Treffer: Explainable Fault Diagnosis: Rule-Based Approach with Improved Discrete Gray Wolf Optimization Algorithm.

The increasing complexity and automation of industrial processes, such as those found in chemical manufacturing, have made fault detection and diagnosis more critical than ever. Faults in such systems can lead to severe safety, environmental, and economic consequences. To address these challenges, this paper proposes a novel, explainable fault diagnosis framework that integrates a rule-based classification approach intelligently generated with an improved Discrete Gray Wolf Optimization (IDGWO) algorithm. The goal is to develop a model that not only achieves high accuracy in fault detection, but also provides transparency and interpretability in its decision-making process. The proposed method is applied to the Tennessee Eastman Process (TEP), a widely used benchmark in fault diagnosis research. The IDGWO algorithm is used to optimize the generation of classification rules, allowing for the extraction of a concise yet powerful rule set. Experimental evaluation using a real TEP dataset demonstrates the effectiveness of the approach, achieving an accuracy of 96.58% with only 9 concise rules, each containing approximately two terms. These rules offer clear insights into the underlying process behavior, enabling domain experts to better understand and trust the system’s decisions. Compared to the traditional techniques in both accuracy and interpretability, the proposed model offers superior performance while maintaining interpretability, thereby supporting safer and more trustworthy industrial systems. [ABSTRACT FROM AUTHOR]

Copyright of International Journal of Information Technology & Decision Making is the property of World Scientific Publishing Company and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)