Fault Diagnosis of Loitering Munitio Controller Based on CNN Algorithm
The application environment of the Loitering Munitionis located is complex and have high re-quirements for security and reliability.The controller is the core component to ensure its safe and reliable flight,and it is of great significance to perform efficient and accurate diagnosis of the controller.First of all,analyze the fault of the sensor and the execution structure of the loitering munition controller,and make mathematical modeling for each failure,and propose the fault model of the controller.Adopting deep learning to diagnose the fault,first design the CNN-based diagnostic model network structure,and set up different learning rates and steps for experiments to improve the CNN algorithm to improve the identifica-tion accuracy and stability of the model.Finally,the network structure of deep learning is deepened,and different optimizers,learning rates,and activation functions are set up as comparison experiments to obtain the most suitable model parameters.Through experimental verification,the improved CNN algorithm can effectively improve the accuracy of the experiment to achieve high-precision fault diagnosis.
loitering munitionfault diagnosisconvolutional neural networkflight control systemma-chine learning
NETL
NSTL
万方数据
