Abstract: A mine low-voltage switchgear was taken as object to study its reliability growth process in the commissioning phase in underground coal mine. It was clarified that number of mine-used low-voltage switchgear failure occurrence accompanied with repair obeys non-homogeneous Poisson process according to Laplace hypothesis-testing theory and Bayes theory. The Bayes point and interval estimations for mean time between failures and failure intensity of synchronous time were presented taking the non-information priori distribution, and the actual data obtained in the commissioning phase was used to predict mean time between failures of the mine-used low-voltage switchgear to verify effectiveness of Bayes methods, which provided basis for finding of safety loophole of the switchgear.