A New Approach Based on Wavelet Packet Transform, ANN and Genetic Algorithm Applied to End Milling of Mild Steel on Vertical Milling Machine

Surface quality is the most significant characteristic in the manufacturing industry for components produced by any of the machining processes which is indicated by surface roughness. Surface roughness is mainly affected by cutting parameters, machine tool vibrations and tool geometry. In this work efforts have been made to acquire vibration data on spindle housing of a vertical milling center while machining by kistler tri-axial accelerometer using Data Physics eight channel analyzer with recording option, measure surface finish and analyze the influence of cutting parameters on machine tool vibration and surface finish using ANOVA in MINITAB for the designed experiments using Taguchi method. Later the vibration signals are imported into MATLAB. Wavelet toolbox is used for further identification of numerically quoted fault frequencies of the machine tool from the signals one by one using a new method based on wavelet packet transform to find the peak to peak amplitudes of the signals. Further these cutting parameters with the amplitudes are used for the prediction of surface roughness using ANN. The cutting parameters are then optimized, which are being sent to CNC machine to improve the surface roughness and control vibration. This work will provide an engineer who designs a machining center with a tool to predict the vibrations and also provide a guideline for industrial engineers.