An AUV thruster fault diagnosis method based on the improved SVDD

In the research of thruster fault diagnosis method using Support Vector Domain Description (SVDD) for Autonomous Underwater Vehicle (AUV), it is difficult to get the optimal kernel function parameters of the SVDD classification model by the traditional parameter optimization method. To solve these problems, an improved SVDD fault pattern classification method is investigated. It improves the classification performance of the SVDD model by describing the distribution forms and rules of the mapping data in a high dimensional feature space and optimizing the kernel function parameters based on the non-Gaussian measurement of maximum entropy principle. The results of the thruster fault simulation experiment of the “Beaver-II” AUV prototype show the effectiveness of the proposed method.