Neural network model to predict compressive strength of steel fiber reinforced concrete elements incorporating supplementary cementitious materials. (2022)

• Record Type: Journal Article
• Authors: Gehlot, Tarun; Dave, Mayank; Solanki, Deepanshu
• Is Part Of: Materials today. Volume 62:Part 12(2022)
• Page Start: 6498
• Page End: 6506
• Publication Date: 2022
• Subjects: Neural Network Model -- Supplementary Cementitious Materials -- Concrete -- Compressive Strength -- Water Binder Ratio -- Workability; Materials science -- Congresses -- Periodicals; 620.1
• DOI: 10.1016/j.matpr.2022.04.327
• Languages: English
• ISSNs: 2214-7853

Abstract

Abstract: During the recent past, the problem of early deterioration of concrete structures and durability of concrete structures has remained major issue posed to engineers. we have reported here the incorporation of Metakaolin, fly ash, GGBS & silica fume at binary and ternary blended system with diverse Water binder ratio. Various concrete mixes with supplementary Cementitious material (SCM) and Normal Concrete of Grade M40 (NC40) has been prepared and Steel Fiber & plasticizer dosage has been varied. This Research Paper Explore the Neural Network Model comprising Compressive strength of various binary and ternary blended Supplementary Cementitious concrete elements. ANN model, then contrasted with MS-Excel, is created with R programming. The right algorithm and neuronal numbers have been determined for optimizing the model architecture via a responsive analysis, . It was found that the prediction of compressive strength with a neuron network was highly accurate. The relative mean squared error, coefficient of decision (R 2 ) and mean absolute relative error is calculated for the experimental outcomes and model outputs. Levenberg Marquardt as algorithm employed for performance analysis. Paramount and Least Value of R 2 is 97.2 % and 95.2 % for training data sets in the ANN model. Statistical performance demonstrate that proposed ANN models for compressive strength is accurate and extremely nearer to the experimental values. There is a close harmonization between actual data and ANN compressive strength outputs. The ANN model therefore seems to be a valuable tool to forecast compressive strength.