实时控制神经网络专家系统结构及推理算法

STRUCTURE AND REASONING ALGORITHMS OF NEURAL NETWORK EXPERT SYSTEM USED FOR REAL TIME CONTROL

为了弥补基于单一人工智能方法所形成的智能控制系统的不足,提高基于产生式规则的控制专家系统的性能,本文提出了一种基于神经网络推理机及结合型外部知识表示的神经网络专家系统模型、基本结构及工作原理;在此基础上提出了一种实时控制神经网络专家系统(RTNNES)结构及其知识表示技术;提出了联合使用神经网络与专家系统技术的实时正向推理算法,基于事件的非单调推理算法。

In order to remedy the defects of Intelligent Control System (ICS) which are based only on Expert System (ES) or Neural Network (NN), and improve the performance of production-rule-based ES, we presented a novel type of ICS in which NN is integrated with ES. The ICS is called Neural Network Expert System (NN-ES). A novel Real Time control NN-ES model, its basic structure and working principle, and knowledge representation method are presented. A Real Time Forward Chaining and an event-based non-monotonic reasoning Algorithm are also give in this paper.