一种新的基于图模型的间歇过程故障监测方法

目前多元统计方法被广泛用于间歇过程故障监测并已经取得了比较好的效果。但是，统计模型的可解释性能比较差，很难直接利用操作人员积累的安全经验。为应对这些不足，提出了一种基于图模型的间歇过程故障监测方法。利用提出的定性建模方法，过程机理及操作人员的安全经验能够方便地表达。利用在线的过程变量数据和正向推理算法推断生产应处于的状态，利用安全知识及时地发现生产异常。当推断的结论与在线测得的结果矛盾或过程变量超过设定的安全限时，给出解释性输出。通过一个间歇反应案例，验证了提出的方法在模型的可解释性和利用安全经验方面的优势。

A novel fault detection method for batch process based on digraph model

So far, the multivariate statistical methods have been widely used to perform fault detection for batch process, and they have shown some good performances. However, the interpretability of statistical model is not good, and it is difficult to directly use the safety experience of operators. In order to cope with these shortcomings, a fault detection method based on digraph model was proposed. By using the proposed qualitative modeling method, the process mechanism and safety experience of operators can be expressed easily. On line data of process variables and forward reasoning algorithm were used to deduce the state of production process. The safety knowledge was used to find abnormal production state in time. When deduced conclusion was not consistent with measured process variable value or process variable value exceeds threshold, an explanation for this situation would be provided. By means of a batch reaction case, the advantages of proposed method on model interpretability and using safety experience were verified.