Dissolved Gas Analysis Method Fusion Based on Machine Learning for Fault Diagnosis and Condition Assessment of Power Transformers 核心 · 已核验

mxds3hvz2ebjqa49

This directory contains the Python codes used to implement the proposed methodology for power transformer fault diagnosis based on Dissolved Gas Analysis (DGA), Dempster–Shafer evidence fusion, and Random Forest classification.

Python Codes:

RF_con_subtipos_FIN.py: Main code of the methodology. This script executes the final Random Forest-based diagnostic process and integrates the diagnostic refinement using subtypes. It represents the main file to be executed for reproducing the proposed methodology.

D_S_para_RF.py: Python code used to generate the consensus pseudo-labels through Dempster–Shafer evidence theory. The outputs generated by this script are used as input labels for the Random Forest training stage.

Doernenburg_RF.py: Python code containing the implementation of the Doernenburg DGA interpretation method used as one of the diagnostic evidence sources.

Rogers_RF.py: Python code containing the implementation of the Rogers DGA interpretation method used as one of the diagnostic evidence sources.

Gas_Key_RF.py: Python code containing the implementation of the Gas Key DGA interpretation method used as one of the diagnostic evidence sources.

Pentagono_1_2_RF.py: Python code containing the implementation of the Duval Pentagon 1 and Duval Pentagon 2 methods. These methods are used for DGA-based diagnosis and diagnostic refinement.

Triangulo_1_4_5_RF.py: Python code containing the implementation of the Duval Triangle 1, Duval Triangle 4, and Duval Triangle 5 methods. These methods are used for conventional DGA diagnosis and physical diagnostic refinement through subtypes.

These codes allow the main stages of the research to be reproduced, including the application of conventional DGA methods, the generation of pseudo-labels using Dempster–Shafer, classification through Random Forest, and diagnostic refinement with Duval subtypes. In this way, a methodology based on Random Forest is proposed for fault analysis and diagnosis in power transformers using DGA.

本卡描述如含来自官方页的原文片段,其版权归原作者,不在本站 CC-BY 4.0 许可范围(见关于与许可)。

落地页

https://data.mendeley.com/datasets/phs5wyfbph

许可证

CC-BY-4.0 (置信:verified_official)

发布年份

2026

发布方

Mendeley Data

分发点