Refined frequency monitoring based on characteristic excitation with application to early fault diagnosis of thin plate damage

Abstract

This article presents a cascade-structured refined frequency monitoring (RFM) method for high-resolution detection with application to early fault diagnosis of thin plate table. Specifically, in frequency monitoring, the proposed RFM framework fully excites characteristic information from vibration signal subjected to uncertainty and noise, which achieves online monitoring for frequency slight change. The adaptive detection threshold is systematically exploited based on the priori information. As a solving skill for monitoring and diagnosis problems, proposed method has two significant advantages over previous methods. First, frequency monitoring is reformulated into a parameter estimation problem associated with the characteristic equation. This procedure aims to decouple frequency from amplitude and phase, which reduces the dependence of the diagnosis on the parameters. Moreover, by combining the benefits of characteristic observer and tracker, frequency resolution can be improved through effective suppression of uncertainty and noise. Secondly, the synthesis strategy of frequency monitoring and fault detection is developed to potentially enhance the reliability of the diagnostic algorithm. Meanwhile, the proposed method can be incorporated into fault-tolerant controller to implement the integrated technology. The effectiveness of the proposed scheme is tested by numerical simulations and platform experiments.