Detection of Power Line Insulators in Digital Images Based on the Transformed Colour Intensity Profiles.

Area of Science:

• Electrical Engineering
• Computer Science
• Image Processing

Background:

• Power line insulator inspection is crucial for electricity infrastructure maintenance.
• Existing detection methods have limitations.
• Damages like fractures and burns necessitate regular condition assessments.

Purpose of the Study:

• To propose and evaluate a novel method for detecting power line insulators in digital images.
• To utilize signal analysis and machine learning for automated insulator detection.
• To enable in-depth assessment of detected insulators.

Main Methods:

• Image acquisition using Unmanned Aerial Vehicle (UAV) for data collection.
• Color intensity profile classification based on identified insulator points.
• Signal transformation using Periodogram or Welch methods.
• Classification using Decision Tree, Random Forest, or XGBoost algorithms.

Main Results:

• The proposed method demonstrated high efficiency, achieving an F1 score of 0.99 in optimal scenarios.
• Successful detection of insulators against diverse backgrounds (sky, trees, power lines).
• Promising classification results indicating practical applicability.

Conclusions:

• The developed method offers an effective solution for automated power line insulator detection.
• The high accuracy suggests a viable tool for enhancing electricity infrastructure maintenance.
• Further research can explore expanded applications and refinements of the technique.