Diagnosis on Looseness of Transformer Winding Based on Improved MPE and K-medoids Algorithm

MA Hongzhong; XUE Jiantong; NI Yiming; WAN Keli; ZE Hengpeng

doi:10.13296/j.1001-1609.hva.2025.09.010

您当前的位置：

首页 >

文章列表页 >

Diagnosis on Looseness of Transformer Winding Based on Improved MPE and K-medoids Algorithm

Research & Analysis | 更新时间：2025-09-05

- Diagnosis on Looseness of Transformer Winding Based on Improved MPE and K-medoids Algorithm
- High Voltage Apparatus Vol. 61, Issue 9, Pages: 73-80(2025)
- 作者机构：
  
  河海大学电气与动力工程学院，南京　211100
- 作者简介：
- 基金信息：
  
  National Natural Science Foundation of China(51577050)
- DOI：10.13296/j.1001-1609.hva.2025.09.010
  CLC：
- Received：20 February 2025，
  
  Revised：17 May 2025，
  
  Published：16 September 2025
- 稿件说明：
移动端阅览
MA Hongzhong, XUE Jiantong, NI Yiming, et al. Diagnosis on Looseness of Transformer Winding Based on Improved MPE and K-medoids Algorithm[J]. High voltage apparatus, 2025, 61(9): 73-80.
DOI：

MA Hongzhong, XUE Jiantong, NI Yiming, et al. Diagnosis on Looseness of Transformer Winding Based on Improved MPE and K-medoids Algorithm[J]. High voltage apparatus, 2025, 61(9): 73-80. DOI： 10.13296/j.1001-1609.hva.2025.09.010.

摘要

为了更加有效地对变压器绕组松动故障进行诊断，针对变压器有载运行时的振动信号，提出了一种基于改进多尺度排列熵（MPE）和K-medoids的变压器绕组松动故障诊断方法。首先采用粒子群优化（PSO）的MPE算法对绕组不同状态下的变压器振动信号进行特征提取，以减少MPE算法中参数设置对故障类型识别精度的影响，然后通过K-medoids聚类算法诊断变压器绕组松动故障，以完成故障的分类识别。对某10 kV变压器的绕组松动故障模拟实验结果表明，绕组不同状态下变压器振动信号的MPE值经PSO参数优化后存在明显差异，诊断效果优于传统经验设置参数的MPE算法，且稳定性得到提高。

Abstract

For effectively diagnosing the looseness of transformer winding

a looseness diagnosis method of transformer winding based on the improved multiscale permutation entropy (MPE) and K-medoids algorithm is proposed for the vibration signal of transformer during the on-load operation. First

the MPE algorithm based on particle swarm optimization (PSO) is used to extract the features of the transformer winding signals under different state of winding so to reduce the influence of parameter settings in the MPE algorithm on the accuracy of fault type ientification. Then

the K-medoids clustering algorithm is used to diagnose the looseness of transformer winding and complete the fault classification and identification. The simulation experiment results of the looseness of winding of a 10 kV transformer show that the MPE values of the transformer vibration signals under different winding states are significantly different after the optimization of the PSO parameters. The diagnosis effect is superior to that of the traditional MPE algorithm with empirical parameters

and the stability has been improved.

关键词

Keywords

references

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Application of Adaptive Hierarchical Multi-class SVM to Transformer Fault Diagnosis

Experimental Research on Mechanical Properties of Copper Wires in Transformer Winding Under Thermal Effects

Prediction Model for Dissolved Gas in Transformer Oil Based on Policy Gradient Optimization

Research on Hybrid Finite Element Method Based on Triangle Mesh and Its Application in Fluid-heat Coupled Calculation of Transformer

Vibration Sound Power Measurement Method of UHV Transformer Based on Location of Main Vibration Point

Related Author

YANG Hong1

GU Shi-fu1

TAO Jia-gui1

GOU Jian2

XU Shurong

LIU Jun

LI Zhongxiang

CHEN Peilong

Related Institution

School of Electrical and Information Engineering,Xihua University

Sichucm Mianyang Province Electric Power Bureau

Guizhou Power Grid Co., Ltd. Electric Power Research Institute

College of Electrical and Information Engineering, Hunan University

TBEA Hengyang Transformer Co., Ltd.

⁰