从属性集角度研究变精度粗糙集模型的属性约简问题,在对象集上定义了一种β下近似二元关系,并利用这种关系建立了属性集及其幂集上的等价关系,由此产生依赖空间。同时利用定义的二元关系和依赖空间给出了变精度粗糙集的β下近似协调集的判定定理,得到一种保持每个决策类的β下近似不变的属性约简方法。最后通过实例验证方法的有效性。 From the point of view of attribute set, we study the problem of attribute reduction of variable precision rough set model, and define a bounded lower approximation binary relation on the object set. By using this relationship, we establish the equivalence relation between attribute set and its power set, The resulting space-dependent. At the same time, by using the definition of binary relation and dependence space, the judgment theorem of β lower approximation coordination set of variable precision rough set is given, and an attribute reduction method is obtained which preserves the β approximately invariant of each decision class. Finally, the validity of the method is verified by an example.