中文摘要：

针对目前实际工业系统不允许在运行中进行实时的故障诊断试验及研究的问题，设计了基于Matlab中GUI的高温硝酸冷却系统的故障诊断仿真试验平台。本试验平台采用符号有向图的深层知识模型和反向推理方法，利用系统模型中的相容通路，推理基本故障的故障诊断规则，引入信息粒度和信息粒化原理，实现了粒矩阵的知识约简算法，构建了基于粒计算-SDG的故障诊断方法。最后通过实验验证了该仿真试验平台的有效性。

英文摘要：

As the real industrial system does not allow research of real-time fault diagnosis in the actual process device, GUI visual program in Matlab was used to design simulation platform for hot nitric acid cooling fault diagnosis. By using Signed Directed Graph（SDG） deep knowledge model and inverse direction inference technique and using consistent path, the basic rules of fault diagnosis were reasoned. Information Granularity and Information Granulation were introduced into construce fault diagnosis method based on GrC-SDG, which achieves Granular Matrix-based knowledge reduction algorithm. Finally, the effective ness of the fault diagnosis simulation plat-form were illustrated with experiments.