Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Supervised Learning based Method for Condition Monitoring of Overhead Line Insulators using Leakage Current Measurement

Mile Mitrovic, Dmitry Titov, Klim Volkhov, Irina Lukicheva, Andrey Kudryavzev, Petr Vorobev, Qi Li, Vladimir Terzija

TL;DR

The paper tackles aging-induced flashover risk in overhead line insulators by predicting the critical voltage $U_{50\%}$ from leakage current features and applied voltage. It introduces a two-stage XGBoost-based pipeline that first classifies insulator wet/dry conditions and then regresses $\%U_{50\%}$ and $\%\sigma_m$ using MRMR-selected LC features, with separate regression models for each condition. Experimental data from natural, cup-and-pin glass insulators collected in a high-voltage lab underpin the method, including 63 LC features derived from time- and frequency-domain analyses. Results show robust classification and regression performance, enabling actionable asset-management decisions and potential live deployment at 110 kV, with future work extending to universal models across insulator types and voltage levels.

Abstract

As a new practical and economical solution to the aging problem of overhead line (OHL) assets, the technical policies of most power grid companies in the world experienced a gradual transition from scheduled preventive maintenance to a risk-based approach in asset management. Even though the accumulation of contamination is predictable within a certain degree, there are currently no effective ways to identify the risk of the insulator flashover in order to plan its replacement. This paper presents a novel machine learning (ML) based method for estimating the flashover probability of the cup-and-pin glass insulator string. The proposed method is based on the Extreme Gradient Boosting (XGBoost) supervised ML model, in which the leakage current (LC) features and applied voltage are used as the inputs. The established model can estimate the critical flashover voltage (U50%) for various designs of OHL insulators with different voltage levels. The proposed method is also able to accurately determine the condition of the insulator strings and instruct asset management engineers to take appropriate actions.

Supervised Learning based Method for Condition Monitoring of Overhead Line Insulators using Leakage Current Measurement

TL;DR

The paper tackles aging-induced flashover risk in overhead line insulators by predicting the critical voltage from leakage current features and applied voltage. It introduces a two-stage XGBoost-based pipeline that first classifies insulator wet/dry conditions and then regresses and using MRMR-selected LC features, with separate regression models for each condition. Experimental data from natural, cup-and-pin glass insulators collected in a high-voltage lab underpin the method, including 63 LC features derived from time- and frequency-domain analyses. Results show robust classification and regression performance, enabling actionable asset-management decisions and potential live deployment at 110 kV, with future work extending to universal models across insulator types and voltage levels.

Abstract

As a new practical and economical solution to the aging problem of overhead line (OHL) assets, the technical policies of most power grid companies in the world experienced a gradual transition from scheduled preventive maintenance to a risk-based approach in asset management. Even though the accumulation of contamination is predictable within a certain degree, there are currently no effective ways to identify the risk of the insulator flashover in order to plan its replacement. This paper presents a novel machine learning (ML) based method for estimating the flashover probability of the cup-and-pin glass insulator string. The proposed method is based on the Extreme Gradient Boosting (XGBoost) supervised ML model, in which the leakage current (LC) features and applied voltage are used as the inputs. The established model can estimate the critical flashover voltage (U50%) for various designs of OHL insulators with different voltage levels. The proposed method is also able to accurately determine the condition of the insulator strings and instruct asset management engineers to take appropriate actions.
Paper Structure (19 sections, 18 equations, 9 figures, 5 tables)

This paper contains 19 sections, 18 equations, 9 figures, 5 tables.

Table of Contents

  1. Introduction
  2. ML-based Insulators' Condition Monitoring Method
  3. Laboratory setup and data collection
  4. Laboratory Facility
  5. Test Samples
  6. Test procedures
  7. Preparation of the test sample
  8. $U_{50\%}$ voltage revealing
  9. Leakage current measurements
  10. Signal feature extraction
  11. Machine Learning for $U_{50\%}$ voltage Estimation
  12. XGBoost model
  13. Maximum Relevance Minimum Redundancy
  14. Evaluation metrics
  15. Results
  16. ...and 4 more sections

Figures (9)

  • Figure 1: Conditions for the classification of insulator string states; $U_{xx\%}$ is the estimated voltage where the probability of flashover of the insulator is $xx\%$.
  • Figure 2: Schematic diagram of assigning a type of insulator string state based on LC measurements.
  • Figure 3: Single-line circuit representing the experimental setup. PD is the protective diodes, OS is the Wi-Fi oscilloscope, R is the non-inductive resistor, T1 is the regulating transformer, T2 is the power transformer, VR is the voltage regulation block, VD is the voltage divider, WP is the operator's workplace, CC is the control console, PC is the computer with Wi-Fi.
  • Figure 4: left - high voltage installation UIV-500 (with voltage level up to 500 kV, load up to 60 kVA, and sustained short circuit current 0.3 A); right - insulator string under testing.
  • Figure 5: Insulator strings without (1.58 $\mu S$) and with natural contamination (4.86 $\mu S$ and 18.33 $\mu S$). Surface conductance was measured under wet conditions by high voltage insulation tester Sew 2804 IN.
  • ...and 4 more figures