天津职业技术师范大学学报

应用神经网络推理技术,设计了一种基于尾气分析的发动机故障诊断专家系统。此系统将传统专家系统与神经网络技术科学地加以综合,弥补了传统专家系统的诸多缺陷。专家系统的知识库由两部分组成,即浅层知识库和深层知识库。推理过程由神经网络完成,并带有解释说明模块,主要说明诊断结果的判定依据。实践证明, 这种方法能较好地解决传统专家系统存在的问题,提高系统的诊断速度和准确率。

Neural network system is applied in expert system of engine fault diagnosis based on exhaust gas analysis. The drawback of traditional expert system is overcome by integrating expert system and neural network. Knowledge data of system is made up of simple knowledge data and deep knowledge data. Inferring course is finished by neural network, and results are interpreted by the explain module. The traditional expert systein is improved by the method, and rapidity and veracity is advanced.