Enhancing Power Transformer Oil Quality Weight Factor using A Genetic Algorithm

Power transformers are critical to electrical power systems but are prone to failures due to factors such as heat, electricity, chemical reactions, mechanical stress, and adverse environmental conditions. Moni-toring the insulating oil effectively is key to preventing these failures. A major challenge in this process is determining the optimal weights for the oil quality index, which lacks a standardized benchmark and often relies on subjective expert assessments. To reduce expert bias and subjectivity, this research utilizes a genetic algorithm to optimize the weightings for five essential parameters: color, water content, break-down voltage (BDV), interfacial tension (IFT), and acidity. The algorithm operates through three stages: crossover, mutation, and selection, and analyzes data from 504 oil tests across various transformers. The mean absolute percentage error (MAPE) is used as the fitness value to assess the algorithm's effective-ness. The optimization process determined the best conditions as 132 iterations, a population size of 180, a crossover rate of 0.2, and a mutation rate of 0.8. These parameters achieved an average MAPE of 1.799% over ten trials, indicating high accuracy. This research not only optimizes the weighting of the oil quality index but also significantly reduces the need for expert input and subjective judgments in trans-former maintenance. The findings are expected to improve the efficiency and reliability of power trans-formers, thereby minimizing failures and associated economic costs.