基于变权理论的模糊推理前提约化的一类方法

A PREMISE REDUCTION METHOD ON FUZZY INFERENCE BASED ON VARIABLE WEIGHTS THEORY

关于传统模糊推理方法在规则前件聚合上存在的一些问题,采用因素空间理论中的变权思想,对模糊推理规则前件进行“综合”,不仅体现了各前件分量在模糊推理中的相对重要性,并且考虑了前件分量的赋值大小对推理结果的均衡影响.另外,指明对规则前件的这种“综合”不改变CRI算法的逼近性,故更有实际价值.

As for some problems on the aggregation of rule premises in traditional fuzzy inference method, a kind of rule premises aggregation method is proposed based on variable weights in factor spaces theory, which not only shows relative importance of every premise in fuzzy inference, but considers the influence of premises' valuations on inference conclusions. In addition, it is pointed out that the aggregation of rule premises doesn't change the approximability of CRI algorithm. There fore, the method is practicable.