该文提出一种基于分子吸收光谱法原理的在线分析仪表故障诊断在工程实践中的新方法.该方法以目标驱动的逆向推理为核心,首先对故障信号应用数据挖掘技术进行初步处理得到故障信息:再依据专家知识,从故障原因出发,采用先分类型层次、后进行逻辑运算的搜索方法,确定合适的模糊关系矩阵及逻辑算子,综合构成模糊专家系统.MYSQL数据库用来存储知识库,JAVA语言实现知识库的调取以及模糊推理机的算法编程,同时搭建友好的人机交互界面.最后以西门子的Ultramat23仪表作为对象进行验证实验,结果表明,该文所设计的模型可以基本确保故障诊断的实时性和准确性.%A new method for fault diagnosis of on-line process analyzer which is based on principle of molecular absorption spectrometry in engineering practice is presented in this paper.This method is focused on backward inference which means the thoughts are driven by goal.First data mining technique is applied in preliminary processing of fault signals for fault information.Then according to expert knowledge,we start from classifying fault causes and then use the search strategy of Boolean operation to determine appropriate fuzzy relationship matrix and logic operator for fuzzy expert system.MYSQL database is used to save the knowledge base.JAVA language is used to obtain knowledge from the base and program for fuzzy inference engine,as well as to build a friendly man-machine interface.At last,we take a verification experiment with Siemens Ultramat23 instrument.The results show that the designed model in this paper can ensure the real-time and accuracy of fault diagnosis essentially.
展开▼