Abstract: It is usually difficult to determine the hyperparameters of components for reliability generic database involved in nuclear power plant PSA.Meanwhile,inaccurate hyperparameter values adversely impact the uncertainty analysis of core damage frequencies and Bayesian updating of component reliability parameters.To seek a solution,with empirical Bayesian approach,research was performed on the statistical models and parameter estimation methods that help to build the reliability generic database from the component failure records.The mathematical mechanisms behind the method of moment estimation and the maximum likelihood method were the focus of discussion,with an example demonstrating the application of these two methods.