Maintenance of medium voltage switchgears is a complex task that requires extensive expertise. However, a substantial portion of this knowledge is often difficult for the next generation of field technicians to access and apply effectively. For instance, maintenance reports – which document system issues and corrective actions – represent a valuable knowledge source yet frequently underutilized due to their unstructured and sparse nature. In this context, recent advancements in Large Language Models (LLMs) and Retrieval-Augmented Generation (RAG) techniques have been explored to analyze historical maintenance reports and provide real-time guidance to field technicians. However, chatbot-based interactions remain constrained by the stochastic nature of responses, challenges related to domain and industry specificity, and risks of hallucination. This study introduces a novel approach that leverages LLMs to systematically extract structured, machine-readable rules – i.e., sequences of actions used to resolve specific issues – directly from maintenance reports. A human-in-the-loop approach is proposed, where a human expert validates each extracted rule, mitigating inaccuracies and enhancing quality control. These validated rules are eventually stored in a dedicated rule corpus and later deployed in a deterministic rule engine, ensuring reliability and consistency. An industrial case study evaluates the effectiveness of this approach by comparing LLM-generated rules with those manually curated by domain experts. By shifting from conventional chatbot-like interaction to a more human-centric structured rule extraction, our methodology enhances reproducibility, reliability, and accessibility in industrial maintenance knowledge management. This approach addresses hallucinations and increases robustness of LLM applications in industrial AI settings.

The performance of machine learning models can be significantly impacted by variations in data quality. Typically, conventional model testing does not examine how robust the model would be in the face of potential data quality deterioration. In an industrial use case, however, data quality is a pertinent issue, as sensors are susceptible to a variety of technical and external issues that may result in poor data quality over time. In order to develop robust machine learning models, industrial data scientists must understand the sensitivity of their models against data quality issues, through the application of an appropriate and comprehensive testing solution. In this work, we propose a generic framework for systematically analyzing the impact of data quality issues on the performance of machine learning models by intentionally applying gradual perturbations to the original time series data. The evaluation is performed using a benchmark industrial process consisting of multivariate time series from sensors in a complex chemical process.

With the increasing application of artificial intelligence (AI) and machine learning (ML), the topic of technical debt management for machine learning systems is gaining more attention. Additionally, industrial systems including manufacturing or logistics processes are also supposed to benefit from AI and ML, which is reported in many publications related to ML application models. However, fewer studies on “how is technical debt managed in context of ML systems” are being published. This contribution fills this gap by reporting findings from 15 semi-structured and in-depth interviews conducted with industrial practitioners. Based on the interview results, suggestions for an initial technical debt management process and two document artifacts that facilitate the process are addressed.

A high degree of automation is an essential factor for smooth operation, to be economically viable and in many parts of the industry a de-facto standard. Nevertheless, manual interventions and operations are ever present, especially in those sectors that deal with large variations and uncertainties. For example, in waste incineration, the composition of waste varies significantly making a steady, efficient, and automated incineration very difficult. Similarly, the recovery from safety-/load-related trips is often not automated due to the large number of potential causes. These manual activities are recorded as part of the regular audit trail and stored in historians or databases. Yet, they are rarely analyzed which makes them a blind spot in the ongoing activities for digitization and Industry 4.0. We provide an overview of state-of-the-art techniques to perform case and workflow mining, our experience in analyzing two industrial data sets and our pipelines established during that activity.

Data preparation for data mining in industrial applications is a key success factor which requires considerable repeated efforts. Although the required activities need to be repeated in very similar fashion across many projects, details of their implementation differ and require both application understanding and experience. As a result, data preparation is done by data mining experts with a strong domain background and a good understanding of the characteristics of the data to be analyzed. Experts with these profiles usually have an engineering background and no strong expertise in distributed programming or big data technology. Unfortunately, the amount of data can be so large that distributed algorithms are required to allow for inspection of results and iteration of preparation steps. This contribution introduces an interactive data preparation workflow for signal data from chemical plants enabling domain experts without background in distributed computing and extensive programming experience to leverage the power of big data technologies.