Embedded Artificial Intelligence for Early-Age Concrete Strength Monitoring: A Scalable IoT System

Abstract

Real-time strength monitoring for pavement quality concrete (PQC), especially at an early age, is critical to determine the optimal time to open the concrete pavement to traffic. In the past, maturity method-based Internet of Things (IoT) enabled systems to have been developed commercially and by individual researchers. They rely on statistical models specific to a concrete grade and do not fully capture the material behaviour. This study aims to replace traditional regression-based IoT systems by embedding an artificial intelligence (AI)- machine learning (ML) model. The proposed model, coupled with an IoT system, predicts the compressive and flexural strength of PQC of M30, M40 and M50 grade mixes with improved accuracy. The AI-ML model is optimised to be directly deployed in a low-cost ESP8266 microcontroller, creating an intelligent system with a standalone sensor. The proposed system demonstrates superior fidelity compared to previous approaches in the field and generalises the strength prediction of multiple PQC grades in a single model. This can establish a framework for promoting automation in quality control and lead towards a data-driven approach in pavement construction.