Transient stability analysis of multi-electric aircraft power system based on relief and CNN-LSTMP

The transient characterization problem caused by high-power nonlinear load switching in a multi-electric aircraft power system is addressed in this work using a stability analysis method based on the combination of Relief and CNN-LSTMP. High correlation features are chosen by applying the Relief algorithm to determine the feature weights. In the meantime, to successfully address the sample imbalance problem, the cost factor is added to the CNN loss function. With the Projection layer, the LSTM is enhanced and optimized while resolving the large-scale computation issue and lowering computational complexity. The transient switching process was simulated in the article to produce time series data, which was then normalized and annotated for model training. A multi-electric aircraft power system simulation model was established using stability analysis, and it performed exceptionally well. Based on the experimental results, the proposed method meets the requirements of practical applications, achieves excellent transient analysis effect, and achieves 96.33% accuracy in transient stability analysis. This offers a workable solution for the application of deep learning to the field of stability analysis of airborne power systems.