Physics-Informed Fast Transient Stability Assessment of Non-Fixed Length in Power Systems

Abstract

Against the backdrop of “dual carbon” goals, the construction of a new power system with new energy as the main component is the main direction and key way for the transformation and upgrading of the power industry. Research into fast and accurate evaluation of transient power angle stability in the context of new power systems is of great significance. To address this, a new transient power angle stability evaluation method is proposed for power systems with grid-forming new energy based on the physics-informed sequence-to-sequence (PI-seq2seq) neural networks and cascaded convolutional neural networks models. First, the PI-seq2seq network structure is used to predict the future power angle trajectory, and a loss function with physical loss terms is constructed to guide the model training process, which avoids the long-duration time-domain simulation to ensure fast transient evaluation. Then, predicted power angle trajectory is taken as input by the cascade convolutional neural networks to evaluate the transient stability and its confidence level. A threshold judgment mechanism for the evaluation confidence level is configured to realize the transient stability judgment of the non-fixed evaluation length, which overcomes the impact of the fixed power angle curve length on the evaluation results. Finally, the method proposed is verified in the Kundur system, and the simulation results show that it has obtained satisfactory results in both the power angle curve prediction and the stability evaluation.

Keywords

• grid-forming new energy
• physics-informed sequence-to-sequence (pi-seq2seq) neural networks
• power angle trajectory prediction
• cascade convolutional neural networks
• transient power angle stability assessment