The traditional reliability analysis has such drawbacks that it must be dependent on the large sample statistical data

and use the probability statistics method to calculate the reliability of gearbox.Based on these

a new method which uses gearbox running status information to implement reliability assessment under the condition of small sample is proposed.It is the reliability evaluation method based on the normalized wavelet information entropy.The reliability assessment based on the normalized wavelet information entropy method is to make decomposition and reconstruction computations on the vibration signals of gearbox by virtue of the second generation wavelet package to get decomposed signals in sub-frequency bands.Then

the relative energy of these decomposed signals and the normalized wavelet information entropy are calculated.Finally

the degree of reliability which acts an important reliability index is acquired by using the normalized wavelet information entropy to evaluate reliability of gearbox.The application of the method to evaluate reliability of gearbox demonstrated that the proposed approach is effective and reasonable.This study provided a way to calculate reliability of gearbox which is lack of large sample statistical data and is independent of probability statistics.