Anomaly Monitoring Method for Uncertain Nonstationary Industrial Processes

An uncertainty and anomaly monitoring technology, applied to instruments, complex mathematical operations, calculations, etc., to improve detection capabilities and avoid over-fitting problems

Active Publication Date: 2022-04-15
SHANDONG UNIV OF SCI & TECH
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

At present, some probabilistic latent variable models have been applied to anomaly monitoring of industrial processes, but they focus on stationary processes rather than non-stationary processes

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Anomaly Monitoring Method for Uncertain Nonstationary Industrial Processes
  • Anomaly Monitoring Method for Uncertain Nonstationary Industrial Processes
  • Anomaly Monitoring Method for Uncertain Nonstationary Industrial Processes

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0191] What is tested in embodiment 1 is the heat transfer coefficient degeneration fault, Figure 3-Figure 8 The monitoring results of different algorithms are summarized. The fault is a typical small fault. In the early stage of the fault, the three methods based on cointegration analysis, stationary subspace analysis and probabilistic stationary subspace analysis all showed different degrees of underreporting. Statistics have the shortest detection latency, such as Figure 7 shown. The advantage of statistics is that they contain non-stationary information. Additionally, if Figure 4 and Figure 5 As shown, the recursive principal component analysis method gives an earlier alarm in the early stage of the fault, but leaves many false alarms as the magnitude of the fault increases. This reflects the inability of the recursive principal component analysis method to discern minor faults from non-stationary trends.

Embodiment 2

[0193] In Example 2, an additive constant sensor deviation fault, which occurs with an amplitude of 1 in the reaction temperature variable, was tested. Compared with the nominal value of the reaction temperature of 430°C, this is a small fault with a smaller magnitude. In this example, the monitoring results of four different methods are as follows Figure 9-Figure 14 shown. from Figure 9 and Figure 12 It can be seen that the methods based on cointegration analysis and stationary subspace analysis can only partially detect the fault, but leave a certain degree of false positives. Both methods fail to model non-stationary components, thus resulting in unsatisfactory monitoring performance for small faults. In addition, the recursive principal component analysis method can hardly detect the failure (see Figure 10 and Figure 11 ), because the fault magnitude is too small to be masked by the non-stationary trend. For probabilistic stationary subspace analysis, the The...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses an abnormality monitoring method for uncertain non-stationary industrial processes, and specifically relates to the field of industrial process abnormality monitoring. On the basis of the stationary subspace analysis method, the invention considers the process uncertainty and proposes a probability stationary subspace analysis method. The method explicitly models the uncertainty and effectively separates the non-stationary trend from the process uncertainty. Considering the mutual coupling between model parameters, the method decouples the parameters by using the expectation maximization algorithm, and deduces the closed-form solution of iterative update. Based on this model, two detection metrics are proposed for anomaly monitoring under the probabilistic framework. Compared with the existing non-stationary process abnormality monitoring method, the method proposed in the present invention eliminates the influence of process uncertainty, improves the detection ability of small faults in non-stationary process; and can avoid the over-fitting problem of model parameters , to build a more accurate generative model for non-stationary data.

Description

technical field [0001] The invention belongs to the field of industrial process abnormality monitoring, and in particular relates to an abnormality monitoring method for uncertain non-stationary industrial processes. Background technique [0002] Anomaly monitoring is critical to ensuring the normal and efficient operation of industrial processes and equipment. Practical industrial processes usually show significant non-stationary characteristics, that is, the statistical properties of process data will change with time. Non-stationary characteristics can be caused by a variety of factors, including raw material changes, load fluctuations, and equipment aging. Non-stationary characteristics affect the application of traditional anomaly monitoring methods such as principal component analysis, and often cause two types of errors in them. The first type of error is false negative, that is to say, the impact of the fault is covered by the non-stationary trend, resulting in a l...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Patents(China)
IPC IPC(8): G06F17/18G06F17/16G06K9/62
CPCG06F17/18G06F17/16G06F18/22
Inventor 周东华吴德浩陈茂银纪洪泉钟麦英
Owner SHANDONG UNIV OF SCI & TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap