Scholarlink

DC capacitors (DCCs) are the key components in power electronic transformers (PETs). It is necessary to monitor the healthy condition of DCCs. Changes in the healthy condition of DCCs result in varied values of capacitance (C) and equivalent series resistance (ESR). The existing data-driven parameter identification methods for DCCs fail to work with variable working conditions. To address the above problem, an adversarial learning network model is proposed. There are four parts in this method: Source-domain Feature Extractor (SFE), Target-domain Feature Extractor (TFE), discriminator and regressor. Firstly, the features of source-domain and target-domain data are extracted by the SFE and TFE respectively, where the feature of source-domain employed as the prior distribution of the target-domain data. Secondly, discriminator is employed to minimize the distribution divergence between the source-domain and target-domain data. Thirdly, regressor is trained by feature and label of source-domain data, then employed to estimate the C and ESR of target-domain data. Finally, the proposed method is verified by the simulation and experimental data of a three-phase AC-DC PET.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An Auto-Encoder Unsupervised Adversarial Learning Network for Parameter Identification of DC Capacitor in Power Electronic Transformer

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Bowen Zhou,
  • Xiaohui Li,
  • Liqun He,
  • Yong Yang,
  • Zhenfeng Wang,
  • Hong Cheng

摘要

DC capacitors (DCCs) are the key components in power electronic transformers (PETs). It is necessary to monitor the healthy condition of DCCs. Changes in the healthy condition of DCCs result in varied values of capacitance (C) and equivalent series resistance (ESR). The existing data-driven parameter identification methods for DCCs fail to work with variable working conditions. To address the above problem, an adversarial learning network model is proposed. There are four parts in this method: Source-domain Feature Extractor (SFE), Target-domain Feature Extractor (TFE), discriminator and regressor. Firstly, the features of source-domain and target-domain data are extracted by the SFE and TFE respectively, where the feature of source-domain employed as the prior distribution of the target-domain data. Secondly, discriminator is employed to minimize the distribution divergence between the source-domain and target-domain data. Thirdly, regressor is trained by feature and label of source-domain data, then employed to estimate the C and ESR of target-domain data. Finally, the proposed method is verified by the simulation and experimental data of a three-phase AC-DC PET.