Technical Language Supervision for Intelligent Fault Diagnosis in Process Industry

Abstract

In the process industry, condition monitoring systems with automated fault diagnosis methods assist human experts and thereby improve maintenance efficiency, process sustainability, and workplace safety. Improving the automated fault diagnosis methods using data and machine learning-based models is a central aspect of intelligent fault diagnosis (IFD). A major challenge in IFD is to develop realistic datasets with accurate labels needed to train and validate models, and to transfer models trained with labeled lab data to heterogeneous process industry environments. However, fault descriptions and work-orders written by domain experts are increasingly digitized in modern condition monitoring systems, for example in the context of rotating equipment monitoring. Thus, domain-specific knowledge about fault characteristics and severities exists as technical language annotations in industrial datasets. Furthermore, recent advances in natural language processing enable weakly supervised model optimization using natural language annotations, most notably in the form of natural language supervision (NLS). This creates a timely opportunity to develop technical language supervision (TLS) solutions for IFD systems grounded in industrial data, for example as a complement to pre-training with lab data to address problems like overfitting and inaccurate out-ofsample generalisation. We surveyed the literature and identify a considerable improvement in the maturity of NLS over the last two years, facilitating applications beyond natural language; a rapid development of weak supervision methods; and transfer learning as a current trend in IFD which can benefit from these developments. Finally, we describe a framework for integration of TLS in IFD which is inspired by recent NLS innovations.

Labelling of Annotated Condition Monitoring Data Through Technical Language Processing

“…Both paths rely on TLP integrated with a pre-trained language model to address the technical language 1, with possibilities for both technical language supervision through joint representation learning (Löwenmark et al, 2021) and technical language labelling (this work). Image-caption pairs are now replaced by signal-annotation pairs.…”

“…TLS was introduced in 2021 (Löwenmark et al, 2021), with an emphasis on IFD and a case-study to showcase the practical usefulness of the approach. Figure 2 showcases a pipeline both for TLS and TLL (introduced next section) based on the NLS pipeline shown in Figure 1.…”

“…Recent progress in machine learning combines visual and textual information through natural language supervision (NLS) (Radford et al, 2021;Kim et al, 2021;Mu et al, 2022), and recent work in transferring NLS to technical language supervision (TLS) (Löwenmark et al, 2021) illustrates the potential of pre-training on hybrid datasets. However, obtaining large annotated industry datasets can be challenging due to intellectual property and privacy restrictions, and the scarce nature of annotations in comparison to signals.…”

Processing of Condition Monitoring Annotations with BERT and Technical Language Substitution: A Case Study

“…Three months later, the component is replaced and a follow-up annotation is written. This annotation is of critical importance for the possibility of technical language supervision (Löwenmark et al, 2021), as it indicates where the associated signal data should be treated as healthy again. However, only three months after this replacement a BPFO is spotted again, initially on the non-locating side.…”