Application of Al2O3/iron-based composite abrasives on MAF process for inner surface finishing of oval-shaped tube: predicting results of MAF process using artificial neural network model

Abstract

The oval-shaped tubes are the valuable product for using to transport the clean gas and high-purity solutions in the field of aerospace, semiconductor, LCD, and medical industries. However, due to their complex shape, it could significantly difficult to improve their inner surface quality by the conventional processes. In this study, the application of Al2O3/iron-based composite abrasives on magnetic abrasive finishing (MAF) process was used to finish the inner surface of oval-shaped SUS 316L tubes. Their inner surface was finished by a mixture of magnetic abrasive tools of Al2O3/iron-based composite abrasives with Fe powders, and light oil. An artificial neural network (ANN) model was used to predict the results of MAF process and then was compared with the actual test. It is found that the results predicted by the ANN model have significant agreement with the actual test results. The correlation coefficient (R2) for both training and testing of surface roughness on span and rise portion is larger than 0.92, and RMSE values for both cases are extremely small (less than 0.01). The actual test results showed that the inner surface of oval-shaped SUS 316L tube was smoothly improved from 0.24 μm to 0.05 μm and 0.04 μm on the rise and span portion, respectively. The application of Al2O3/iron-based composite abrasives on MAF process is an efficient method for inner surface finishing of the oval-shaped tubes.

Keywords

• Al2O3/iron-based composite abrasives
• MAF process
• Oval-shaped SUS 316L tube
• Artificial neural network (ANN)
• Surface roughness
• EDS elemental mapping