Applying Artificial Intelligence to Improve on-Site non-Destructive Concrete Compressive Strength Tests

• Main Authors: NGO TU QUYNH LOAN, 吳秀瓊鸞
• Other Authors: YU, REN- WANG
• Format: Others
• Language: en_US
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/handle/938qa7

博士 === 國立高雄科技大學 === 土木工程系 === 107 === In the construction field, Non –Destructive Testing (NDT) methods are used to inspect the compressive strength of concrete because they are easy to carry and economical. The two popular NDT methods are Rebound Hammer (RH) test and Ultrasonic Pulse Velocity (UPV) test. However, the estimated results from RH test give a large percentage of error when comparing with the results from using destructive testing methods; and for UPV testing method, the estimated results are not quite accurate. In this study, different artificial neural network models are applied to the on-site Silver Schmidt RH and UPV tests to fully investigate the effect of various artificial intelligence models and their parameters on the prediction results. Data from on-site beam of concrete are collected to develop and validate the models. The research results deliver valuable insides when using RH and UPV tests in addition with Artificial Neural Network (ANN), Support Vector Machine (SVM) and Adaptive Neural Fuzzy Inference System (ANFIS). In this research, data of NDT methods of 98 concrete samples are collected in an on-site residential building by a local material laboratory. The Artificial Intelligence (AI) prediction models are developed then validated using these data from the laboratory. As a result, AI prediction models give more precise results than statistical regression models do. The research results show significant enhancement when using RH and UPV tests with the application of ANNs, SVM and ANFIS to estimate concrete compressive strength.