Differential Evolution Algorithm Optimized ELM Fault Diagnosis of Flight Control System Actuator

The effect of the algorithm based on ELM (Extreme Learning Machine) and improved differential evolution algorithm on fault diagnosis of the flight control system actuator is studied. As a machine learning algorithm, ELM has the characteristics of fast speed and high precision. In order to further improve the generalization performance of ELM training, the differential evolution algorithm is combined with the ELM algorithm to improve its fault diagnosis effect. First, the 3DOF(three-degree-of-freedom) aircraft fault diagnosis experimental platform is used to collect the fault data of the actuator; then the global iterative optimization of the differential evolution algorithm is used to do optimization on the network parameters trained by the ELM with the fault data as well as the threshold, the training results show that the network algorithm has better generalization performance; finally, the trained model is used to detect the same type of actuator fault data. According to the experimental results, the method performs well in fault diagnosis, which shows the effectiveness of the method.