Scholarlink

<p>This study presents a new and well-validated method using a particle filter for detecting Partial Discharge (PD) in power transformers. Recognizing the limitations of traditional methods in noisy settings, our creative approach effectively merges multi-modal acoustic and Ultra - High Frequency (UHF) sensor data, enhanced by advanced signal processing and improved feature extraction techniques. At first, our basic algorithm showed higher error rates than earlier research, highlighting the necessity for major enhancements. We tackled this issue by creating an optimized particle filter algorithm, inspired by the precise Monte Carlo localization methods used in satellite navigation and deep-space tracking. This improved algorithm, supported by comprehensive MATLAB simulations and detailed quantitative analysis, significantly lowered localization errors, achieving outstanding performance gains with sub-millimeter accuracy in ideal conditions and consistently maintaining sub-6&#xa0;mm precision across various testing scenarios.</p><p>A key contribution is the combined use of advanced Time Difference of Arrival (TDOA) techniques with our dependable Particle Filter framework, which is improved by adaptive mechanisms tailored for transformer environments. This strong combination greatly boosts fault detection and noise immunity, while also offering outstanding resistance to environmental changes that are often faced in transformer diagnostics. When compared to traditional methods, our integrated, sensor-driven framework consistently shows better performance in terms of detection precision and localization accuracy.</p><p>This improved ability is crucial for enabling dependable predictive maintenance plans, avoiding costly breakdowns, and extending the lifespan of high-voltage equipment. The proven practicality and success of this combined method emphasize its potential to revolutionize transformer condition monitoring. Future work will focus on enhancing computational efficiency for real-time uses and expanding the framework’s applicability to a broader range of high-voltage devices, thus solidifying its importance as a flexible and vital tool in modern electrical asset management.</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

Particle filter- based localization approach for PD in power transformers

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月30日

查看全文
  • Niravkumar J. Patel,
  • Jalpa Thakkar,
  • Kalpesh Dudani

摘要

This study presents a new and well-validated method using a particle filter for detecting Partial Discharge (PD) in power transformers. Recognizing the limitations of traditional methods in noisy settings, our creative approach effectively merges multi-modal acoustic and Ultra - High Frequency (UHF) sensor data, enhanced by advanced signal processing and improved feature extraction techniques. At first, our basic algorithm showed higher error rates than earlier research, highlighting the necessity for major enhancements. We tackled this issue by creating an optimized particle filter algorithm, inspired by the precise Monte Carlo localization methods used in satellite navigation and deep-space tracking. This improved algorithm, supported by comprehensive MATLAB simulations and detailed quantitative analysis, significantly lowered localization errors, achieving outstanding performance gains with sub-millimeter accuracy in ideal conditions and consistently maintaining sub-6 mm precision across various testing scenarios.

A key contribution is the combined use of advanced Time Difference of Arrival (TDOA) techniques with our dependable Particle Filter framework, which is improved by adaptive mechanisms tailored for transformer environments. This strong combination greatly boosts fault detection and noise immunity, while also offering outstanding resistance to environmental changes that are often faced in transformer diagnostics. When compared to traditional methods, our integrated, sensor-driven framework consistently shows better performance in terms of detection precision and localization accuracy.

This improved ability is crucial for enabling dependable predictive maintenance plans, avoiding costly breakdowns, and extending the lifespan of high-voltage equipment. The proven practicality and success of this combined method emphasize its potential to revolutionize transformer condition monitoring. Future work will focus on enhancing computational efficiency for real-time uses and expanding the framework’s applicability to a broader range of high-voltage devices, thus solidifying its importance as a flexible and vital tool in modern electrical asset management.