Multi-objective Optimization of Supervisor Simplification in AMS Based on Petri Nets and NSGA-II

• Published in: International Conference on Automation, Control and Robotics Engineering (Online) pp. 241 - 246
• Main Authors: Guo, Bingtao, Chen, Lei, Chang, Xingyuan, Chen, Chen, Li, Xiangning, Hu, Hesuan
• Format: Conference Proceeding
• Language: English
• Published: IEEE 16.07.2025
• Subjects: Automated manufacturing systems; Computational modeling; Genetic algorithms; Genetics; Manufacturing systems; NGSA-II; Optimization; Petri nets; Robots; Robustness; Sorting; Supervisor simplification; Supervisory control
• ISSN: 2997-6278
• DOI: 10.1109/CACRE66141.2025.11119579

Automated manufacturing systems (AMSs) are often modeled by Petri nets. Supervisors, in terms of a set of inequalitie, are common in supervisory control theories. Supervisor simplification aims to remove the redundant ones. However, in most cases, the number of redundant inequalities is very limited in AMSs with multiple resource usages in type and quantity. One possible way to further simplify the supervisor is constructing a new set of inequalities by adjusting the coefficients that enforce restrictive control on the siphons, with the sacrifice of permissiveness. This leads to the optimal coefficient adjustment problem, which aims to minimize the scale of the simplified supervisor while maximizing the number of reachable markings. It is a typical multi-objective combinatorial optimization problem and can be solved by the non dominated sorting genetic algorithm II (NSGA-II). NSGA-II provides a set of Pareto-optimal solutions for supervisor simplification. Compared with our previous method, NSGA-II has a better performance in finding the Pareto-optimal solutions and improves computational efficiency. An example is provided for illustration and comparison.