Effective assessment of short-term voltage in power systems using RNN-LSTM

Short-term Voltage Stability assessment plays an important role in maintaining power supply security. However, current assessment methods show low accuracy and poor computational efficiency. To tackle these issues, a Short-term Voltage Stability assessment model based on Recurrent Neural Network and Long Short-Term Memory is introduced. The model applies Hierarchical Clustering and Least Square Generative Adversarial Network to perform cluster analysis and data augmentation. It uses Back Propagation Neural Network to predict short-term voltage state information. Based on the predicted state information, the model further conducts Short-term Voltage Stability assessment through a hybrid deep learning framework. The result shows that, under the performance analysis experiments, the proposed model obtains a coefficient of determination equal to 0.927. As compared to other benchmark models, the assessment accuracy of this proposed model is improved. Moreover, when the observation cycle equals 6.0, the accuracy of assessment is equal to 0.963. Additionally, the computation time of the model is equal to 7.51 s. The computation time is significantly lower than that of other benchmark models. The result shows that the proposed model improves accuracy and computation efficiency to complete the Short-term Voltage Stability assessment task. The proposed model is beneficial to the sustainable development of smart grid and shows great value in engineering applications.