参数自适应的析取云模糊置信规则识别方法

Disjunctive Cloud Fuzzy Belief Rules for Recognition with Adaptive Parameters

为获得准确的模糊置信规则结构参数,提出了参数自适应的析取云模糊置信规则识别方法 .为完成模糊域的自适应划分,提出了基于频数统计的双门限检测方法和基于包含度的双门限检测方法 .用云模型作为模糊集,改变熵系数和超熵系数,实现对模糊集形状的调整;前提属性的联接设置为析取逻辑关系,改进了证据的基本概率赋值方式,对规则权重和属性权重进行了优化.实验结果表明,与其他方法相比,本文方法的正确识别率提高了5%~15%,规则可解释性更强.

In order to obtain the accurate structural parameters of fuzzy rules, a disjunctive cloud fuzzy belief rules based recognition method with adaptive parameters is proposed. In order to complete the adaptive division of fuzzy domain,a double threshold detection method based on frequency statistics and inclusion degree separately are proposed. Cloud model is used as fuzzy set, of which by changing entropy coefficient and super entropy coefficient to adjust the shape of fuzzy set, and the connection of premise attributes is set as disjunctive logic relation. The basic probability assignment of conflict evidence is improved, and a programming model is built to optimize the rule weight and attribute weight. The experimental results show that compared with other methods, the correct recognition rate of our method is improved by5%~15%, and the rules are more interpretable.