Prediction of Ultrasonic Pulse Velocity of Concrete

Ultrasonic pulse velocity test (UPV) is one of the main and popular non-destructive tests of concrete which examine the cracks, quality, homogeneity, and defects within the concrete. In this investigation, the values of UPV were predicted using the three soft computing models, such as M5 rule (M5PR), M5P tree (M5PT), and random forest. The study area of the study is the old hostel of Panipat Institute of Engineering and Technology, Samalkha, Panipat, which was found to have some signs of distress and surface cracks. The compressive strength value of concrete, half-cell potential, chloride, carbonation, and Transit time are the input parameter whereas UPV is the output parameter. The total data set contained 149 observations out of which 50 (one-third) observations were used for testing and 100 were used in training. The modeling results suggest that all the three soft computing models could predict the UPV very well, although Ran-For has an edge over M5PT and M5PR in the prediction of UPV in terms of root mean square error and correlation coefficient values. Hence, the Ran-For model can be used in the prediction of UPV in case of the non-availability of data.