Time-frequency analysis of driving sound and application to electric machinery diagnosis

In this paper, we analyze a driving sound model by time-frequency analysis and apply to electric machinery diagnosis. First, a simplified rotator (electric machinery) is represented by a differential equation model. The equation can express the various types of unusual or fault conditions of the rotator by selecting the parameter values. Next, two high resolution time-frequency representations (prescribed minimum square error criteria and minimum cross-entropy criteria) obtained by a set of multiple windows of discrete-time short-time Fourier transform are shown. By implementing as Matlab/SimuLink software systems, various conditions of rotator such as misalignment bearing are analyzed by the high resolution time-frequency methods. Further, the actual pump is diagnosed to show the effectiveness of our method.