Vibration Signal Analysis for Surface Roughness Prediction in Machining

Signal processing was employed in this study to monitor and predict the surface condition during the turning process. Vibration signals of the cutting tool were collected during the machining of a stainless-steel workpiece with a carbide insert, and subsequently processed. Initially, the Fourier Transform (FFT) was used to separate the machine's vibration signature and noise. Subsequently, the optimal frequency band was determined through Power Spectral Density (PSD) estimation using the Welch method, extracting crucial information for analysis. The correlation between vibration signals and surface roughness was examined by calculating the Root Mean Square acceleration (GRMS) of the obtained PSD curves. Finally, a mathematical prediction model was constructed via a simple linear regression equation between GRMS and surface roughness. The results demonstrated strong agreement between predicted and experimental data, with low coefficients of Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and Mean Absolute Error (MAE) (0.025, 0.1581, and 0.1174, respectively), affirming the accuracy of the developed model. The findings of this study significantly contribute to enhancing manufacturing efficiency and cost optimization by leveraging signal processing in machining processes.