Harmonic-Wavelet and Approximate-Entropy Extracted Weak Noise

The objective of this paper is to introduce Newland Harmonic wavelets and Pincus (1991) approximate entropy are applied to extract weak cycle jumps. Because some weak vibration signals contained strong noises are very important for fault forecast. Based on harmonic wavelet mesh plot of the vibration signals, we can induce contours of the mesh map with log space, then, we offer the time-frequency profile plot from the mesh map. This time-frequency profile plot is corresponding to decompose level on harmonic wavelets. Final, by computing approximately entropy in each level time-frequency profile plot, the weak cycle signal can be easily extracted from strong noise.