Hybrid Taguchi-entropy weight-based GRA optimization and neural network approach to predict the tribological responses of 3D printed PLA, ABS, and their carbon fibre composites

Polylactic Acid (PLA) and Acrylonitrile butadiene styrene (ABS) are thermoplastic polymers with wide variety of industrial applications like in packaging industry, implants, equipment housings etc. Attention of researchers is largely focused on minimising wear inflicted losses in industry. The aim of this work is to optimize the tribological responses i.e. coefficient of friction (CoF) and specific wear rate (SPWR) of polymers PLA and ABS as well as their carbon fibre reinforced composites. In the present work, variation of speed and load is investigated for different materials. Samples were printed using FDM process in accordance to ASTM F732 A2 with printing conditions were identified as 70% infill density, 0° (vertical) orientation, grid infill pattern, and 0.3 mm layer thickness. It was observed that on increasing sliding speed, there is a decrease in CoF and SPWR, while on increasing load, CoF decreases and SPWR increases. Taguchi’s L16 orthogonal array is applied for design of experiments. i.e. CoF by 8.9% and SPWR by 20%. A novel hybrid optimizationprediction framework integrating Shannon entropy-weighted GRA with ANN modelling was developed for tribological analysis of FDM-printed polymer composites-. Techniques such as analysis of variance (ANOVA) and artificial neural network (ANN) are employed for accurate validation and prediction. As opposed to ANOVA. It was observed that predictions made using the ANN approach were very similar to those obtained from actual experiments.