model selection;  Bayes method;  Uncertainty

In previous studies on model selection, it is generally assumed that a model is selected, and then the distribution family determined by the model is compared to find the optimal distribution function and numerical characteristics, ignoring the uncertainty of the model itself. The method and application of Bayes method in model selection are introduced, and an example is given to show that using Bayes method to select a model can not only reduce the influence of model uncertainty in model selection, but also extend the model according to the actual situation and the deepening of problem understanding.