Prediction of critical condensation temperature for high voltage switchgear based on TimeGAN-augmented data

In southern China, the "Returning Dampness" climate often causes moisture condensation in high-voltage switchgear, leading to equipment malfunctions or failures. The complex mechanism of moisture condensation makes it difficult to accurately model and predict the critical condensation temperature, especially with limited data samples. Addressing the challenge of building a prediction method that does not rely on precise models in data-scarce situations is an urgent need. This paper proposes a method for predicting the critical condensation temperature of high-voltage switchgear based on TimeGAN-augmented data, aiming to improve prediction accuracy. First, to address the issue of data scarcity, TimeGAN is employed to augment the training samples. Then, a Improved Grey Wolf Optimizer (IGWO) is used to iteratively optimize the hyperparameters of a BiLSTM model, constructing the IGWO-BiLSTM model to predict the critical condensation temperature. Finally, simulation validation is performed using real-world data from a 35kV high-voltage switchgear in Fujian, China. The simulation results show that the proposed method significantly improves prediction performance compared to model-driven approaches, providing a foundation for condensation prevention in high-voltage switchgear.