Research on aviation engine fault diagnosis based on an improved ABC algorithm and CNN-LSTM

To enhance the accuracy and robustness of aviation engine fault diagnosis, this study proposes a fusion method based on an improved artificial bee colony (ABC) algorithm and CNN-LSTM. The diagnostic framework is constructed by achieving global hyperparameter optimization through the improved ABC algorithm, combined with feature extraction from convolutional neural networks and the temporal modeling capability of long short-term memory networks. Experimental results demonstrate that this method achieves 97.3% accuracy, 96.5% recall, and 96.8% F1-score, representing significant improvements over traditional models. This research validates the effectiveness of the integrated optimization strategy, providing a reliable approach for intelligent diagnostics under complex operating conditions.