Multilayer perceptron modelling of geopolymer composite incorporating fly ash and GGBS for prediction of compressive strength

Abstract

ABSTRACT In today’s world of advanced technology tacked with Artificial Intelligence and neural networks, dealing with geopolymer composites that involve multiple variables, there is tremendous scope for methods that can accurately predict the strength parameters, as no standard mixed design IS code has been developed. Since the quantities of silica modulus, Na2SiO3 material, water to binder ratios, curing time, curing temperature all have a significant impact on compressive strength. A multi-layer ANN can therefore be built for geopolymer compressive strength prediction by the use of the backpropagation architecture. Such computing tools will be better, cheaper, time-saving, and labour-saving in terms of resource allocation. In the current study, compressive strength prediction of geopolymer composite containing fly ash and ground granulated blast furnace slag was performed using a type of artificial neural network (ANN), specifically multilayer perceptron (MLP). In this paper, a total of three hundred seventy-six fact points were accumulated from the literature. The coefficient of determination (R2) was used to assess the proposed model’s accuracy in this analysis. The MLP model proposed will forecast geopolymer compressive strength with an R2 value of 0.968. The algorithm trained and tested the model.