Fault Detection in Weak AC Power Grids with Double-fed Wind Turbines using Hankel-EfficientNet

In the past decade, renewable energy has emerged as a pivotal component of modern power systems. However, fault detection in weak AC power grids, characterized by a high proportion of renewable energy generation, has become increasingly challenging. Addressing this challenge necessitates enhancing the reliability of the power system. This paper presents a fault detection algorithm leveraging the Hankel-EfficientNet approach within a deep learning framework. The methodology involves simulating output data via a simulation model and transforming it into the requisite image dataset using Hankel transformation. Subsequently, the EfficientNet neural network is deployed to train the dataset, and an appropriate loss function is selected for model optimization. Experimental results demonstrate that the average accuracy of the Hankel-EfficientNet network attains 100%, thereby fulfilling fault detection requirements in AC power grids.