Piezo Based Structural Health Monitoring of Concrete Systems Using Machine Learning

Abstract

Monitoring the health of infrastructure has become imperative now a days due to vast infrastructure development in the last few decades which has now aged. Proper monitoring of the structures during the construction as well as design life stage can avoid catastrophic failures. Strength and durability are the two main aspects of reinforced concrete structures and providing a real time monitoring of these are the big challenges which almost all the concrete researchers are working around the world. The field of structural health monitoring (SHM) using piezo sensors via electro-mechanical impedance (EMI) has grown tremendously in recent years can provide a solution to these issues. The application of machine learning (ML) techniques is experiencing exponential growth in SHM domain using sensors because of the immense capability of handling voluminous datasets and making past and future predictions based on input data used for training. The aim of the present research is to utilize smart sensors, namely PZT sensors, in different configurations and analyse its sensitivity for strength monitoring, durability studies and suitably propose its application in real-life. ML models were developed using the sensor data to predict the strength of different cementitious systems and concrete systems. The research was further extended to develop ML models to predict the baseline/healthy and future EMI data of different blended RC structures (conventional, fly ash blended and fly ash based geopolymer) subjected to a chloride-laden environment. Also, different corrosion phases have been identified based on the famous Tuutti’s model for the RC and prestresses structures subjected to corrosion. Further pioneering work has been carried out on different concrete systems subjected to combined environmental and mechanical loading wherein physical models for structural parameter deterioration were developed and also empirical relations between equivalent stiffness and surface concentration were established. The developed strength prediction models can be used to predict the strength of the newly developed concrete non-destructively which will aid for proper project management. The ARIMA model can be used for prediction of baseline/healthy data for existing structures to study durability. It is expected that this research work will serve as new important guidelines to the industry as well as to the research community working in the field of structural health monitoring.