中文摘要：

基于模糊K近邻算法对模糊支持向量机中模糊类别隶属度的计算进行了改进，将距离与样本之间的关系相结合进行加权弥补了FSVM算法的不足．引入CCA算法对语音特征矢量进行降维处理，有效减小了特征之间的冗余信息，通过识别实验对传统的SVM、FSVM以及基于模糊K近邻的FSVM的算法性能进行了比较和分析．

英文摘要：

In the process of calculating the fuzzy class membership function, the distance and relationships between the samples are weighted based on fuzzy K neighbor algorithm. In the experiment, firstly, the dimensions of the speech feature vector are reduced by CCA method. Then, with the results of experiment, the performance of traditional SVM, Fuzzy SVM, Fuzzy SVM based on Fuzzy K neighbor nearest algorithm is compared.