Research on Surface Roughness Prediction of Turning Parts Based on BP Artificial Neural Network

Surface roughness of parts is an important index for the quality of processing. An accurate and efficient model for surface roughness prediction can provide a reliable constraint or objective function for the processing parameter optimization. In the part design and actual processing, the surface roughness value (Ra) is taken from the national standard specification series, these series can be regarded as the corresponding category. Therefore, in this research, a new method is presented to predict the surface roughness by classifying the Ra under different cutting conditions. The effects of network structure and learning algorithm on the prediction results were discussed. Finally, a 5-3-3-3 BP Artificial Neural Network (ANN) structure with Levenberg-Marquardt (LM) algorithm were used to build the model. The results show that the prediction accuracy of the model was as high as 97.44%, and the surface roughness value of the turning parts can be predicted well.