Abstract: A Bayesian network fault diagnosis model is established based on analysis of maintenance data in order to solve the problem of testing trouble in medical laboratory equipments resulted from the complicated structure of medical instruments and inadequate maintenance data. The Pgmpy toolkit is introduced based on the analysis of maintanace data, and the process of establishment of network and learning of the network parameters through a case study of the construction of diagnostic model of urinograph ulyt 500B. The experimental results show that the reasoning results of the diagnosis system constructed by this method are consistent with the reality.

Key words: Bayesian network, urine analyzer, fault diagnosis, Pgmpy