Research on Fault Fuzzy Clustering Method for Reducing False Alarm by New Three States Demarcation

Reducing false alarm have important signification for accurate diagnosis of equipments´ faults. The classification of built in test is introduced in the paper. Then the equipments status is redefined to three kinds of state (namely as fault state, quasi fault state, health state). Among them, the intermittent fault, transient fault, and the critical fault are included in the quasi fault. The 3-state Markov model is contrasted and analyzed with 2-state Markov model on theory. The advantage to divide the equipments status as new three kinds of states is pointed out. Joining together the method of fuzzy clustering theories, running fault condition of rotor system are classified and identified. The intermittence fault and transient fault of rotor are successfully separated on the verifying platform. It is show by computation and analysis that the method of judging the threshold value of the equipments status is viable by choosing the F-statistic. The method introduced in the paper has important reference value for reducing false alarm of the equipment.