Machine learning-based monitoring and predicting the compressive strength of different blended cementitious systems using embedded piezo-sensor data

Abstract

This paper presents strength monitoring and prediction of blended cementitious systems using embedded piezo sensor via machine learning (ML) models. Experiments were conducted on three cementitious systems such as ordinary Portland cement (OPC), fly-ash blended cement (FA), and limestone calcined-clay cement (LC3) in which piezo sensor is embedded inside the cement pastes to acquire electro-mechanical impedance (EMI) signatures in the form of conductance and susceptance during strength development. Further, an equivalent stiffness parameter (ESP) was extracted from EMI signatures by developing physical model based on spring and damper element. Furthermore, ML models were developed to predict the compressive strength based on ESP and destructive compressive strength data. Experimental result indicates ESP (non-destructively) and compressive strength (destructively) value of LC3 exhibits higher than other two cementitious systems (OPC and FA). The developed ML models show promising results in prediction of compressive strength for all the cementitious systems with R2 = 0.99.