PREDICTION OF FLANK WEAR AND SURFACE ROUGHNESS BY RECURRENT NEURAL NETWORK IN TURNING PROCESS

Abstract

Tool wear and surface roughness plays a significant role for proper planning and control of machining parameters to maintain product quality in order to achieve sustainable manufacturing. The machining process is complex, thus it is very difficult to develop a comprehensive model. This study proposes an innovative model of flank wear and surface roughness prediction for turning of AISI 1040 steel based on a recurrent neural network (RNN). In this study, the flank wear and surface roughness was measured during turning at different cutting parameters. Full factorial experimental design applied aims to increase the confidence limit and reliability of the experimental data. The input variables for the proposed RNN network were cutting speed, feed rate, depth of cut and the homogeneity extracted from the surface texture images obtained by using grey level co-occurrence matrix. The result shows that the accuracy of the flank wear and surface roughness prediction using RNN can reach as high as 97.05% and 96.58%, respectively.