Probabilistic Stacked Denoising Autoencoder for Power System Transient Stability Prediction With Wind Farms
Tong Su, Youbo Liu, Junbo Zhao, Junyong Liu
Abstract
To address the uncertainties of renewable energy and loads in transient stability assessment with credible contingencies, this letter proposes a stacked denoising autoencoder (SDAE)-based probabilistic prediction method. The correlations among wind farms have been effectively considered through the variable transformation via the Cholesky decomposition. SDAE allows learning the mapping relationship between operational features and the transient stability margin. The possible operation scenarios are sampled under different confidence levels to generate appropriate inputs for SDAE to assess the probabilistic transient stability distribution. Results on the modified IEEE 39-bus system show that our proposed method can achieve a similar level of accuracy as the benchmark and improved Monte Carlo simulations-based methods while having much higher computational efficiency.