Page 16 - Read Online

Liu et al. Intell Robot 2023;3(2):131-43 I http://dx.doi.org/10.20517/ir.2023.07 Page 139

(a) (OURS) (b) JMMD

(c) CDA (d) MKMMD

Figure 3. Confusion Matrix of four different methods on gearbox dataset

By using a joint domain adaptation migration network to de-target the alignment to reduce the joint distribu-
tion differences between two different domains, the accuracy of our proposed method in this fault type has
been dramatically improved. At the same time, a conditional confrontation training module was introduced
to help improve the alignment effect to deal with domain drift. Finally, the most significant differences be-
tween the different categories were obtained. The above-mentioned results provide sufficient evidence of the
transferability of our proposed fault diagnosis method.

4.3. Case 3: SEU gearbox datasets with mixed fault
4.3.1. Data description
We use the bearing and gearbox dataset from Southeast University in China in this experiment [27] . The ex-
perimental platform, DDS, consists mainly of a motor, a planetary gearbox, and a parallel gearbox. The fault
signals are obtained under two different working conditions, 20Hz-0V and 30Hz-2V. The dataset for the gear-
box includes the fault signal of the planetary gearbox in the , , and directions. There are four types of
faults: broken teeth, missing teeth, root faults, and surface faults, and one normal type for healthy working
conditions. The bearing data are available for four types of faults: inner ring, outer ring, rolling element, and
mixed inner and outer rings. In order to evaluate the performance of our approach when dealing with mixed
fault types, gear and bearing fault data from the SEU dataset were combined into a mixed dataset. There are
ninefaulttypesinthismixeddataset, includingfourgearfaults, fourbearingfaults, andonenormaldata. There