Abstract: This study aims to improve the measurement accuracy of the voltage divider ratio of capacitive voltage transformers (CVTs) and explore a calibration scheme that adapts to complex operating conditions. Through in-depth analysis of the causes of CVT voltage divider ratio errors, combined with improved particle swarm optimization algorithm, a dynamic correction model that balances temperature drift and harmonic interference was constructed. Adaptive weighting, mutation operation, and neighborhood topology adjustment strategies were designed to optimize the search process of correction parameters, and the effectiveness of the model was verified through experimental platforms. Research has shown that this method is significantly superior to traditional hardware compensation and linear regression correction methods in terms of calibration accuracy, response speed, and environmental adaptability.