Solving the puzzle: Leveraging machine learning for effective root cause analysis

Root Cause Analysis (RCA) is an established and important methodological tool widely used in manufacturing to determine the causes of repeating machine breakdowns, defects or to improve overall efficiency. Traditionally, specialists may use quality management tools like 5-Why and Ishikawa diagrams or statistical methods to systematically identify and address the underlying causes. With increasing digitalization, more and more data are being collected providing a good basis for systematic data driven analysis. However, it is difficult and time-consuming to extract valuable insights from this data. Today’s established methods today are reaching their limits dealing with the sheer amount of data and the complex, non-linear interrelationships of modern manufacturing processes and process chains. To navigate these complexities, the integration of Explainable AI (XAI) and causal discovery techniques offers promising avenues for enhancing RCA. XAI provides transparency in machine learning models, enabling practitioners to gain impulses from inspecting how a model uses the data. Meanwhile, causal discovery focuses on distinguishing between mere statistical associations and genuine causal relationships, allowing for the identification of causal paths. This talk will provide an overview of these methodologies, discussing their application in root cause analysis, as well as the challenges and opportunities they present in improving decision-making processes.