Abstract:Structural health monitoring (SHM) tasks like damage detection are crucial for decision-making regarding maintenance and deterioration. For example, crack detection in SHM is crucial for bridge maintenance as crack progression can lead to structural instability. However, most AI/ML models in the literature have low latency and late inference time issues while performing in real-time environments. This study aims to explore the integration of edge-AI in the SHM domain for real-time bridge inspections. Based on edge-AI literature, its capabilities will be valuable integration for a real-time decision support system in SHM tasks such that real-time inferences can be performed on physical sites. This study will utilize commercial edge-AI platforms, such as Google Coral Dev Board or Kneron KL520, to develop and analyze the effectiveness of edge-AI devices. Thus, this study proposes an edge AI framework for the structural health monitoring domain. An edge-AI-compatible deep learning model is developed to validate the framework to perform real-time crack classification. The effectiveness of this model will be evaluated based on its accuracy, the confusion matrix generated, and the inference time observed in a real-time setting.

What problem does this paper attempt to address?

### Problems the Thesis Tries to Solve This paper aims to explore the integration of Edge - Artificial Intelligence (Edge - AI) technology in the field of Structural Health Monitoring (SHM) to achieve real - time bridge detection. Specifically, the thesis attempts to solve the following key problems: 1. **Real - time Problem**: - Most existing AI/ML models have problems of low latency and long inference time in real - time environments. In this paper, by introducing Edge - AI technology, the response time and computing speed of the model are optimized, thus achieving real - time inference. 2. **Security and Privacy Problems**: - Traditional cloud servers or network connection methods have security and privacy risks during data transmission. Edge - AI technology allows inference and calculation on the device side, thereby improving data security and privacy protection. 3. **Bandwidth and Storage Requirements**: - Real - time data processing requires a large amount of bandwidth and storage resources. Edge - AI technology reduces the requirements for bandwidth and storage by performing data processing on the device side. 4. **Stability Problem**: - Systems relying on cloud servers cannot work properly in case of network disconnection or network instability. Edge - AI technology improves the stability and reliability of the system. 5. **Cost Optimization**: - By performing inference on the device side, Edge - AI technology can reduce the use cost of cloud services and achieve more cost - effective AI integration. ### Specific Research Objectives - **Develop a Light - weight Neural Network Model**: This paper proposes a light - weight Convolutional Neural Network (CNN) model for real - time crack classification tasks. - **Optimize the Neural Network Response Time**: Through methods such as model pruning, quantization, and weight clustering, the inference time of the model is optimized, reduced from an average of 3 seconds to 20 milliseconds. - **Verify the Validity of the Framework**: Using the Kneron KL520 platform, the effectiveness and performance of the proposed Edge - AI framework in practical applications are verified. ### Main Contributions 1. **Propose a New Edge - AI Framework**: An Edge - AI framework suitable for the field of Structural Health Monitoring. 2. **Develop and Deploy an Optimized CNN Model**: Perform crack classification on image data in a real - time environment. 3. **Significantly Optimize the Inference Time**: Optimize the average inference time for crack identification from 3 seconds to 20 milliseconds. ### Conclusion This paper proposes a new Edge - AI framework for real - time inference tasks in the field of Structural Health Monitoring. By developing and optimizing a light - weight CNN model and testing it on the Kneron KL520 platform, the effectiveness and performance of this framework are verified. Experimental results show that this model has a high accuracy rate (92.4%) in crack classification tasks and significantly shortens the inference time.

Integrating Edge-AI in Structural Health Monitoring domain

An Investigation Into the Advancements of Edge-AI Capabilities for Structural Health Monitoring

Edge Integration of Artificial Intelligence into Wireless Smart Sensor Platforms for Railroad Bridge Impact Detection

Damage Detection in Bridge Structures: An Edge Computing Approach

From Cognitive Bias Toward Advanced Computational Intelligence for Smart Infrastructure Monitoring

Deep Learning for Structural Health Monitoring: Data, Algorithms, Applications, Challenges, and Trends

Advancing Bridge Structural Health Monitoring: Insights into Knowledge-Driven and Data-Driven Approaches

Integration of Railway Bridge Structural Health Monitoring into the Internet of Things with a Digital Twin: A Case Study

Real-Time Video Surveillance Based Structural Health Monitoring of Civil Structures Using Artificial Neural Network

Crack Detection of Structures using Artificial Intelligence System

Unsupervised deep learning framework for temperature-compensated damage assessment using ultrasonic guided waves on edge device

Deep learning-based structural health monitoring

Artificial Intelligence and Structural Health Monitoring of Bridges: A Review of the State-of-the-Art

Edge Structural Health Monitoring (E-SHM) Using Low-Power Wireless Sensing

Accelerating Classification on Resource-Constrained Edge Nodes Towards Automated Structural Health Monitoring

Weakly supervised crack segmentation using crack attention networks on concrete structures

Lightweight Real-time Detection of Components Via a Micro Aerial Vehicle with Domain Randomization Towards Structural Health Monitoring

Review of Bridge Structural Health Monitoring Aided by Big Data and Artificial Intelligence: From Condition Assessment to Damage Detection