Lab-Scale Vibration Analysis Dataset and Baseline Methods for Machinery Fault Diagnosis with Machine Learning

The monitoring of machine conditions in a plant is crucial for production manufacturing. A sudden failure can stop and cause loss revenue. vibration signal good indicator its condition. This paper presents dataset signals from lab-scale machine. contains four different types conditions: normal, unbalance, misalignment, bearing fault. Three learning methods (SVM, KNN, GNB) evaluated the dataset, perfect result was obtained by one on onefold test. performance algorithms using weighted accuracy (WA), since data are balanced. results show that best-performing algorithm SVM with WA 99.75% fivefold cross-validations. provided form CSV files an open free repository at https://zenodo.org/record/7006575 .