Abstract
Reduction methods are widely used to simplify complex models, particularly for Petri nets, which model discrete event systems. Although effective in limiting combinatorial explosion, these methods have potentially critical flaws, especially by reducing the state space in a way that may obscure behaviours or states essential to comprehensive analysis. This paper proposes an approach to address these shortcomings by integrating reachability methods into the reduction process. By leveraging the ability of reachability methods to ensure the attainability of critical states while maintaining efficient state space reduction, this solution enhances the accuracy of complex system analysis while optimising computational resources. Two practical case studies of manufacturing system and task management system illustrate this approach and demonstrate its potential to improve the rigor of large-scale model analyses.