Fault Detection of CNC Machines from Vibration Signals Using Machine Learning Methods

Abstract

Every machine used in industry has become indispensable in today’s production process. As a result, malfunctions of these machines have devastating consequences. In this context, it is very important to diagnose faults before or during the start-up phase. In this paper, the vibration signals obtained from the computerized CNC machine drill bit whiledrilling the metal plate. For fault diagnosis, feature extraction was first performed. A total of 21 properties were extracted from the vibration data in the time domain, frequency domain, and time-frequency domain. The features that were taken at certain intervals from the dataset were doubled to get more accurate results. After the feature extraction process, the data was normalized. The normalization process was determined as min-max and z-score. After the normalization process, classification was carried out by Support Vector Machine (SVM) and K-nearest neighbors (kNN) methods. As a result of these processes, %96,45 and %94,37 accuracy rates for SVM and kNN were obtained respectively.