CausalTrace: A Neurosymbolic Causal Analysis Agent for Smart Manufacturing

TL;DR

CausalTrace is a neurosymbolic causal analysis module embedded in SmartPilot that integrates data-driven causal discovery with ontologies and knowledge graphs to provide real-time root cause analysis, counterfactual reasoning, and explainable decision support for smart manufacturing. Built on a modular AI framework (C3AN), it combines LiNGAM and DiffAN for causal graph learning, bootstrap-based edge stability, and a total effect computation, while grounding in a manufacturing knowledge graph and process ontology to offer semantic explanations. The approach is evaluated on an academic rocket assembly testbed, achieving high agreement with domain experts (ROUGE-1: 0.91; RCA metrics MAP@3: 94%, PR@2: 97%, MRR: 0.92, Jaccard: 0.92) and an overall C3AN score of 4.59/5, demonstrating strong robustness, intelligence, and trustworthiness. Deployment is staged from virtual prototyping to live deployment, with plans to transition to decentralized edge-enabled architectures and public hosting to scale across manufacturing environments, enabling reliable, interpretable, and actionable causal insights for operators and SMEs.

Abstract

Modern manufacturing environments demand not only accurate predictions but also interpretable insights to process anomalies, root causes, and potential interventions. Existing AI systems often function as isolated black boxes, lacking the seamless integration of prediction, explanation, and causal reasoning required for a unified decision-support solution. This fragmentation limits their trustworthiness and practical utility in high-stakes industrial environments. In this work, we present CausalTrace, a neurosymbolic causal analysis module integrated into the SmartPilot industrial CoPilot. CausalTrace performs data-driven causal analysis enriched by industrial ontologies and knowledge graphs, including advanced functions such as causal discovery, counterfactual reasoning, and root cause analysis (RCA). It supports real-time operator interaction and is designed to complement existing agents by offering transparent, explainable decision support. We conducted a comprehensive evaluation of CausalTrace using multiple causal assessment methods and the C3AN framework (i.e. Custom, Compact, Composite AI with Neurosymbolic Integration), which spans principles of robustness, intelligence, and trustworthiness. In an academic rocket assembly testbed, CausalTrace achieved substantial agreement with domain experts (ROUGE-1: 0.91 in ontology QA) and strong RCA performance (MAP@3: 94%, PR@2: 97%, MRR: 0.92, Jaccard: 0.92). It also attained 4.59/5 in the C3AN evaluation, demonstrating precision and reliability for live deployment.

Figures (4)

• Figure 1: Multi-agent architecture of SmartPilot
• Figure 2: (a) Architecture of CausalTrace agent: PredictX provides predicted anomalies, ForeSight offers forecasted demand, both feeding into InfoGuide. Data flows through Data Loader and Feature Selector before entering the Causal Discovery Engine. Discovered causal graphs support Root Cause Analysis. Causal graphs and RCA results are validated in the Causal Evaluator. InfoGuide responses are enhanced via neurosymbolic integration, combining a manufacturing knowledge graph, process ontology, and causal graph prompts. Final results are stored in the Memory Module. (b) Interactive User Interface: Causal graph visualization and allows causal graph editing. Enhances graphs with ontology-derived descriptions, types, and units. Tooltip metadata is enriched using a smart manufacturing knowledge graph.
• Figure 3: Illustration of a structured causal graph with categorized query types: structure, reasoning, and RCA.
• Figure 4: End-to-end deployment pipeline of SmartPilot with virtual and real-world integration