The fault diagnosis of rolling bearings is of utmost importance in industrial applications to ensure mechanical systems' reliability, safety, and economic viability. However, conventional data-driven fault diagnosis techniques mainly depend on a pre-existing dataset with complete failure modes and knowledge to serve as the training data, which may not be available or accessible in some crucial industrial scenarios. This can limit the practicality of these methodologies in real-world industrial applications. This paper addresses this issue by developing a novel digital twin-enabled domain adversarial graph network (DT-DAGN). The main contributions of this paper are as follows: 1) the development of a comprehensive and accurate digital twin model for rolling bearings that includes a dynamic simulation of the bearing's operational status using only its structural parameters and failure severity/size to obtain the system's vibration response, and 2) the development of a novel graph convolutional network-based transfer learning framework to transfer knowledge from simulated datasets to measured datasets, enabling effective fault diagnostics of bearings with limited knowledge. A series of experiments are applied to validate the efficacy of the developed methodology.

Convolutional neural network (CNN)-based fault detection approaches based on multisource signals have attracted increasing interest from the research community and industrial practices, thanks to the powerful feature representation capability of CNN and the rapid development of sensor technology. Various strategies have been applied in existing CNN-based diagnostic models to learn features from 1-D real-valued multivariate data. However, the distribution gap and the intrinsic correlations among multisource mechanical signals during the learning process have been rarely considered, which may lead to suboptimal fault identification results. To tackle this issue, this article proposes a cross-modal fusion convolutional neural network (CMFCNN) for mechanical fault diagnosis, which performs modality-specific and cross-modal feature representation on multisource data. Specifically, CMFCNN adopts two parallel modality-specific networks and a cross-modal knowledge-sharing network to fully explore independent and shared features from the multisource mechanical signals. To achieve effective feature propagation and fusion, a cross-modal fusion module is introduced to integrate cross-modal features and pass the fused information to the next layer. Moreover, to alleviate overfitting and achieve a better diagnostic performance of the framework, an online soft-label training algorithm is adopted in the CMFCNN training phase. Extensive experimental results on the cylindrical rolling bearing dataset and the planetary gearbox dataset validate that the proposed CMFCNN outperforms seven state-of-the-art methods significantly, especially under strong noise conditions.