SMART CONSTRUCTION MONITORING USING IOT SENSORS AND MATHEMATICAL MODELING FOR STRUCTURAL HEALTH PREDICTION

Abstract

 The safety, durability and cost efficacy of construction projects are aspects that demand the structural integrity of the same. The conventional ways of inspection are occasionally aperiodic, laborious, and subject to human errors, and hence they cannot easily observe the early signs of structural degradation. The following research puts forward a smart construction monitoring system which considers a combination of IoT sensor networks with predictive mathematical simulation to make structural health assessment and forecasting real time. In the research, there were crust gauges, a tempering sensor, tilt, environmental sensors, and strain sensors on a comprehensive apparatus, which measures strain, vibration, tilt, temperature, and humidity in a period of six months. There were four predictive models used Multiple Linear Regression (MLR), Support Vector Regression (SVR), Random Forest Regression (RFR), and Long Short-Term Memory (LSTM) networks. The experimental outcomes revealed that LSTM gave the best predictive accuracy with R 2 = 0.94, RMSE = 0.58 MPa and MAE = 0.45 Mpa, when compared to RFR, which is robust predicting tilt (R 2 = 0.90). MLR and SVR represented good baselines, as the R 2 values were 0.81 and 0.87 each. The new strategy would help preventive maintenance and increase safety as the possible structural failures could be identified early. The science document is a scalable, experimentally validated approach to predicting the structural real-time health and is part of developing an intelligent, sustainable and resistant construction monitoring system.