Abstract
The increasing demand for efficient, reliable, and sustainable monitoring systems has driven research in the areas of measurement systems, sensors and energy harvesting. New sensors enable real-time monitoring of physical and environmental conditions with efficient measurement systems. The integration of these technologies has the potential to enhance system performance, reduce operating costs, and improve maintenance processes. This special issue is devoted to state-of-the-art modelling, data processing and condition monitoring methods for applications ranging from Industry 4.0 to environment monitoring, enabled by the Internet of Things (IoT).
Introduction
As the global landscape shifts toward Industry 4.0 and faces escalating environmental challenges, the necessity for robust monitoring systems has never been more critical. Real-world monitoring environments are inherently complex; sensors often grapple with cross-sensitivity—such as temperature sensors responding to mechanical strain—while data transmission is frequently hampered by noise and bandwidth limitations. Furthermore, deploying sophisticated artificial intelligence in remote areas requires high energy efficiency.
This special issue of the Journal of Dynamics, Monitoring and Diagnostics highlights six research papers that push the boundaries of sensor technology, IoT, and industrial health monitoring. By integrating deep learning, fuzzy logic, and novel mechanical designs, these studies significantly improve the accuracy and efficiency of environmental and industrial diagnostics.
Methodologies in Advanced Sensing
The research presented covers a broad spectrum of methodologies designed to bridge the gap between theoretical modeling and practical application.
Structural and Environmental Hardware
To address the problem of strain interference in temperature sensing, researchers developed a Fiber Bragg Grating (FBG) sensor encapsulated in an arc-shaped cavity within a composite structure. This design achieves near-perfect mechanical decoupling, allowing for pure temperature readings even in high-stress environments.
In the field of structural health monitoring, computer vision has emerged as a powerful non-contact tool. A Gaussian-based vibration detection method was utilized to track wind turbine tower edge points, providing a viable alternative to traditional physical accelerometers for detecting early foundation settlement.
Modeling and Signal Processing
Advanced modeling techniques, such as Finite Element Modeling (FEM) combined with Model-Assisted Probability of Detection (MAPOD) analysis, were employed to enhance eddy current testing. By integrating multiple response signals—including coil impedance and magnetic flux—the system's ability to detect minute flaws was significantly improved.
For manufacturing processes, specifically Fused Filament Fabrication (3D printing), Fast Kurtogram analysis was applied to acoustic emission signals. This allows for real-time monitoring of material flow states, identifying stability domains to ensure product quality.
Intelligent Diagnostics and Edge AI
A significant focus of this issue is the deployment of "intelligence at the edge."
TinyML for Environmental Protection
To combat illegal logging, researchers developed an ensemble network of Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNN). Through quantization, this model was optimized for low-power microcontrollers (TinyML), achieving 88% accuracy in classifying forest audio. This demonstrates the viability of long-term, battery-powered environmental monitoring.
Fuzzy Logic in Industrial Maintenance
In the context of aero-engine maintenance, an "Attribute-Driven Fuzzy Fault Tree" (FFT) was developed to handle the complex and often conflicting indicators found in lubricant failure diagnosis. This approach increased diagnostic accuracy from 73.4% to 92.6% compared to existing methods.
Key Results and Insights
The findings across these studies highlight several critical insights:
- Mechanical Decoupling: The FBG composite structure achieved a sensitivity of 26.3 pm/°C with 99.65% isolation from mechanical strain.
- Multi-Factor Analysis: In eddy current testing, integrating multiple signals outperforms conventional single-factor analysis for small flaw detection.
- Vision-Based Monitoring: Computer vision is capable of high-precision structural monitoring without the need for mounted sensors.
- Edge Efficiency: Sophisticated ensemble models can be shrunk for microcontrollers without sacrificing significant accuracy, enabling scalable IoT solutions.
Conclusion and Future Directions
The guest editorial acknowledges the contributions of authors and reviewers in advancing the state-of-the-art in measurement systems. While the current research offers significant advancements, future work must focus on scaling these "smart" devices into larger, interconnected IoT networks and improving the explainability of AI models in safety-critical industrial applications. Testing the long-term durability of novel composite sensors in extreme outdoor environments remains a priority for future research.
The guest editorial would like to thank all contributing authors, reviewers, and the journal’s production and editorial teams for their assistance with this special issue.
Summary updated