Scholarlink

<p>The moisture content in transformer oil is a critical indicator for maintaining power system reliability. In this research, a novel model combining broadband ultrasonic detection technology coupled with an Improved Whale Optimization (IWOA) and an Elman neural network is proposed for predicting the moisture content in transformer oil. Initially, the study analyzed 240 transformer oil samples, measuring their moisture content using a moisture content meter. A custom-designed broadband ultrasonic detection platform revealed compelling correlations between ultrasonic signatures and moisture levels across samples. Subsequently, an optimized Elman neural network model was developed to predict the moisture content in transformer oil. The input features included ultrasonic amplitude, phase, and attenuation coefficients at different frequencies, while the output represented actual moisture content. Through rigorous validation, the findings indicate that the IWOA-Elman model achieves a prediction accuracy of up to 96%. The prediction accuracy meets the national standard GB/T 7600, and the integration of broadband non-destructive ultrasonic sensing technology with the IWOA-Elman algorithm establishes a new framework for online predicting of trace moisture content in transformer oil.</p>

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

Research on Moisture Detection in Transformer Oil Using an Improved Neural Network Model and Broadband Ultrasonic Technology

  • 期刊论文

  • 发表日期：

    2025年6月16日

查看全文
  • Zhao-Yu Qin,
  • Sa-Sa Kong,
  • Guo-Zhi Zhang,
  • Si-Da Xu,
  • Lei Yang,
  • Jia-Bin Huang

摘要

The moisture content in transformer oil is a critical indicator for maintaining power system reliability. In this research, a novel model combining broadband ultrasonic detection technology coupled with an Improved Whale Optimization (IWOA) and an Elman neural network is proposed for predicting the moisture content in transformer oil. Initially, the study analyzed 240 transformer oil samples, measuring their moisture content using a moisture content meter. A custom-designed broadband ultrasonic detection platform revealed compelling correlations between ultrasonic signatures and moisture levels across samples. Subsequently, an optimized Elman neural network model was developed to predict the moisture content in transformer oil. The input features included ultrasonic amplitude, phase, and attenuation coefficients at different frequencies, while the output represented actual moisture content. Through rigorous validation, the findings indicate that the IWOA-Elman model achieves a prediction accuracy of up to 96%. The prediction accuracy meets the national standard GB/T 7600, and the integration of broadband non-destructive ultrasonic sensing technology with the IWOA-Elman algorithm establishes a new framework for online predicting of trace moisture content in transformer oil.