Scholarlink

<p>Transformer insulation oil deterioration critically impacts transformer reliability and lifespan, often leading to operational disruptions and financial losses. Accurate forecasting of oil ageing is therefore essential for proactive maintenance. This study presents a novel approach that combines a Deep Convolutional Neural Network (DCNN) with a Variance-Controlled Hunger Game Search (VC-HGS) algorithm to optimise the DCNN’s dense layers, reducing prediction errors. Experimental results show that the VC-HGS-DCNN model outperforms traditional methods, achieving 7% higher accuracy with lower MAE and RMSE. These improvements enable more precise maintenance scheduling, minimise unnecessary oil replacements, prevent unplanned transformer failures, and ultimately extend transformer service life while reducing operational costs.</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

Ageing assessment of power transformation insulation oil by using an improved hunger game search algorithm

  • 期刊论文

  • 发表日期：

    2026年5月11日

查看全文
  • Harkamal Deep Singh,
  • Jashandeep Singh

摘要

Transformer insulation oil deterioration critically impacts transformer reliability and lifespan, often leading to operational disruptions and financial losses. Accurate forecasting of oil ageing is therefore essential for proactive maintenance. This study presents a novel approach that combines a Deep Convolutional Neural Network (DCNN) with a Variance-Controlled Hunger Game Search (VC-HGS) algorithm to optimise the DCNN’s dense layers, reducing prediction errors. Experimental results show that the VC-HGS-DCNN model outperforms traditional methods, achieving 7% higher accuracy with lower MAE and RMSE. These improvements enable more precise maintenance scheduling, minimise unnecessary oil replacements, prevent unplanned transformer failures, and ultimately extend transformer service life while reducing operational costs.