Axial Flow Pump Fault Diagnosis Based on VMD Hybrid Domain Feature and DBN
Since traditional fault diagnosis methods cannot fully capture the fault features,this paper proposes a fault diagnosis method based on mixed domain features of variational mode decomposition(VMD)and deep belief network(DBN).The pressure pulsation of the faulty axial flow pump is decomposed into several intrinsic mode functions using VMD,the energy and entropy characteristics of each IMF component with large correlation are calculated,and combined with the time domain and frequency domain features extracted from the original signal to form a mixed domain feature.Genetic algorithm is introduced to optimize the combination of fault features,remove redundant features,and input the optimized fault features into DBN for recognition and classification.The results show that compared with the single feature set,the fault diagnosis based on the mixed domain features offers better performance;Compared with support vector machine,deep belief network has more advantages in fault diagnosis,the classification accuracy can reach 95％.This indicates that the pressure pulsation signal of the pump housing can better reflect the fault characteristics of the axial flow pump model,the pump housing measurement point can be used as the fault detection point.
NETL
NSTL
万方数据
