Development of a bi- directional multi- input- multi-output predictive model for the fused deposition modelling process using co-active adaptive neuro-fuzzy inference system

Abstract

Abstract In the automated manufacturing industries, modelling and prediction of the process parameters of additive manufacturing plays an important role. This paper proposes a computationally intelligent method using coactive-adaptive neuro-fuzzy inference system to establish relationships between the process parameters and the responses, in both forward and backward directions, for the fused deposition modelling process. Experimental data have been statistically analyzed and regression equations have been generated to produce large training samples. The model has been built using six inputs each with non-linear Gaussian membership function distributions, and three responses, each with linear membership function distributions for the forward-directed mapping. Similarly, three inputs and six outputs from the same training data set have been used to formulate the backward-directed inference model. The parametric study for the used back propagation algorithm has been conducted and validation has been accomplished with the optimal settings using actual experimental data.