Support Vector Machines (SVMs)

• Published in: Condition Monitoring with Vibration Signals pp. 259 - 277
• Main Authors: Ahmed, Hosameldin, Nandi, Asoke K
• Format: Book Chapter
• Language: English
• Published: Chichester, UK Wiley 2019; John Wiley & Sons, Ltd
• Edition: 1
• Subjects: binary classification problem; feature extraction; feature selection; machine fault diagnosis; multiclass SVM approach; normalisation; support vector machines; SVM classifier; Support vector machines; Training; Condition monitoring; Vibrations; Rotating machines; Machine learning; Kernel
• ISBN: 9781119544623; 1119544629
• DOI: 10.1002/9781119544678.ch13

Support vector machines (SVMs) are one of the most popular machine learning methods used to classify machine health conditions using the selected feature space. In machine fault detection and diagnosis, SVMs are used for learning special patterns from the acquired signal; then these patterns are classified according to the fault occurrence in the machine. This chapter presents essential concepts of the SVM classifier by giving a brief description of the SVM model for binary classification. Then, it explains the multiclass SVM approach and different techniques that can be used for multiclass SVMs. A considerable amount of literature has been published on the application of SVMs and variants in diagnosing machine faults. Most of these studies introduced pre‐processing techniques that include normalisation, feature extraction, transformation, and feature selection. The data produced during the pre‐processing step represent the final training set that is used as input to SVMs.