Machine Fault Detection Using Vibration Signals and Improved Fuzzy Clustering Algorithm

This paper will present a new solution for machine fault detection based on the vibration signals. The solution will used in improved fuzzy Gustaffson – Kessel clustering method to generate the classification data centers characteristic for different states of the machines. The Gustaffson – Kessel method offers a modified euclidian distance, which allows betters separation borders between data clusters. The model will be tested with the vibration signals collected from the standard CASE Bearing Data Sets to show the high accuracy of the results.