Response Surface Modelling and Effective Application of Adaptive Neuro-Fuzzy Inference System to Analyze Surface Roughness of Al/Gr/Cp5 MMC Machined using WEDM

Mangesh Phate,Shraddha Toney,Vikas Phate,Mangesh Phate,Shraddha Toney,Vikas Phate
DOI: https://doi.org/10.1080/14484846.2021.1913852
2021-04-26
Australian Journal of Mechanical Engineering
Abstract:<span>This paper presents an investigation of wire electrical discharge machining (WEDM) of new fabricated aluminium base composite with 5% graphite by weight, i.e., Al/Gr/Cp5. The new fabricated metal matrix composite (MMC) has a complete application in space, defence, automobile, and other concerned metal industries. In this work, adaptive neuro-fuzzy inference system (ANFIS) and the response surface method (RSM) were selected for the analysis and modelling purposes. ANFIS is a soft computing technique used in nonlinear data training. Taguchi's L<sub>27</sub> plan was employed to conduct the experiments. The process parameters such as pulse-on time (PON), pulse-off time (POFF), a feed rate of wire (FRW) and the input current (IC) were selected for the experimentation. Based on the experimental findings, PON is identified as the most dominant parameter and the input current. WEDM performance was measured in terms of surface roughness (SR). For the optimised (minimum) surface roughness of 1.8625 microns, the best set of process parameters was PON (108 μs), POFF (52 μs), FRW (4 m/min) and the input current (12 Amp). ANFIS-predicted results were compared with the experimental results. An acceptable agreement is observed with the high value of the correlation coefficient (R<sup>2</sup> = 0.9624). The soft computing technique, like ANFIS-RSM, shows a good competency and reliability for the presented work.</span>
What problem does this paper attempt to address?