TY - JOUR ID - 168920 TI - Prediction of Compressive Strength of Corncob Ash Concrete for Environmental Sustainability Using an Artificial Neural Network: A Soft Computing Techniques JO - Journal of Soft Computing in Civil Engineering JA - SCCE LA - en SN - AU - Abhishek, R AU - Gowda, B S Keerthi AU - Naveen, D C AU - Naresh, K AU - Sundarakannan, R AU - Arumugaprabu, V AU - Varsha, Amogha AD - Research Scholar, Department of Civil Engineering, Visvesvaraya Technological University, Mysuru, India AD - Department of Civil Engineering, Visvesvaraya Technological University, Mysuru, India AD - Department of Civil Engineering, Sri Venkateswara College of Engineering, Tirupati, India AD - Department of Aerospace Engineering, Khalifa University of Science and Technology, Abu Dhabi, U.A.E. AD - Institute of Agricultural Engineering, Saveetha School of Engineering, SIMATS, Chennai, India AD - Department of Mechanical Engineering, Kalasalingam Academy of Research and Education, Tamilnadu, India AD - Research Scholar, Department of Civil Engineering, PESCE (VTU), Mandya, Karnataka, India Y1 - 2023 PY - 2023 VL - 7 IS - 2 SP - 115 EP - 137 KW - Soft-computing techniques KW - Levenberg-Marquardt KW - Agriculture waste KW - Construction materials KW - Pozzolanic material DO - 10.22115/scce.2023.347663.1471 N2 - Agricultural waste materials are increasingly being used as partial replacements for cement in concrete. Several experimental studies are available to evaluate the mechanical properties of plastic waste reinforced concrete but there are limited evaluations on agricultural waste material. In this study, an attempt is made to investigate the compressive strength of Corn Cob Ash (CCA) concrete at different replacement levels by implementing an Artificial Neural Network (ANN). As the percentage of CCA increases, workability, density and compressive strength decreases, hence the developed ANN model consists of 3 input parameters (cement content, CCA content, and curing ages) in the input layer, 4 hidden neurons in the hidden layer and 3 output parameters (slump, density, and compressive strength) in the output layer. Training is done by adopting Levenberg-Marquardt back-propagation algorithm by considering 80% of experimental data with log-sigmoid activation function for both hidden and output layers. The developed model has a high correlation coefficient of 0.999 for both the training and testing data sets. It has low MSE and MAPE values of 2.2768x10-5 and 1.25 for training data respectively and 3.0463x10-5 and 1.37 for testing data respectively. Hence, it is concluded that the developed model predicts the output at an average rate of 98% accuracy. The predicted 2.5% replaced CCA concrete shows the best performance at all curing ages. Therefore, this percentage level is considered as an optimum replacement level which does not much affect the hardened properties of concrete. UR - https://www.jsoftcivil.com/article_168920.html L1 - https://www.jsoftcivil.com/article_168920_2b51a9342cfebf2d6732620b68d76246.pdf ER -