A smart diagnosis system based on automatic recognition of multiple rotor faults

Abstract

Diagnosing rotor faults is considered as one of the most vital tools for health maintenance of rotating machinery. In this work, a smart diagnosis system based on automatic recognition of multiple rotor faults is developed. As rotor faults are comparatively complicated, the reasoning mechanism of composite nesting probability reasoning network knowledge expression and multiple reasoning methods are adopted. Besides, methods such as fuzzy pattern recognition and image analysis are also applied to conduct automatic recognition of fault symptoms like rotor vibration spectrum, shaft centerline orbit, and transient features. Also, this article attempted to propose artificial neuron network study and diagnosis methods on the basis of fractal features of faults, set up a smart diagnosis system based on automatic recognition of multiple rotor faults, and verify the feasibility of system diagnosis using a rotor practical fault diagnosis case.