ARSM方法的改进及其在冷机两器故障诊断中的应用

目前多数的近似推理都是基于Zadeh提出的模糊关系合成（CRI）的方法，但是CRI本身存在着缺陷．基于相似度度量的近似推理方法可以很好地弥补CRI推理的缺陷，因而目前受到广泛关注．在前人研究的基础上，对基于相似度度量的推理方法提出了改进，提出了对于不同的规则形式，采取不同的近似推理计算方法．故障诊断专家系统是目前研究的热点，以两个具体的推理案例为例说明了新计算方法在冷水机组蒸发器和冷凝器故障工况诊断过程的应用．通过给出的推理案例可知，此种改进的近似推理方法较CRI方法更易理解，且精度较高．

Improved ARSM and Its Application in Fault Detection and Diagnosis of Evaporator and Condenser of Chillers

Most of the present approximate reasoning （AR） systems are based on the method of compositional rule of inference （CRI） presented by Zadeh,yet there are some obvious flaws in the CRI methodology. As approximate reasoning based on similarity measure （SM） can make up for such flaws ,it has been attracting lots of attentions recently. Approximate reason- ing methods were applied according to different rules＇ form to enhance the traditional ARSM methodology. FDDES （fault detection and diagnosis expert system） is a heated research area recently and the proposed reasoning methodology was applied in the centrifugal chiller＇s FDDES. Field experimental results show that the proposed reasoning methodology is easier to understand and is more precise compared with traditional CRI.