Integrating Expert Knowledge With Domain Adaptation for Unsupervised Fault Diagnosis

Abstract

Data-driven fault diagnosis methods often require abundant labeled examples for each fault type. On the contrary, real-world data is often unlabeled and consists of mostly healthy observations and only few samples of faulty conditions. The lack of labels and fault samples imposes a significant challenge for existing data-driven fault diagnosis methods. In this article, we aim to overcome this limitation by integrating expert knowledge with domain adaptation (DA) in a synthetic-to-real framework for unsupervised fault diagnosis. Motivated by the fact that domain experts often have a relatively good understanding on how different fault types affect healthy signals, in the first step of the proposed framework, a synthetic fault dataset is generated by augmenting real vibration samples of healthy bearings. This synthetic dataset integrates expert knowledge and encodes class information about the faults types. However, models trained solely based on the synthetic data often do not perform well because of the distinct distribution difference between the synthetically generated and real faults. To overcome this domain gap between the synthetic and real data, in the second step of the proposed framework, an imbalance-robust DA approach is proposed to adapt the model from synthetic faults (source) to the unlabeled real faults (target) which suffer from severe class imbalance. The framework is evaluated on two unsupervised fault diagnosis cases for bearings, the CWRU laboratory dataset and a real-world wind-turbine dataset. Experimental results demonstrate that the generated faults are effective for encoding fault type information and the DA is robust against the different levels of class imbalance between faults.

Delamination prediction in composite panels using unsupervised-feature learning methods with wavelet-enhanced guided wave representations

“…Here, the initial state of the structure is damaged with an edge notch. However, we can use this dataset considering the initial state as a baseline and upcoming states as delaminated [29].…”

Deep Unsupervised Domain Adaptation with Time Series Sensor Data: A Survey

“…Sensors . Common types of sensors used for rolling bearing signal acquisition are accelerometer sensors [ 48 , 52 ] as well as microphones [ 46 ]. Accelerometer sensors measure the health of machinery by being fixed directly near the machinery to collect the vibration signals generated by the mechanical vibrations.…”

Combining Knowledge and Deep Learning for Prognostics and Health Management

“…Especially inducing data-driven models with knowledge based on experience and physics has been successful. Recent work in the PHM domain has shown to increase explainability and enable root cause analysis by taking into account input from human experts (Steenwinckel et al, 2021), make models more robust and decrease the amount of data needed through the simulation of data based on expert knowledge (Wang, Taal, & Fink, 2021), and increase overall performance by extending the feature space (Chao, Kulkarni, Goebel, & Fink, 2022). Even though recent work in the direction of combining knowledge with data is promising, a lot of untapped knowledge in the shape of physical models (e.g.…”