Composite Fault Diagnosis of Aviation Generator Based on EnFWA-DBN

Because of the existence composite faults, which consist both short-out and eccentricity characteristics output voltage internal magnetic field aviation generators are less different than those single faults; this causes fault to be difficult identify. In order solve problem, paper proposes a diagnosis method using an enhanced fireworks algorithm (EnFWA) optimize deep belief network (DBN). The generator model is built according finite element (FEM), whereas combinations faults obtained simulations. EnFWA used train DBN obtain best structure. Meanwhile, extreme learning machine (ELM) classifier performs classification on test data. results show that pinpoint accuracy can achieved proposed in generators.