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。 Traditional expert system fuzzy knowledge rule base is relatively simple, the system in the process of reasoning can hardly gain new knowledge, can not be reasoned self-learning, when the system encounters complex parallel comprehensive fuzzy rules, reasoning mechanism is almost impossible to achieve. Therefore, it is very important to process the complex fuzzy rules in the forward reasoning process, and take the processed preconditions as the input of the neural network to establish a new fuzzy inference engine based on the advantages of radial basis function networks. By reasoning the number of iterations n Changes to change the learning rate factor approach to network reasoning self-learning.