Wavelet Fractal Technology and Its Applications to Nonstationary Fault Diagnosis
Zi Yanyang,He Zhengjia,Zhang Zhousuo
(Xi'an Jiaotong University, Xi'an 710049,China)

Abstract: Wavelet fractal technology (WFT) is proposed according to the unity that possesses multiscale decompositions and self-similarity in both wavelet transform and fractal theory. In order to overcome disadvantage of the general method calculating box dimension (BD) of discrete signals, an improved algorithm of BD is given. An empirical formula is derived for choosing scale-invariant region of periodic or quasi-periodic signals. Successful application has been achieved to detect looseness fault in bearing bushing of turbo-generator set. The results show that WFT can provide a new effective technology to reveal nonstationary fault of mechanical equipment.
Keywords: wavelet fractal;box dimension;nonstationary;mechanical equipment;fault diagnosis