期刊
IEEE TRANSACTIONS ON POWER ELECTRONICS
卷 35, 期 5, 页码 4689-4702出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TPEL.2019.2945931
关键词
Electromagnetic interference (EMI); graphics processor (GPU); high-voltage direct current (HVdc); insulated-gate bipolar transistor (IGBT); massively parallel processing; modular multilevel converter (MMC); wideband modeling
资金
- Sino-Foreign Joint Program Scholarship of North China Electric Power University
The analysis of electromagnetic interference (EMI) noise of power electronic circuits involves both the transient characteristics of power semiconductor devices and the wideband stray parameters of passive equipment. Modular multilevel converters (MMCs) used in high-voltage direct current (HVdc) transmission systems contain thousands of submodules (SMs), which makes it considerably challenging to perform device-level simulation on the traditional commercial software. This article presents an accurate and fast method for wideband modeling and simulation of MMC-HVdc system for the assessment of conducted EMI during the design stage. Physical characteristics of the semiconductor devices, parasitic parameters of the insulated-gate bipolar transistor (IGBT) packages, and stray capacitances of the SMs are all taken into consideration, and massively parallel transient simulation of the wideband MMC model is carried out on the graphics processor (GPU). The accuracy and efficiency of the GPU-based parallel algorithm are validated by the comparison with the experimental measurement of an 11-level full-bridge MMC prototype. Furthermore, the stray capacitance network of the valve tower in HVdc project is extracted, and a matrix partition method based on the shielding plate configuration is utilized to conduct the computation in a fully parallelized manner. The developed GPU program is used to run the large-scale case of a 201-level two-terminal MMC-HVdc system, and the primarily affected frequency range by various factors is analyzed. Execution time test is conducted for different level topology, and it is demonstrated that the GPU can achieve a remarkable speedup over multicore CPUs, especially when the system scale is more substantial.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据