Bayesian network parameter learning method integrating expert prior knowledge

The invention provides a Bayesian network parameter learning method integrating expert prior knowledge, and aims at improving the precision of a parameter learning result obtained under the small sample condition. The method comprises the implementation steps that normal distribution represented by the value possibility of a Bayesian network parameter is acquired; the value range of a normal distribution standard deviation is acquired; a target function needing to be solved for approaching normal distribution by adopting beta distribution is acquired; a target function expression is simplified, whether the target function has a minimum value or not, if yes, values of a position parameter and a shape parameter of beta distribution are calculated, and otherwise, coefficients in the target function expression are adjusted slightly; a parameter learning model integrating the expert prior knowledge is acquired; and probability distribution estimation values of various variables integratingthe expert prior knowledg