An Algorithm for Optimizing Heuristics is Proposed for Reconstructing Data in Ultra High Voltage Direct Current Control and Protection System

This paper presents a heuristic approach grounded in gradient penalty optimization for enhancing Wasserstein Generative Adversarial Network (WGAN) performance in mitigating the challenges associated with real-time data acquisition deficits within Ultra High Voltage Direct Current (UHVDC) control and protection systems. Through iterative training of the generator and discriminator components, the synthesized data can accurately conform to the intricate distribution patterns inherent in the original dataset. Incorporating gradient penalty serves to enforce Lipschitz continuity within the discriminator, thereby augmenting the training stability and efficacy of the model. The method proposed herein demonstrates favorable outcomes across varying degrees of data loss, spanning from 5% to 40%. Particularly notable is its proficiency in data reconstruction under a 40% data loss scenario, surpassing alternative methodologies.