Research on Predicting Machining Surface Roughness Based on Neural Network

The quality of the machined surface largely affects some physical properties such as wear resistance, corrosion resistance and fatigue resistance of parts, as well as functions such as service life and reliability. Surface quality is one of the requirements of customers for parts processing. The biggest indicator of the surface quality of machining is the size of the roughness. A neural network is used to predict the surface roughness of the machining workpiece, and stainless steel is selected as the workpiece material. Different cutting experiments were carried out at different cutting speeds, different feed rates, and different cutting depths. Aiming at the change of cutting conditions in the cutting process, the neural network model is used to predict the surface roughness during the process. The results of neural network prediction of workpiece surface roughness show that the numerical error predicted by this method is small and the accuracy is high. The prediction model has good prediction accuracy, and its maximum error does not exceed 2%. The model can predict the surface roughness after processing under different combinations of cutting speed, feed and depth of cut parameters, and has important guiding significance for the control of the surface quality of parts.