Abstract: With the wide use of helicopters, it becomes more and more significant to detect the fatigue crack in critical component during the period of its high strength running, and conventional non-destructive testing technology cannot monitor the generation and changes of cracks in real time. TB6 titanium alloy, a common component material of helicopter, was fatigue fracture tested. The acoustic emission (AE) signals during the test were collected by AE detection technology. Wavelet energy coefficient historical variety analysis method, combined with characteristic parameter historical variety analysis and wavelet analysis, was adopted to study the characteristic variation of signal. The results show that the fatigue fracture process of the material has experienced four stages, and the AE characteristics of each stage have obvious turning points. At the beginning of crack initiation, characteristic parameter and wavelet energy coefficient have obvious jump, and the center frequency of the AE signal tends to move toward the high frequency with time. These characteristic laws provide a reference for the AE dynamic monitoring and health evaluation of TB6 titanium alloy components.