Fault Diagnosis of Heat Removal Pump Based on Bayesian Networks

The Bayesian fault diagnosis networks of heat removal pump were built based on its fault tree. As the basic conversion method of FT/BN was shown, the expression of multi-formed logic and condition probability table of non-determinism was analyzed. The root nodes’ probability was gain by the subsection voting with specialists. The node significance, sensitivity analysis and bidirectional diagnoses provide decision for equipments in designing, using and maintaining.