Scholarlink

<p>The accurate prediction of the thermal characteristics of underground cables is needed for reliable operation of underground power cables, optimisation of ampacity, and insulation degradation protection. However, the thermal model of underground cables is a complicated task because of the uncertainty of the thermal resistivity of soil and ambient temperature changes, and the dynamic loading conditions. Traditional numerical methods like finite element methods (FEM) are very accurate but cannot be used for real-time monitoring or for dynamic thermal monitoring because they are very consuming. In this paper, an AI-based type-2 fuzzy logic system (IT2-FLS) for interval prediction of uncertainty for underground cable systems is presented. The suggested model explicitly models the variability of the environmental and operational parameters by using a Footprint of Uncertainty (FOU) and calibrates the membership functions by applying Particle Swarm Optimisation (PSO) technique based on the available data. The training and evaluation were performed with a comprehensive set of 500 Monte Carlo scenarios (50,000 observations) created by a transient thermal model consistent with IEC 60287. The IT2-FLS has RMSE = 0.132&#xa0;K, MAE = 0.104&#xa0;K and the maximum prediction error = 0.395&#xa0;K compared to its benchmark in FEM. The RMSE is decreased by 73.6% using ANN and 58.2% for type-1 ANFIS. The proposed technique is the lowest RMSE cable temperature predicted by any published work related to cable thermal surrogates. It is observed that the degradation in the RMSE during robustness analysis with input noise ranging from 5 to 20% is significantly lower, which points to the advantage of interval type-2 uncertainty representation.</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

AI-driven interval type-2 fuzzy logic for uncertainty-aware thermal field modelling of underground power cable systems

  • 期刊论文

  • 发表日期：

    2026年6月23日

查看全文
  • Arnika Jain,
  • W. Ancy Breen,
  • M. Sujatha,
  • B. Natarajan,
  • Challapalli Sujana,
  • Elangovan Muniyandy

摘要

The accurate prediction of the thermal characteristics of underground cables is needed for reliable operation of underground power cables, optimisation of ampacity, and insulation degradation protection. However, the thermal model of underground cables is a complicated task because of the uncertainty of the thermal resistivity of soil and ambient temperature changes, and the dynamic loading conditions. Traditional numerical methods like finite element methods (FEM) are very accurate but cannot be used for real-time monitoring or for dynamic thermal monitoring because they are very consuming. In this paper, an AI-based type-2 fuzzy logic system (IT2-FLS) for interval prediction of uncertainty for underground cable systems is presented. The suggested model explicitly models the variability of the environmental and operational parameters by using a Footprint of Uncertainty (FOU) and calibrates the membership functions by applying Particle Swarm Optimisation (PSO) technique based on the available data. The training and evaluation were performed with a comprehensive set of 500 Monte Carlo scenarios (50,000 observations) created by a transient thermal model consistent with IEC 60287. The IT2-FLS has RMSE = 0.132 K, MAE = 0.104 K and the maximum prediction error = 0.395 K compared to its benchmark in FEM. The RMSE is decreased by 73.6% using ANN and 58.2% for type-1 ANFIS. The proposed technique is the lowest RMSE cable temperature predicted by any published work related to cable thermal surrogates. It is observed that the degradation in the RMSE during robustness analysis with input noise ranging from 5 to 20% is significantly lower, which points to the advantage of interval type-2 uncertainty representation.