提出了一种基于模糊粗糙集理论的模式识别方法，将动态聚类法和方差分析法引入连续属性模糊化，获取模糊隶属函数，避开了粗糙集理论属性离散化过程带来的信息丢失；利用 检验判断分类的合理性，克服了人为确定分类数目的缺点；应用模糊化得到的模糊决策表进行条件属性约简，通过属性值约简，提取了清晰、简明的故障模式规则。轴承故障模式识别结果表明，该方法对比一般粗糙集理论，有效地提高了模式识别精度，在实际模式识别中具有很好的应用价值。

A method of patterns recognition was presented based on fuzzy rough sets. Dynamic clustering algorithm and method of anaiysis of variance is introduced to fuzzify the continuous condition attribute, and fuzzy membership functions is derived, which avoided losing information caused by discretization in rough set theory. test is introduced to judge the valid analysis of clustering, which has overcome the disadvantage of determining artificially the class number of clustering. The fuzzy decision table obtained by attribute fuzzified is used to attributes reduction, then values of attributes are reducted, and clear and concise pattern rules are obtained. The application to in fault of bearing shows the proposed algorithm can raises the accuracy greatly in contrast with normal rough sets, and is a good method in applications to f patterns recognition.