Abstract: For fault problems of micro-nano satellite attitude control system using pulsed plasma thruster as the actuator, the support vector machine technology was used to detect and isolate two kinds of common faults of pulsed plasma thruster. The wavelet kernel function parameters of Mexico hat were optimized by adaptive genetic algorithm (AGA). Combined with wavelet analysis, the hyperplane optimization efficiency and generalization ability of SVM classifier were improved by these methods. Finally, the experimental results prove that the method can accomplish fault diagnosis tasks quickly and accurately, and also prove the superiority and effectiveness of wavelet kernel function support vector machinesin fault diagnosis.

Key words: micro-nano satellite, support vector machine, wavelet analysis, fault diagnosis, pulse plasma thruster, genetic algorithm