Classification of acceleration signal in milling process of FCD 450 cast iron for surface roughness using tuned support vector mechanics

Accurate identification and characterisation of surface roughness in industrial applications play a crucial role in ensuring product quality, reliability, and performance. The tuned Support Vector Mechanics (SVM) is used to effectively process and analyse the acceleration data of a cutting process for surface quality identification. In the machining process, a series of feature extraction techniques are employed to capture acceleration signals from cutting tool and workpiece. The SVM model incorporates the extracted features to build a robust classification framework capable of accurately identifying different levels of surface roughness. Experimental results show a positive but nonlinear correlation (r = 0.6543) between acceleration and surface roughness (Ra). In the first experiment, the tuned medium Gaussian SVM achieved an accuracy of 85.53% and an F1 score of 84.93%, outperforming other tested SVM kernels. The model also demonstrated stable performance in a second experiment with an accuracy of 84.0%. Furthermore, surface quality was successfully categorised into five discrete levels. The tuned of medium Gaussian SVM consistently achieves high classification accuracy across different surface roughness levels, exhibiting robustness and reliability in machining processes.