Modelling efficiency of industrial waste utilised for microsurfacing using artificial neural networks

Abstract

There was a need develop a model to determine the optimum proportion of waste materials which ensure the quality of designed micro surfacing mix. Artificial neural network (ANN) has been used to create a model for prediction of the optimum proportion of mineral filler and additive due to non-linearity of data. In the present study, since there are five inputs (dimensions) and two outputs having nonlinear relationship, ANN modelling suits to be best for output prediction. The Bayesian regularisation algorithm was used to train the network. The micro surfacing characteristics are a function of five input performance parameters namely mixing time, cohesion (30 min), cohesion (60 min), setting time and wet track abrasion test. The two output parameters are filler proportion and control additive balance. The model tool developed shall ease in determining the mix design parameters, i.e. filler content and additive percentage to achieve the desired effect.