Search

In order to deal with the problem of low classification accuracy in thruster fault identification based on single fault feature, a fault identification algorithm based on time-domain energy and time-frequency entropy of fusion signal is proposed. Firstly, the fault singular signals from two single aspects, such as surge speed and control voltage, are fused into a fusion signal to reflect fault information more comprehensively. Then, the peak region energy feature of the fusion signal is extracted in time domain, and the entropy feature of the fusion signal is extracted in time-frequency domain, so as to obtain the multi-domain fault features. Finally, based on support vector data description algorithm, a multi-classifier is established, and the relative distance between fault sample and each hypersphere in the multi-classifier is calculated. The fault severity corresponding to the fault sample is determined by the minimum relative distance. The experimental results of an experimental prototype in a pool show that the classification accuracy of the proposed method is 95.2%. In comparison to the classification models corresponding to the surge speed and control voltage, the classification accuracy is increased by 5.2% and 22.8% respectively.