Hybrid Data-Physics-Driven Modeling Method for Real-Time Simulation of Cascaded Power Electronics Systems
Year: 2025
Type: article
Abstract: Existing real-time simulation methods for cascaded power electronic systems face significant challenges. Physics-driven approaches often suffer from limited applicability when critical system information is unavailable, as well as inefficiency due to the extensive manual effort required for model derivation. Conversely, data-driven methods are constrained by large data demands and prolonged training times. To address these limitations, this article proposes a hybrid data-physics-driven (HDPD) modeling method. The cascaded subcircuits, referred to as “modules,” are represented by discretized state-space equations with associated transition matrices. At the module level, a data-driven approach based on the least squares method is employed to automatically generate the transition matrices, utilizing data obtained from short-duration offline simulations. At the system level, a physics-driven approach is applied to directly compute the equivalent model of the cascaded system, leveraging the inherent characteristics of series and parallel connections. The proposed HDPD method is applied to the cascaded solid-state transformer using an established CPU-field-programmable gate array (FPGA) real-time simulation platform. Results demonstrate that the proposed approach effectively balances the strengths of data-driven and physics-based modeling, providing high efficiency and flexibility in model development, while maintaining strong parallelism and low computational complexity during simulation. Consequently, the method facilitates convenient model construction and enables large-scale simulations with improved scalability.
Authors: Chenxiang Gao, Pan Wu, Jin Xu, Keyou Wang, Yubo Song, Frede Blaabjerg
Institutions: Shanghai Jiao Tong University, Aalborg University
Language: English
Subfield: Control and Systems Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
Funder: State Grid Corporation of China
OpenAlex ID: W4413212140