基于自适应神经网络的模糊推理

传统专家系统的模糊知识规则库比较简单,系统在推理过程中几乎不能获得新知识,不能进行推理自学习,当系统遇到复杂并行综合模糊规则时,推理机制几乎无法实现.所以将复杂模糊规则进行正向推理处理就显得重要,并将处理后的规则前件作为神经网络的输入,结合径向基函数网络的优点建立新的模糊推理机,并通过推理迭代次数n的变化来改变学习率因子的方法进行网络推理自学习.

Fuzzy Reasoning on the Basis of the Neural Network of the Self-adaptation

The regular storehouse of fuzzy knowledge of the traditional expert system is simpler, the system can't almost obtain new knowledge in the course of reasoning and can't carry on reasoning study, when the system meets the complicated and comprehensive fuzzy rule, the reasoning mechanism is almost unable to realize. So it is important that the complicated fuzzy rule is reasoned forward, the result which the system had been treated will act as the input of neural network, combined advantage of the radial network, the system sets up the new fuzzy reasoning machine and will reason and study through adopting the reasoning pile in order to take the place the change of number's times.