基于专家思维的多维度故障诊断方法

Multi-dimensional Fault Diagnosis Method Based on Expert Thinking

针对目前专家系统诊断准确度不高的问题,提出一种基于专家故障诊断思维方式的多测点、多时间点、多敏感参数相融合的故障诊断方法。从故障诊断专家对设备异常进行分析诊断的思路出发,依据故障机理、故障响应特点及故障劣化规律,选择相关测点、多时间点的数据来构建敏感参数矩阵,并生成对应的故障矩阵和权重矩阵,对设备异常状态进行诊断。利用该方法对某石化一离心压缩机异常状态进行分析诊断,结论准确,且能避免不同时间点诊断结论冲突问题。

In view of the low diagnosis accuracy of expert system at present, a multi-dimensional fault diagnosis method based on expert thinking is proposed, which fuse information of related points, multi-times and multi-sensitive parameters. Following the thinking when fault diagnosis expert analyze and diagnose the machine＇s abnormal state and base on the failure mechanism, response characteristics and deterioration law, select the data of related-points in multi-times to establish the matrix of sensitive parameters, then generate the fault matrix and weight matrix to diagnose the machine＇s abnormal state. The result is accurate and conflicts between different times can be avoided when using this method to analyze and diagnose the abnormal state of a machine in real petrochemical.