A general regression neural network approach for the evaluation of compressive strength of FDM prototypes

Soft computing (SC) methods are well known for their remarkable ability of learning from experimental data set to describe nonlinear and interaction effect with great success. Due to complex mechanism and uncertainty of the fused deposition modelling (FDM) process, of late, SC methods are preferred compare to theoretical model (physics-based) for measuring output responses of the FDM process. In the present study, performance modelling of FDM prototype has been carried out using two potential SC methods such as multi-gene genetic programming (MGGP) and general regression neural network (GRNN). The effect of three input factors namely, layer thickness, orientation, raster angle on output compressive strength of the prototype was studied using the two SC models. Data generated from the experimental study are fed into the cluster of MGGP and GRNN for the formulation of mathematical models. Based on the experimental datasets, the proposed SC models predict compressive strength of FDM fabricated prototype in terms of input process parameters. The predictions of compressive strength by these models are evaluated against the data generated in experimental study. Results conclude that the model generated by GRNN has better goodness of fit compare to the MGGP model and thus can be a promising alternative for optimizing the FDM process.