Analysis and prediction of uncertain responses using regression and fuzzy logic for friction stir welding of AA2014 under n-MQL

AA2014 is an Al-Cu alloy friction stir welded under different combinations of rotational speed (800, 1000 and 1200 rpm) and transverse speed (44, 60, 72 mm/min) under minimum quantity lubrication condition with graphene nanofluid as coolant. Design of experiments is performed using Taguchi L9 orthogonal array. Analysis of variance technique is adapted to find the most influencing input parameter (rotational speed, transverse speed) of each output response (ultimate tensile strength, % elongation, microhardness and grain size). Regression and fuzzy logic based models are developed to predict the output responses. The reliability of the predicted results is tested by calculating the correlation coefficient. The predicted results from regression and fuzzy logic are then compared with the experimental results. The results of trend analysis exhibit the substantial influence of both the input parameters on the output responses. The results from ANOVA reveals that the rotational speed highly influences ultimate tensile strength and grain size while transverse speed majorly influences microhardness. The error in prediction using fuzzy model is observed to be significantly limited with correlation coefficients in the range of 0.70-0.96. The developed models are observed to be highly efficient and therefore can be used for prediction in any uncertain engineering applications.

Automated summary

This study uses Taguchi design and analysis of variance techniques to predict the output responses of friction stir welded AA2014 alloy using fuzzy logic models. The results show that rotational speed and transverse speed have a significant influence on ultimate tensile strength, microhardness, and grain size. The developed models are highly efficient.