Trestward: An AI Powered Intelligent Infrastructure Monitoring System for Predictive Structural Health Analysis of Critical Civil Infrastructure

Critical civil infrastructure has been degrading (bridges, dams, tunnels, and flyovers) and presents an increased risk to public safety especially in areas with a rapidly aging stock of infrastructure. Conventional periodic inspection methods do not have the ability to detect minor structural anomalies and, if left unchecked, the anomaly will lead to catastrophic failures. In this paper, we describe a cost effective prototype system called "Trestward" that supports continuous and realtime structural health monitoring (SHM) of critical infrastructure assets via an Internet of Things (IoT). The prototype system combines multiple hardware components including an ESP32 microcontroller, an Arduino Uno, an MPU6050 sixaxis IMU, an SW 420 vibration sensor, and a DHT11 environmental sensor. Sensor data from these hardware components are sent over IEEE 802.11 b/g/n Wi-Fi using the MQTT protocol to a dashboard hosted on the cloud where anomaly detection based on thresholds classifies the structural conditions into three actionable categories: safe, warning, and critical. Two hand-built scaled physical models of a highway bridge span and a gravity dam were created and validated to demonstrate the functionality of the prototype system. Initial results show that the Trestward prototype was able to provide real-time communications and reliable threshold-based structural classifications under lab conditions. Future work will involve developing a supervision Random Forest machine learning classifier for the next stage of artificial intelligence (AI) that will be trained with a labelled dataset collected using the operational prototype.