重构与趋势分析相结合的工业过程异常原因诊断

针对传统重构方法难以分离方向相似的产品质量或运行状态相关的故障或异常,提出了一种重构与定性趋势分析相结合的异常原因诊断方法.首先,采用主元分析算法进行建模和监控,并利用重构方法提取故障方向和估计故障幅值;接下来,分别利用角度相似度和定性趋势分析定义故障方向和故障幅值序列的相似度;然后,根据故障方向和故障幅值变化趋势的相...

Reconstruction and Trend Analysis Based Abnormality Cause Diagnosis of Industrial Processes

In order to overcome the limitation of traditional reconstruction methods on isolating qualityrelevant or operational faults with similar directions, a new abnormality cause diagnosis method that incorporates reconstruction and trend analysis method is proposed. First, the principal component analysis algorithm is used for modeling and monitoring while the reconstruction method is used to extract the fault direction and estimate the fault amplitude. After that, the angle similarity and qualitative trend analysis are used to define the similarity of the fault directions and the fault amplitude sequences respectively. A new comprehensive similarity index that incorporates the fault direction similarity and fault amplitude trend similarity is defined. The fault cause is diagnosed using the comprehensive similarity index between the current fault and historical fault candidates. Finally, data collected from a Monte Carlo simulation experiment are used to compare the proposed method with the traditional methods. The results demonstrate the effectiveness of the proposed method.