Comparative Analysis of Grey Relational Analysis Integrated with the Principal Component Analysis and Analytic Hierarchy Process for Multiobjective Optimization of Inclined Laser Percussion Drilling in Carbon Fiber Reinforced Composites
Yadvendra Kumar Mishra,Sanjay Mishra,S. C. Jayswal
DOI: https://doi.org/10.1142/S0219686721500475
2022-02-13
Journal of Advanced Manufacturing Systems
Abstract:Multiobjective optimization (MOO) helps to achieve simultaneous improvement of more than one output characteristic in machining processes where complex interaction between the input parameter exists. This study focuses on the comparative analysis of design of experiment (DoE)-based grey relational analysis (GRA) combined with principal component analysis (PCA) and analytic hierarchy process (AHP). Experiments were conducted with millisecond (ms) duration pulsed Nd: YAG laser using the Box–Behnken design (BBD) approach of the response surface methodology (RSM) at three different levels of input parameters. The output parameters, i.e., hole circularity at top (HCT), hole circularity at bottom (HCB), and hole taper (HT), were determined for various input parameters like pulse current (I), pulse width ( Pw ), gas pressure ( Gp ), workpiece thickness ( Ti ), and incidence angle ( θ ) during laser percussion inclined hole drilling (LPIHD) in the carbon fiber reinforced polymer (CFRP) of three different thickness, i.e., 1mm, 3mm, and 5mm at incidence angles of 0, 10, and 20 degrees. Multiobjective function based on RSM has been developed for GRA-PCA and GRA-AHP and further optimizations were performed using the desirability approach of RSM. The analysis revealed that the angle of incidence is the most significant factor for controlling the output parameters. Interaction of pulse current and thickness ( I×Ti ) has a major impact on output responses. The GRA-PCA approach gives the average improvement of 2%, 9%, and 37%, respectively, for HCT, HCB, and HT, whereas in the case of GRA-AHP, the corresponding improvements are only 1%, 6%, and 11%. Therefore, the GRA-PCA approach is a more effective tool for the MOO of LPIHD in CFRP.