Classification of Acceleration Signal in Machining Process for Surface Roughness Identification using Tuned Support Vector Mechanics

Abstract

In industrial applications, accurate surface roughness identification and characterization are essential for ensuring product quality, dependability, and performance. The suggested technique efficiently processes and examines the acceleration data of a cutting operation for surface quality detection using customized Support Vector Mechanics (SVM). The suggested method extracts pertinent data from the acceleration signals using a number of feature extraction approaches. Incorporating the collected features, the improved SVM model creates a strong classification framework that is capable of precisely recognizing various degrees of surface roughness. An extensive dataset made up of acceleration signals from various machining operations and surface roughness conditions is used to assess the performance of the proposed approach. Using this dataset, the hyper-tuning of the SVM model is trained and tested to determine its classification precision and generalizability. The experimental findings show that, when compared to conventional classification methods, the customized SVM model performs better. The suggested method regularly demonstrates durability and reliability while achieving excellent classification accuracy across a range of surface roughness levels. The suggested method provides a workable and effective solution for automating surface roughness identification, enabling in-process quality control and real-time monitoring.