Abstract: Aiming at the problem of basic probability assignment of D-S theory, a hyper-sphere support vector machine method is proposed. Hyper-sphere support vector machines are used to train classification model in fault spaces generated by sensors. The basic probability assignment can be got by point-class membership function and class-class membership function. D-S theory is used to fuse evidence and believe function is used to decision making. Numeric experiments show that the accuracy and speed of classification are improved, and this method is suitable for practical use.

Key words: Support Vector Machine, D-S theory, fault diagnosis

摘要： 针对使用多传感器信息融合技术进行故障诊断时，故障模式较多、基本概率赋值难以确定的问题，提出一种基于超球支持向量机与D-S证据理论相结合的故障诊断方法。该方法使用超球支持向量机针对每一个传感器的故障空间训练分类模型，根据类内隶属度与类-类相似度得到各故障类别的基本概率赋值，利用D-S证据理论进行证据融合，基于信任函数进行故障决策。试验结果表明该方法提高了故障识别能力，有一定实践意义。

关键词: 支持向量机, D-S证据理论, 故障诊断