Hierarchical Causal Abduction: A Foundation Framework for Explainable Model Predictive Control

2026-05-11

AI Summary

This study addresses the challenge of explaining decisions made by nonlinear Model Predictive Control (MPC) systems, whose complexity, stringent constraints, and opaque optimization processes hinder human interpretability and real-world deployment. To this end, the authors propose a Hierarchical Causal Abduction (HCA) framework that uniquely integrates domain reasoning via physical knowledge graphs, sensitivity analysis using KKT multipliers, and temporal causal discovery through PCMCI to generate faithful and interpretable justifications for MPC actions. The method produces high-fidelity explanations across diverse scenarios without requiring per-domain parameter tuning and generalizes to other predictive decision-making systems. Evaluated on greenhouse climate control, building HVAC, and chemical process applications, HCA achieves an explanation accuracy of 0.478—53% higher than LIME—and further improves to 0.88 after only 2–3 days of domain calibration, with ablation studies confirming the necessity of each component.

Abstract

Model Predictive Control (MPC) is widely used to operate safety-critical infrastructure by predicting future trajectories and optimizing control actions. However, nonlinear dynamics, hard safety constraints, and numerical optimization often render individual control moves opaque to human operators, undermining trust and hindering deployment. This paper presents Hierarchical Causal Abduction (HCA), which combines (i) physics-informed reasoning via domain knowledge graphs, (ii) optimization evidence from Karush--Kuhn--Tucker (KKT) multipliers, and (iii) temporal causal discovery via the PCMCI algorithm to generate faithful, human-interpretable explanations for control actions computed by nonlinear MPC. Across three diverse control applications (greenhouse climate, building HVAC, chemical process engineering) with expert validation, HCA improves explanation accuracy by 53\% over LIME (0.478 vs. 0.311) using a single set of cross-domain parameters without per-domain tuning; domain-specific KKT-threshold calibration over 2--3 days further increases accuracy to 0.88. Ablation studies confirm that each evidence source is essential, with 32--37\% accuracy degradation when any component is removed, and HCA's ranking-and-validation methodology generalizes beyond MPC to other prediction-based decision systems, including learning-based control and trajectory planning.

Problem

• Model Predictive Control
• Explainability
• Nonlinear Dynamics
• Safety Constraints
• Human Trust

Innovation

• Hierarchical Causal Abduction
• Explainable Model Predictive Control
• KKT multipliers
• Causal Discovery
• Knowledge Graphs