Assessment of the Aging State for Transformer Oil-Barrier Insulation by Raman Spectroscopy and Optimized Support...

Area of Science:

• Electrical Engineering
• Materials Science
• Spectroscopy

Background:

• Transformer insulation degradation impacts power system safety and reliability.
• Accurate assessment of oil-immersed barrier insulation aging is critical.

Purpose of the Study:

• To develop and validate a method for assessing transformer oil-barrier insulation aging states.
• To utilize Raman spectroscopy and machine learning for accurate classification of insulation degradation.

Main Methods:

• Indoor accelerated thermal aging experiments to create samples of varying insulation degradation.
• Raman spectroscopy for characterizing aged oil-immersed barrier insulation.
• Data processing including baseline correction, smoothing, and principal component analysis (PCA).
• Support vector machine (SVM) classification with optimized parameters (C and gamma) using grid search, PSO, and GA.

Main Results:

• Raman spectra successfully characterized different aging states of transformer insulation.
• An optimal subset of PCA features enhanced SVM classification accuracy.
• The SVM classifier achieved a high classification accuracy of 94.44% for external validation samples.
• Grid search proved efficient for optimizing SVM parameters.

Conclusions:

• The developed Raman spectroscopy and SVM-based method provides effective assessment of oil-barrier insulation aging.
• This approach offers technical support for condition monitoring and maintenance strategies in power systems.
• The findings contribute to ensuring the reliable operation of power infrastructure.