基于克隆选择的粗糙集属性约简方法

基于免疫克隆选择的原理，提出了一种新的粗糙集属性约简方法，将属性集合的分类近似质量作为进化目标，利用免疫反应的分布性特点通过局部并行搜索实现全局优化，并采用抗体更新和亲和力抑制手段来维持群体的多样性，保证了各抗体局部优化解的稳定性，从而获得了多个优化约简集合，通过机械故障诊断数据的实例应用，表明该方法可以获得多个符合分类质量要求的属性约简集合，因此满足了设备故障诊断的特征优化选择要求。

Reduction of Rough Set Attribute Based on Immune Clone Selection

A novel attribute reduction approach of rough set based on immune clone selection was proposed. In this method, the approximation quality and attribute set were used as evolution objection and antibody, respectively. On the basis of the inherent distribution within the immune response, the global optimization of antibody was realized through parallel local optimization. Moreover, the diversity of antibody population was maintained with the affinity suppression and renewal of antibody. Thus the stable multilocal optimal solutions can be preserved. In addition, the machinery fault data were analyzed by this method, and the attribute reduction sets were obtained furthest to satisfy the demand of feature selection in machinery diagnosis.