Investigating the Tribological Characteristics of Burnished Polyoxymethylene—ANFIS and FE Modeling

Polymers are utilized in numerous tribological applications because of their excellent characteristics; for example, accommodating shock loading and shaft misalignment. A high surface finish is required to ensure consistently good performance and extended service life of manufactured polymeric components. Burnishing is the best choice as a finishing process for this study due to its ability to increase hardness, fatigue strength, and wear resistance and also introduce compressive residual stress on the burnished workpiece. Due to the complexity and uncertainty of the machining processes, soft computing techniques are preferred for anticipating the performance of the machining processes. In this study, ANFIS as an adaptive neuro-fuzzy inference system was applied to anticipate the workpiece hardness and surface roughness after the roller burnishing process. Five burnishing variables, including burnishing depth, feed rate, speed, roller width, and lubrication mode, were analyzed. A Gauss membership function was used for the training process in this study. The predicted surface roughness and hardness data were compared with experimental results and indicated that the Gauss membership function in ANFIS has satisfying accuracy as high as 97% for surface roughness and 96% for hardness. Furthermore, the generated compressive residual stress on the burnished surface was studied by a 2D finite element model (FEM). The simulated results of residual stress were validated with the experimental results obtained from X-ray diffraction (XRD) tests.