A Health Data Map-Based Ensemble of Deep Domain Adaptation Under Inhomogeneous Operating Conditions for Fault Diagnosis of a Planetary Gearbox

Abstract

A deep learning model trained under a specific operating condition of the gearbox often experiences an overfitting problem, which makes it impossible to diagnose faults under different operating conditions. To solve this problem, this paper proposes an ensemble of deep domain adaptation approaches with a health data map. As a fundamental approach to alleviate the domain shift problem due to inhomogeneous operating conditions, the vibration signal is transformed into an image-like simplified health data map that visualizes a tooth-wise fault of the gearbox. The simplified health data map enables the use of a conventional convolutional neural network (CNN) model. To solve the remaining domain shift problem even with the simplified health data map, this study employs a maximum classifier discrepancy (MCD), which is a typical domain adaptation method. To further enhance its performance, a discrepancyscale factor-based MCD and its ensemble approach are proposed. The proposed method is demonstrated with a 2 kW planetary gearbox testbed operated under stationary and non-stationary speed conditions. The results present that the proposed method outperforms conventional CNN and MCD even under the inhomogeneous operating condition of the gearbox.

Domain Knowledge-Informed Synthetic Fault Sample Generation with Health Data Map for Cross-Domain Planetary Gearbox Fault Diagnosis

“…HDMap can visualize vibration signals in a more intuitive manner by considering modulation characteristics caused by the revolving planet gears, which allows for the independent interpretation of fault signatures from the gears. Subsequent validation studies have demonstrated that HDMap can outperform time, frequency, and time-frequency domain signals when used as input to the deep learning-based fault diagnosis model of the planetary gearbox [11]. However, a previous study has shown that HDMap still requires further domain adaptation to solve the remaining domain shift problem when the domain gap is significant.…”

“…One such physics-guided signal representation approach for gearbox health monitoring is health data map (HDMap) that has recently been introduced [10]. It has been shown that HDMap can enhance the performance of DL-based fault diagnosis methods compared to raw input signals [11].…”

A review on deep learning based condition monitoring and fault diagnosis of rotating machinery

“…Ha and Youn, 2021 126 performed fault diagnosis of planetary gearbox with the use of CNN and maximum classifier discrepency (MCD). In this study, vibration signals were first converted into image map using time syncronous averaging (TSA) which visualizes the fault toothwise.…”

Gear Fault Diagnosis Method Based on Multi-Sensor Information Fusion and VGG

“…Although traditional machine learning methods such as SVM have good performance on small sample datasets, these methods require a lot of expertise and domain knowledge, and the feature engineering steps need to be done manually. Since deep learning does not require manual feature extraction, deep learning has been applied to gear fault diagnosis by scholars [ 15 , 16 , 17 , 18 , 19 , 20 ]. Nguyen et al [ 21 ] used ANC and DNN to build a sensitive gear fault diagnosis model that can effectively and accurately diagnose the types of faults with different shaft speeds and have high classification accuracy.…”