Abstract: Nuclear power plant (NPP) is a very complex system, which needs to collect and monitor vast parameters, so it’s hard to diagnose the faults of NPP. An ensemble learning method was proposed according to the problem. And the method was applied to learn from training samples which were the typical faults of nuclear power plant, i.e., loss of coolant accident (LOCA), feed water pipe rupture, steam generator tube rupture (SGTR), main steam pipe rupture. And the simulation results were carried out on the condition of normal and invalid and absent parameters respectively. The simulation results show that this method can get a good result on the condition of invalid and absent parameters. The method shows very good generalization performance and fault tolerance.