Fault Diagnosis Method of Morphing Aircraft Based on Particle Swarm Optimization and Convolutional Neural Network

To address the challenges of sensor fault diagnosis in the complex design and dynamic flight conditions of morphing aircraft, this paper proposes an intelligent fault diagnosis method that combines Particle Swarm Optimization (PSO) with Convolutional Neural Networks (CNN). This method utilizes PSO to optimize CNN hyperparameters, including learning rate, convolution kernel size, and number of convolution kernels, to enhance the CNN's feature extraction and fault classification capabilities. This optimization improves the accuracy of fault detection and identification. Traditional CNNs often require manual hyperparameter tuning, which can lead to suboptimal results and struggle to adapt to dynamic environments like those of morphing aircraft. Although swarm algorithms offer advantages, they face practical challenges such as slow convergence and high computational complexity. However, the proposed PSO-CNN method demonstrates robustness by addressing these issues, achieving a diagnostic accuracy exceeding 99% under various sensor fault conditions, and significantly outperforming traditional CNN models. These results confirm the effectiveness of the proposed method in enhancing flight safety and reliability.