Application of machine learning models to investigate the performance of concrete reinforced with oil palm empty fruit brunch (OPEFB) fibers

The use of agricultural waste as a sustainable fibre material for structural concrete manufacturing has been gaining traction owing to its eco-friendliness, availability, and cost effectiveness. The oil palm empty fruit brunch (OPEFB) fibre is cheap and readily available in most tropical countries. In this study, the performance of using OPEFB-fibre for structural concrete production was investigated experimentally and using a machine-learning approach. An architectural study of the artificial neural network (ANN) with different training algorithms and activation functions in a range of 1–10 neurons in a single hidden layer was carried out to select the optimal network for prediction. Also, hyper-parametric tuning of the adaptive neuro-fuzzy inference system (ANFIS) model clustered with fuzzy c-means (FCM) was studied for optimal model selection. The best models for each output were selected after evaluating the model’s performance using relevant statistical metrics like root mean square error (RMSE), mean absolute deviation (MAD), mean absolute percentage error (MAPE), and correlation coefficient (R 2 ). FCM-clustered ANFIS models with five, five, six, and four clusters for co-efficient of water absorption, density, compressive strength, and tensile strength, respectively, outperformed other models with RMSE values of 0.0016, 3.183, 0.6357 and 0.0234 and R 2 values of 0.9994, 0.9985, 0.9966 and 0.9919, respectively. The experimental result revealed that the maximum compressive strength was obtained with 0.2% OPEFB-fibre at 60 days curing, maximum tensile strength with 0.2% OPEFB-fibre at 7 days curing and maximum density with 0.4% OPEFB-fibre at 7 days. Graphical abstract