate that the optimal cone angle for a quartz fiber-optic-based sensitive unit ranges from 45.2° to 92°. A prototype array element with a cone angle of 90° was fabricated and used as the core component for feasibility experiments in static and dynamic two-phase (gas and liquid) identification. During static identification, the reflected optical power differs by an order of magnitude between the two phases. For dynamic sensing of multiple microbubble positions, the reflected optical power varies from 13.4 nW to 29.3 nW, which is within the operating range of the array element’s photodetector. In theory, assembling conical quartz fiber-based sensitive units into fiber-optic probes and configuring them as arrays could overcome the resolution limitations of the panoramic bubble imaging sub-sensor. Further discussion of this approach will be presented in a subsequent paper. Full article