基于神经网络专家系统的推理研究

为了克服传统专家系统知识获取难、学习适应能力差、推理效率低等问题,许多专家提出将神经网络与规则专家系统相结合,构建基于神经网络的专家系统模型。文中设计了一种基于神经网络专家系统模型的混合推理机制,通过对基于神经网络推理算法、规则推理算法以及神经网络与规则的混合推理算法进行实验比较,证明本文提出的混合推理机制在改善专家系统推理准确率方面的有效性。

Research on reasoning based on neural network expert system

In order to overcome the shortcomings of traditional expert system, such as hard to acquaint knowledge,poor learning adaptation ability, low efficiency of reasoning and etc, many experts proposed combining neural networks and expert system rules to build expert systems model based on neural network. This paper designed a hybrid inference mechanism based on the neural network expert system model. Through experimental comparison between neural network inference algorithm, reasoning algorithm based on rules and hybrid inference mechanism combing neural network and rules, this paper showed that the effectiveness of hybrid inference mechanism to improve the expert system inference accuracy.