Fault Diagnosis System For Rotating Machinery Based On Order Tracking Analysis

Vibration signals generated by rotating machinery contain lots of information about running status and fault type, thus we can fulfill device status monitoring and fault diagnosis through vibration analysis. Order tracking analysis is a signal processing method to analyze the relationship between the signal and the frequency base. Using rotating speed as the frequency base to analyze the vibration, the resonant vibration component caused by equipment failure can be extracted. Thus order tracking analysis is a useful method for device fault diagnosis. In this article, we have discussed the shortcomings of the existing order tracking algorithms and proposed a new resampling method for order tracking. The experimental results demonstrate that the proposed method improves computation accuracy and efficiency.This article studied the vibration characteristics of rotating machinery and applied pattern recognition method on device fault diagnosis. Then a fault diagnosis system was designed for rotating machinery, which is built on order tracking analysis and support vector machine method. The turbocharger quality inspection experiment proof that SVM has a better ability to identify mechanical fault type and device condition monitoring than artificial neural network method.The system software is built on the LabVIEW platform, which has very high scalability, versatility and good algorithm integration ability. The system hardware is based on PCI data acquisition board and acceleration sensor, which is used to capture vibrations. Then software algorithms are used to process vibration signals and recognize device fault type. The main content and research work of this article are as follows:(1) Studied the machinery failure mechanism and the vibration frequency characteristic of rotating machinery. Proposed a new resampling method for computed order tracking, which improves computational accuracy and lower algorithm complexity.(2) Applied the SVM method on device monitoring and product quality inspection. Through turbocharger quality inspection experiment, it was proved that SVM has very excellent detection capabilities on product quality inspection. And the SVM method has a better classification accuracy than artificial networks. Then we have designed a SVM based multi-type classifier for fault diagnosis.(3) Completed system software coding, debugging and testing work. Built the motor vibration testing experimental platform, verify the robustness and classification accuracy and computation efficiency of the system.