论文部分内容阅读
化工过程的回路众多、关系复杂,一个回路的振荡可能会波及其他回路,多个回路的异常往往只由一个故障点引起。因此分析多回路的振荡传播路径并找到故障根源对于消除故障十分重要。流程工业普遍存在的低频噪声会干扰一般振荡检测方法的准确性,本文以TE过程为例,用频域滤波后的自相关函数(ACF)解决低频干扰严重且频率非单一的振荡检测问题,通过对多回路过程输出的功率谱提取主元,并根据主元相似性对回路聚类,分析振荡的传播路径,最后借助双相干等非线性检测方法定位厂级振荡的根源。同时本文基于上述方法开发了振荡检测与诊断工具箱,对多回路系统的数据进行自动分析。
The chemical process has many loops and complex relationships. The oscillation of one loop may affect other loops, and the abnormality of multiple loops is often caused by only one fault point. Therefore, it is very important to analyze the multi-loop oscillation propagation path and find the root cause of the fault. The low-frequency noise prevailing in the process industry interferes with the accuracy of the general oscillation detection method. In this paper, taking the TE process as an example, the frequency-domain filtered autocorrelation function (ACF) is used to solve the oscillation detection problem with severe low frequency interference and non-uniform frequency. The principal components are extracted from the power spectrum output from the multi-loop process, and the loop is clustered according to the similarity of the principal components to analyze the propagation path of the oscillation. Finally, the root cause of plant-level oscillation is located by nonlinear detection methods such as bicoherence. At the same time, this paper develops an oscillation detection and diagnosis toolbox based on the above method to automatically analyze the data of multi-loop system.