Scholarlink

To address poor repeatability and limited accuracy in field measurements of loop resistance within GIS (Gas-Insulated Switchgear), this study conducts controlled experiments and discovers that using a earthing switch-gear as the voltage lead terminal introduces significant contact resistance deviations. These deviations do not align with predictions from the ideal four-wire (Kelvin) method, indicating non-ideal behavior in the actual testing system. A non-ideal physical model is then established, incorporating equivalent parasitic resistance and the instrument’s effective internal resistance. An in-situ calibration framework based on Bayesian Markov Chain Monte Carlo (MCMC) is proposed. This framework, combined with measurements of a set of standard resistors, simultaneously identifies unknown resistances under test. Simulations show that with a R_target of 44.5μΩ， the mean measurement error can be reduced from 4.96 μΩ to 0.12 μΩ, representing an 81.8% decrease in error.

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

Non-Ideal Modeling and Bayesian Correction of Systematic Errors in GIS Loop Resistance Field Testing

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Nan Zhang,
  • Miao Xu,
  • Ruixi Liu,
  • Fei Wang,
  • Tian Qi,
  • Hongpeng Zang,
  • Hongjiang Sun

摘要

To address poor repeatability and limited accuracy in field measurements of loop resistance within GIS (Gas-Insulated Switchgear), this study conducts controlled experiments and discovers that using a earthing switch-gear as the voltage lead terminal introduces significant contact resistance deviations. These deviations do not align with predictions from the ideal four-wire (Kelvin) method, indicating non-ideal behavior in the actual testing system. A non-ideal physical model is then established, incorporating equivalent parasitic resistance and the instrument’s effective internal resistance. An in-situ calibration framework based on Bayesian Markov Chain Monte Carlo (MCMC) is proposed. This framework, combined with measurements of a set of standard resistors, simultaneously identifies unknown resistances under test. Simulations show that with a R_target of 44.5μΩ， the mean measurement error can be reduced from 4.96 μΩ to 0.12 μΩ, representing an 81.8% decrease in error.