基于高斯型隶属函数新的模糊插值推理方法

在稀疏规则库的条件下,当输入的事实落入规则空隙时,采用传统的CRI方法是得不到任何推理结果的。KH线性插值推理解决了这个问题,从而产生了模糊插值推理方法。但现有的插值推理方法几乎都是基于三角形隶属函数的,很少有高斯型隶属函数。本文在线性插值推理方法的基础上,给出一种基于高斯型隶属函数的顶点和拐点距离比的模糊插值推理方法。这为模糊插值推理又提供了一个十分有用的工具。

New Method of Fuzzy Interpolative Reasoning Based on Gaussian-type Membership Function

When rule base is sparse, no reasoning result was obtained by traditional CRI method as the factual exent input was in the gap between two neighboring antecedents. Koczy and Hirota had proposed a linear interpolative reasoning method, which gave a solution to the problem, so fuzzy interpolative reasoning was bom. But now, all of the interpolative reasoning methods were almost based on triangular-type membership function, a little based on Gaussian-type membership function. The method of fuzzy interpolative reasoning based on the proportion of vertex and inflection point of Gaussian-type membership function was presented, which provided a useful tool for fuzzy interpolative reasoning.