Structural Health Monitoring System Using AI

Abstract

Structural Health Monitoring (SHM) is essential for ensuring the long-term safety, durability, and reliability of civil infrastructure by enabling early detection of cracks, abnormal vibrations, tilting, and deformation. Conventional inspection techniques remain limited due to their manual nature, lack of real-time capabilities, restricted accessibility, and susceptibility to human error. To address these limitations, this project proposes an intelligent SHM framework that integrates Internet of Things (IoT) sensing, cloud computing, and deep learning–based crack detection. The system utilizes an ESP32 microcontroller interfaced with an ADXL345 accelerometer, a vibration sensor, and a flex sensor to continuously monitor structural responses such as tilt, tremors, and surface bending. Sensor data is displayed locally through an LCD module and uploaded to the ThingSpeak cloud for remote access and long- term trend analysis. A Telegram Bot is incorporated to deliver instant alerts whenever abnormal sensor patterns are detected, ensuring rapid awareness and timely intervention. Furthermore, an AI-driven crack detection module based on YOLOv5 analyzes high-resolution surface images and accurately identifies crack presence using advanced object-detection techniques. By combining IoT-based sensing with real-time cloud analytics and deep learning image interpretation, the proposed system offers a cost-effective, scalable, and highly reliable solution for preventive maintenance and enhanced structural safety in modern infrastructure.

Introduction

Structural health is critical for the safety, stability, and longevity of civil infrastructure, including buildings, bridges, and industrial facilities. Traditional manual inspections are limited by cost, labor, subjectivity, and their inability to detect internal or micro-level defects continuously. To address these challenges, modern SHM systems leverage IoT sensors, cloud computing, and Artificial Intelligence (AI) to enable continuous, real-time monitoring and predictive maintenance.

The proposed system integrates an ESP32 microcontroller with an ADXL345 accelerometer, vibration sensor, and flex sensor to measure tilt, vibrations, and crack-induced deformations. Sensor data is displayed locally on a 16×2 LCD and transmitted to the ThingSpeak cloud for remote visualization and analysis. A Telegram Bot provides instant alerts for abnormal readings, ensuring rapid response to potential structural issues.

For automated crack detection, the system employs YOLOv5, a deep learning-based object detection framework, which processes high-resolution images to identify cracks accurately and efficiently. By combining sensor-based monitoring with AI-powered image analysis, the system achieves a multi-layered, hybrid SHM architecture that enhances precision, reliability, and responsiveness.

This IoT–AI SHM system is cost-effective, scalable, and intelligent, suitable for smart cities, industrial infrastructure, academic research, and disaster management. It enables early fault detection, predictive maintenance, improved safety, and extended operational lifespan of civil structures.

Key Features:

• Continuous real-time monitoring of structural parameters (tilt, vibration, deformation).

• AI-driven crack detection using YOLOv5.

• Cloud-based data visualization and long-term analytics.

• Instant alerts via Telegram for abnormal conditions.

• Scalable and adaptable for diverse infrastructure types.

Conclusion

The development of an AI- and IoT-enabled Structural Health Monitoring (SHM) system presents an efficient, reliable, and real-time approach for assessing the stability and safety of buildings and other civil infrastructures. By integrating multiple sensors—including the ADXL345 accelerometer, vibration sensor, and flex sensor—with the ESP32 microcontroller, the system effectively captures crucial parameters associated with tilt, oscillation, and crack-induced deformation. Continuous data transmission to the ThingSpeak cloud facilitates remote monitoring and long-term behavioural analysis, while the incorporation of a Telegram Bot ensures immediate alert notifications in abnormal conditions, thus enabling rapid response and reducing the risk of structural hazards. The inclusion of the YOLOv5 deep learning model significantly enhances the system’s capabilities by delivering accurate, automated crack detection from captured surface images. The results confirm that this hybrid sensing and AI- based approach is substantially more effective than conventional manual inspection methods. It provides early- warning signals, reduces dependency on periodic human assessments, and offers a cost-effective, scalable, and easily deployable solution that strengthens preventive maintenance practices and improves overall infrastructure safety. Although the proposed system establishes a strong foundation for real-time structural monitoring, several enhancements can further elevate its performance and practical applicability. Future developments may include integrating additional sensor modalities such as strain gauges, ultrasonic sensors, humidity sensors, and temperature sensors to capture a broader range of structural and environmental conditions. The crack detection module can be upgraded by training YOLOv5 on larger, more diverse datasets or by adopting next-generation architectures such as YOLOv8, EfficientDet, or transformer- based vision models to improve robustness under challenging lighting and texture variations. Predictive analytics using advanced machine learning models—such as LSTM networks, anomaly detection algorithms, and survival analysis—can be employed to forecast structural deterioration trends, enabling proactive and data-driven maintenance strategies. Large-scale deployments across bridges, highways, multi-storey buildings, and dams can be supported through wireless sensor networks and mesh communication architectures, ensuring multi-node connectivity and centralized oversight. Beyond the technical advancements, the proposed SHM system demonstrates the broader transformational potential of integrating IoT, AI, and cloud computing into modern infrastructure management.

References

[1] R. Avilés, M. L. Romero, and J. Duarte-Salazar, “Automatic detection of concrete cracks using deep learning-based computer vision techniques,” Construction and Building Materials, vol. 295, pp. 1–12, 2021. [2] Russell, S., & Norvig, P. (2020). Artificial Intelligence: A Modern Approach (4th ed.). Pearson. This is one of the most widely used AI textbooks, covering fundamentals, machine learning, and deep learning. [3] Bochkovskiy, C.-Y. Wang, and H.-Y. M. Liao, “YOLOv4: Optimal speed and accuracy of object detection,” arXiv preprint arXiv:2004.10934, 2020. [4] G. Jocher et al., “YOLOv5 — You Only Look Once, Version 5,” GitHub Repository, Apr. 2022. [Online].Available: https://github.com/ultralytics/yolov5 [5] H. Yang, Z. Li, and S. Yu, “Concrete crack detection using U-Net and image enhancement methods,” Journal of Building Engineering, vol. 42, pp. 1–10, 2021. [6] Tom M. Mitchell (1997) – Machine Learning, this book is one of the foundational textbooks on machine learning, covering key concepts, algorithms, and theoretical principles. It is widely used. [7] McEwen, A. (2013). Designing the Internet of Things. This book provides an introduction to IoT design, architecture, and real-world applications. It covers hardware, software, networking, and security aspects, making it essential for IoT-based Structural Health Monitoring (SHM) projects. [8] H. Zhao, Y. Guo, and L. Sun, “DeepCrack: A deep hierarchical feature learning architecture for crack segmentation,” Neurocomputing, vol. 359,pp. 221–233, 2019.

Copyright

Copyright © 2025 Aravinda T V, Krishnareddy K R, Ramesh B E, B Zaid Ismail, Chethan L, Ganesh Patel M, Lokateja K M. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.