# Predicting and optimizing control parameters of stir casting of Al alloy/MWCNT/RHA composite using artificial neural network and Taguchi-Grey relational analysis for multi-objective outcomes

## TL;DR

This study uses artificial neural networks and statistical methods to optimize the production of a composite material for better mechanical properties.

## Contribution

A novel combination of ANN and Taguchi-Grey relational analysis is used to optimize multi-objective composite casting parameters.

## Key findings

- The ANN model achieved a high R2-score of 99.65% in predicting tensile strength.
- MWCNT content had the most significant impact on composite properties with 48.26% contribution.
- Optimized parameters combination A2B3C3D2E2 produced the best composite properties.

## Abstract

In the present investigation, the influence of various casting parameters viz. stirrer time, stirrer speed, and processing temperature and reinforcement content on the mechanical properties of AlP0507/CNT/RHA composite is assessed. The optimum parameter combination that produces greater multi-objective performance was obtained using the GRA method. The comparison of all R2-score showed that the ANN model is best fitted to predict the tensile strength of HAMMC with highest R2- score of 99.65%. GRA established that the MWCNT content has most significant influence on the response parameters followed by stirring time, RHA content, stirring speed and processing temperature; and the best properties of stir cast HAMMCs was obtained by the combination A2B3C3D2E2. ANNOVA performed on GRA indicated that MWCNT content with contribution of 48.26% exerted maximum impact on the properties of the fabricated HAMMC, followed by stirring time with contribution 19.4%. Processing temperature contributed least with meagre contribution of 2.17%. The predicted value of GRG (0.830775) was found very close to the GRG value of the highest-ranked experiment (0.79643454) confirming the accuracy of the optimization and its validation. The improvement in GRG value by 0.09792454 shows that the optimized parameters provided the optimal results and can be recommended.