Modelling and Validation of Power Electronics Converter Systems using Coloured Petri Nets

TL;DR

This paper develops a formal, timed hierarchical Coloured Petri Nets model for power converter systems, enabling integrated validation and comparison with Simulink and real hardware, improving system design reliability.

Contribution

It introduces a formal CPN-based modeling approach for power electronics systems, addressing Simulink's limitations in integrated system validation.

Findings

CPN model aligns with Simulink simulation results.

Model accurately reflects real hardware measurements.

Closed-loop and open-loop configurations show consistent steady-state behavior.

Abstract

We apply Coloured Petri Nets (CPNs) and the CPN Tools to develop a formal model of an embedded system consisting of a power converter and an associated controller. Matlab/Simulink is the de-facto tool for embedded control and system design, but it relies on informal semantics and has limited support for transparent and integrated specification and validation of both the power converter electronics, controller (hardware), and the control logic (software). The contribution of this paper is to develop a timed hierarchical CPN model that mitigates the shortcomings of Simulink by relying on a Petri net formalisation. We demonstrate the application of our approach by developing a fully integrated model of a buck power converter with controller in CPN Tools. Furthermore, we perform time-domain simulation to verify the capability of the controller to serve the control objectives. To validate the developed CPN model, we compare the simulation results obtained in an open-loop configuration with a corresponding implementation in Simulink. The experimental results show correspondence between the CPN model and the Simulink model. As our CPN model reflects the fully integrated system, we are able to compare CPN simulation results to measurements obtained with a corresponding implementation in real hardware/software and compare closed-loop with open-loop configuration. The results show alignment for the steady state while further refinement of the control algorithm and validation is required.