The invention relates to fault detection technology, and provides a fault detection method applied to a
nuclear power plant system aiming at the defects of time consumption and labor consumption existing in fault clearance schemes of the conventional
nuclear power plant systems. The method comprises the following steps of: setting a root node of a fault tree as an operating node; determining a possible fault node based on a
logic gate for connecting the operating node and each sub-node and
fault probability; acquiring operating parameters of equipment; judging whether the fault node is a bottom node of the fault tree when judging that fault exists in the equipment corresponding to the possible fault node according to fault conditions associated with the possible fault node and the operating parameters; if so, outputting the equipment which corresponds to the possible fault node and serves as a fault point; and otherwise, setting the possible fault node as the operating node and returning to the second step. The technical scheme provided by the invention completely automatically runs during the clearance of failed equipment, and the clearance speed is high, so the method can actually save time and labor.