Synthesis of Free-Labeled Petri Nets With Inhibitor Arcs: An Exact and Efficient Approach
IEEE Transactions on Automation Science and Engineering · 2026 · 23 · 10730-10749
DOI: 10.1109/tase.2026.3695285 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Automation Science and Engineering, 2026. Verfasst von Wei Cheng und 2 weiteren Autor:innen. Die Arbeit nennt 55 Literaturangaben. Gefördert von: European Union’s Horizon Europe Research and Innovation Program, China Scholarship Council.
Abstract (Originalsprache)
This paper investigates the synthesis of free-labeled Petri nets with inhibitor arcs from prefix-closed language samples. Inhibitor arcs are crucial for capturing priority logic, such as blocking behavior, in discrete-event systems, and are therefore essential for faithful Petri net models. We formulate the synthesis problem as an integer linear program that seeks a minimum Petri net that reproduces the language samples. To improve scalability, we exploit the structure of the ILP and develop a decomposition strategy that enables exact solution through branch-and-price. The proposed approach is validated on three case studies: the classical producer–consumer system, a parallel production line, and a lab-scale LEGO manufacturing plant. The results demonstrate improved robustness and efficiency over a state-of-the-art algorithm on larger instances. This provides a practical foundation for supervisor control and digital twin development in discrete-event systems. Note to Practitioners—This paper is motivated by the practical challenge of building trustworthy digital twins for complex manufacturing and automation systems. A digital twin is an active virtual replica of a physical plant, used to simulate system behaviors, monitor operations, and optimize performance. However, to build digital twins, engineers currently face a dilemma. Manually constructing traditional simulation models is highly time-consuming and error-prone, whereas data-driven neural networks often yield opaque models that are difficult to interpret and trust. To bridge this gap, we propose an automated tool that directly converts routine event logs into transparent, flowchart-like models (Petri nets). These models explicitly capture how processes flow, share limited resources, and prioritize tasks, including complex interactions like system blocking. Consequently, practitioners can rapidly turn historical data into a working, understandable digital twin. This enables engineers to visually pinpoint the root causes of bottlenecks, analyze deadlocks, and safely evaluate new dispatching policies before physical deployment. One current limitation is its strict data requirement: the tool assumes the log captures every possible sequence of events up to a certain length. This completeness is challenging in practice due to rare behaviors or unrecorded events. Ongoing work focuses on constructing suitable inputs from real-world logs and characterizing exactly how much data is needed to reliably generate these digital twins.
Förderung
- European Union’s Horizon Europe Research and Innovation Program
- China Scholarship Council
Zitieren
APA
Cheng, W., Zhu, L., & Matta, A. (2026). Synthesis of Free-Labeled Petri Nets With Inhibitor Arcs: An Exact and Efficient Approach. IEEE Transactions on Automation Science and Engineering, 23, 10730-10749. https://doi.org/10.1109/tase.2026.3695285
Vancouver
Cheng W, Zhu L, Matta A. Synthesis of Free-Labeled Petri Nets With Inhibitor Arcs: An Exact and Efficient Approach. IEEE Transactions on Automation Science and Engineering. 2026;23:10730-10749. doi:10.1109/tase.2026.3695285
BibTeX
@article{Cheng2026,
author = {Cheng, Wei and Zhu, Lulai and Matta, Andrea},
title = {Synthesis of Free-Labeled Petri Nets With Inhibitor Arcs: An Exact and Efficient Approach},
journal = {IEEE Transactions on Automation Science and Engineering},
year = {2026},
volume = {23},
pages = {10730--10749},
doi = {10.1109/tase.2026.3695285}
}
EUVIMED