用于专家系统规则库的冗余校验方法研究

Research of detecting redundancy method for rule-based in expert system

产生式规则是目前应用较多的一种知识表示方法。在用于确定发酵过程生物量软测量混合模型结构的专家系统中,当向产生式规则知识库添加新的规则时,冗余的存在会影响推理的效率以及推理的准确性。提出了一种用于该专家系统规则库的冗余校验方法,给出了冗余规则的判别、冗余规则的处理以及冗余校验的实现方法。实验结果表明,该冗余校验方法可以根据输入条件和已有规则,判断出新添加的规则是否冗余,并在消除冗余对推理效率影响的同时,降低模型复杂度,有利于优化混合模型的结构。

Production rules is a knowledge representation applied in numerous fields presently.In expert system for confirming structure of biomass soft-sensor hybrid model in fermentation process,when adding new rule into rule-base,exist of redundancy affects reasoning efficiency and reasoning veracity.A detecting redundancy method for rule-based in the expert system is introduced in this paper.Distinguish of redundancy,disposal of redundancy and realization of redundancy verification are presented.Experiment shows that,the detecting redundancy method can judge whether rule added is redundant according to inputting conditions and existent rules,and can debase complexity of model when avoiding effect of redundancy to reasoning efficiency,which favors to optimize structure of hybrid model.