摘要： 本文在分析影响专家系统效率的主要因素基础上,提出了一种基于光并行匹配的专家系统模型ESBOPM及其光电混合型硬件支持系统。在ESBOPM中,为了适应光并行匹配的需要采用了知识的类产生式表示方法、知识编码、知识匹配操作并行化等技术。在硬件支持系统方案中,利用光信息处理技术并行化特点实现匹配操作并行化的硬件支持。性能分析和实验结果表明ESBOPM及其硬件支持系统是有效的、合理的并且能显著地提高专家系统响应速度。

关键词: 专家系统, 光计算, 光电混合系统

Abstract: After analysis of main factors affecting an expert system, the paper proposes an expert system model based on optical paralled matching (ESBOPM) and a scheme of electro-optical hybrid hardware support system. The ESBOPM adopts quasi-production representation of knowledge, knowledge coding and knowledge matching operation parallellism techniques to suit the requirements of optical parallel matching. In the scheme of hardware support system, the parallel characteristics of optical information processing technique is used to realize the hardware support matching operation parallellism. The performance evaluation and experiment results show ESBOPM and its hardware support system are valid and increase the matching speed to 30.46 times (as compared with serial matching) or 7.55 times (as compared with indexed matching).