一种鲁棒性较强的新神经元模型及其在模糊推理中的应用

A NEW ROBUST NEURON MODEL AND IT′S APPLICATION IN FUZZY INFERENCE

提出一种具有较强鲁棒性的神经元模型 ,能实现与、或和混合 -并模糊逻辑运算。利用该神经元模型 ,构造了一个神经网络 ,基于该网络的模糊推理方法远比Zadeh的CRI方法更易于满足模糊推理的一致性要求 ,可望在面向医学和人文社会科学等领域的模糊专家系统及模糊决策支持系统中获得应用。

A new fuzzy neuron model proposed in the paper is robust, and can realize fuzzy logic operations of AND, OR and MIXED UNION. Consisting of the new neurons, a fuzzy neural network is constructed here. Based on the neural network, the new method of fuzzy inference is much easier to satisfy consistency principle of fuzzy inference than Zadeh′s CRI. It is likely that the new model is applicable to fuzzy expert systems and fuzzy decision support systems in medicine and literal humanity fields.