The technology of accelerated degradation testing has become an efficient approach to evaluating the reliability of the degrading product. However

the method of analyzing accelerated degradation data

which excessively dependents on subjective experience

results in the inaccuracy of the reliability evaluation. In the paper

a more objective method based on acceleration factor constant principle is proposed. First

the changing rules of the parameters of degradation models are deduced according to the acceleration factor constant principle. Next

the effectiveness of the degradation data under each accelerated stress is identified through the parameter equation independent of accelerated stress. The key is that a t statistic is constructed to verify whether the parameter estimates satisfy the parameter equation. Then

the acceleration models of the parameters dependent on accelerated stress are constructed. Last

the effective accelerated degradation data is utilized to estimate parameters

so the reliability under the normal stress level can be extraplolated. The proposed method is demonstrated by taking the inverse Gaussian process as an example. Both the simulation test and case application indicate that the study of the paper provides a more objective and reasonable technical approach to reliability evaluation based on accelerated degradation data.