Root cause analysis is the process of investigating the cause of a failure and providing measures to prevent future failures. It is an active area of research due to the complexities in manufacturing production lines and the vast amount of data that requires manual inspection. We present a combined approach of causal neuro-symbolic AI for root cause analysis to identify failures in smart manufacturing production lines. We have used data from an industry-grade rocket assembly line and a simulation package to demonstrate the effectiveness and relevance of our approach.
Smart Manufacturing, or Industry 4.0, represents a wave of innovations and new technologies that are shaping the future of manufacturing with the goal of achieving more efficient production lines. Advanced sensors and Internet of Things devices collect data about various aspects of the production process, marking the onset of data-driven manufacturing. The collected data is utilized for predictive maintenance, process optimization, and root cause analysis (RCA). Traditional RCA is a costly and time-consuming process that involves manual inspection by domain experts, potentially leading to production delays. The automation of RCA is an active research area in smart manufacturing, aimed at minimizing downtime and ensuring costeffective production lines. Current work on RCA leverages ontologies, knowledge graphs (KG), and neuro-symbolic methods to store expert knowledge, model production line dependencies, and conduct reasoning to identify the time, location, and cause of failures [1]. In order to better understand and explain the root cause of failures and provide preventive measures, it is important to comprehend and model the causal associations in the data. However, parametric causal AI approaches for RCA do not consider prior knowledge about the relationships and parameters in the data [2]. Furthermore, traditional causal association approaches, which predict causal relations between variables, do not scale well for large volumes of data [3]. Effective