Sparse System Dynamics Modeling for High-Dimensional Decision-Making in Industrial Automation

The increasing complexity of industrial automation systems has introduced significant challenges in modeling and analyzing high-dimensional decision-making environments. Traditional system dynamics (SD) models often struggle with scalability and computational efficiency when faced with numerous interdependent variables and feedback loops. In this study, we propose a Sparse System Dynamics Modeling (SSDM) approach that leverages sparsity-aware techniques to identify and retain only the most influential causal relationships within complex industrial systems. The SSDM framework introduces a structure reduction mechanism based on variable correlation thresholds and influence-weight pruning, enabling the construction of lightweight yet expressive models. By applying this method to a case study involving automated production line optimization, we demonstrate that SSDM maintains the predictive integrity of full-scale SD models while reducing computational overhead by up to 60%. The model also facilitates faster scenario simulations and more interpretable decision pathways, making it suitable for real-time industrial planning and control. Our results highlight the potential of sparse modeling in addressing the curse of dimensionality in industrial environments, providing a scalable and interpretable alternative for decision-makers in smart manufacturing and Industry 4.0 applications.