Abstract

A new method of condition monitoring is proposed. Wavelet transform and fractal geometry are the means. A variety of condition monitoring techniques are currently in use for diagnosis of machinery faults. However, little research into the detection of fractal characteristics in processing signals has been done. This paper establishes wavelet transformation based on the noise signal for analysis of the correlation dimension of the typical working condition. Correlation dimension of the decomposing coefficient of wavelet transformation is calculated to identify the working condition. A field experiment shows the correlation dimension of similar working conditions has similar values while different working conditions show distinct characteristics. The correlation dimensions of the typical working condition of a pump plotted here have a different domain. The experimental results and correlation graph confirmed the proposed method’s feasibility and validity.

This content is only available via PDF.
You do not currently have access to this content.