Recent advances in reasoning artificial intelligence (AI) agents powered by language models (LMs) and custom tools open new opportunities for AI-assisted condition monitoring (CM) involving unlabelled but annotated, complex industrial data. Technical language annotations written by domain experts include unstructured information regarding machine condition, maintenance actions, and tacit knowledge. This thesis investigates how LMs and LM agents can improve human-machine interaction and facilitate training of AI models on CM industry data using annotations as surrogate ground-truth labels. The main contribution is the introduction of technical language supervision (TLS) to address the long-standing gap between idealised labelled lab datasets and complex unlabelled field data, and the development of AI agents for condition monitoring including a multimodal vector store with domain specific retrieval and generation modules.

Specific contributions are: (1) the introduction and implementation of TLS for CM through contrastive learning with annotated sensor data, including a literature survey and implementations of zero-shot fault diagnosis on unlabelled industry data; (2) the creation of a method to improve technical language processing by augmenting out-of-vocabulary technical words with natural language descriptions and evaluating semantic similarities of technical language representations; (3) the development of a human-centric method for language-based fault classification using visualisation and clustering; (4) the development of an open source chatbot agent which facilitates natural language interaction with industry data and models through a custom CM vector store with data-specific retrieval augmented generation, LM analysis of annotations and hierarchy data, and LM assisted CM; (5) the compilation of a publicly available annotated industry dataset; (6) an investigation of specific CM data processing challenges, such as different data modalities, time-delays between annotations and signal properties, component-specific noise and feature levels, and non-linear fault development over time.

The results of the studies indicate that annotations are a viable substitute for labels when processed with regard to the technical language therein, and integrating LM-based agents on annotated CM data facilitates answering queries corresponding to industrial analysis tasks. By augmenting out-of-vocabulary technical words with natural language descriptions, LM performance can be improved, as demonstrated in initial work on classifying technical fault descriptions with the BERT LM improving accuracy from 88.3% to 94.2%, thereby halving the error rate.

In the industrial datasets analysed, gathered from kraftliner paper machines over four years, the most common faults and alarms were cable and sensor faults, while bearing faults were the most common causes of follow up analysis and maintenance stops. Clustering CM data based on both signal and language properties indicates that cable and sensor faults can be differentiated from bearing faults with an F1-score of 92.6%. The usefulness of the developed agents was evaluated in typical CM workflows, and the results indicate that AI agents with custom tools are capable of generating historic insight and meaningful fault descriptions. In particular, using a custom multimodal CM retrieval augmented generation approach with a custom CM vector store, the false alarm rate for sensor and cable faults is shown to be lowered from over 80% in current work flows, to under 30% with the proposed method. This suggests that false and redundant alarms which negatively impact maintenance planning by prompting time-consuming human analysis can be reduced.

The main takeaways of this thesis are that annotations can facilitate the development of AI models on field industry data, and bring meaningful historic insights. This approach has the potential to augment existing CM practices by reducing false alarm prevalence, providing more meaningful alarms, and improving upskilling.