An improved multipoint optimal minimum entropy deconvolution adjusted method for the diagnosis of rotating machinery under variable speed conditions

Deconvolution algorithms have been proved to be effective tools for fault feature extraction and diagnosis of rotating machinery. As a non-iterative algorithm, multipoint optimal minimum entropy deconvolution adjusted (MOMEDA) shows its advantage in computation efficiency. However, MOMEDA focuses only on the extraction of impulses with precise periodicity while the fault feature excited under time-varying speed may not meet this requirement. To solve this problem, this paper proposes an improved method named as MOMEDAang to extend the application scenarios of MOMEDA under time-varying speed condition. Instead of using a periodic target vector, the new method calculates the time-domain intervals of fault impulses according to their nature of isometric distribution in angle domain, which can be free from the variation of speed. Moreover, the harmonic modulation intensity and average Gini index are further defined in this paper to highlight the inherent cyclic modulation order of filtered signal and conduct the health condition identification of monitoring equipment. The results of simulation and experimental cases reveal effectiveness and efficiency of the proposed method.