Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Multi-modal process fault detection method based on weighted SVDD

A fault detection and multi-modal technology, applied in the direction of kernel methods, data processing applications, instruments, etc., can solve the problems that the design of weighting factors needs to be improved, rarely consider the nonlinear and non-Gaussian characteristics of multi-modal data, and achieve effective Conducive to modeling, sensitive identification, and avoiding the effect of overfitting

Active Publication Date: 2021-01-05
HUAZHONG UNIV OF SCI & TECH
View PDF10 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] To sum up, the existing multimodal monitoring methods lack the utilization of the potential characteristics of multimodal data, and rarely consider the nonlinear and non-Gaussian characteristics of multimodal data; while the design of weighting factors in weighted SVDD needs to be improved

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
  • Multi-modal process fault detection method based on weighted SVDD
  • Multi-modal process fault detection method based on weighted SVDD
  • Multi-modal process fault detection method based on weighted SVDD

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0063] A multimodal process fault detection method based on weighted SVDD, such as figure 1 As shown, including: offline modeling phase and fault detection phase;

[0064] The offline modeling phase includes:

[0065] Obtain normal data of different modalities to form an initial data set X, and use a sliding window of length w to slide on the initial data set X to obtain a window data sequence;

[0066] Calculate the statistics of each window data in the window data sequence, respectively as a sample, and calculate the local reachable density ratio of each sample respectively, as the weight factor corresponding to each sample, so as to use the weighted SVDD algorithm to establish a hypersphere model; The accessible density ratio is the normalized value of the local accessible density of the sample;

[0067] Optimizing the hypersphere model to obtain the Lagrangian factor α i , and the center a and radius R of the hypersphere; among them, 1≤i≤N, N represents the length of th...

Embodiment 2

[0122] A computer-readable storage medium includes a stored computer program. When the computer program is executed by a processor, the device where the computer-readable storage medium is located is controlled to execute the weighted SVDD-based multimodal process fault detection method provided in Embodiment 1 above.

[0123] The beneficial effects obtained by the present invention will be further explained in conjunction with specific application scenarios below.

[0124] In one of the application scenarios, 200 low-density banana-shaped two-dimensional samples are generated as mode 1, 100 high-density normal distribution two-dimensional samples are generated as mode 2, and 3 outliers close to high-density samples point. The global density and weighting factor of each sample are respectively as figure 2 and 3 As shown, the 1st to 200th sampling points are the sampling points of mode 1, the 201st to 300th sampling points are the sampling points of mode 2, and the 301st to ...

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 a multi-modal process fault detection method based on weighted SVDD, and belongs to the technical field of multi-modal monitoring of an industrial manufacturing process. The method comprises the steps: obtaining normal data of different modals, forming an initial data set X, and sliding on the X through a sliding window with the length of w, and obtaining a window data sequence; calculating statistics of each window data as samples, respectively calculating a local reachable density ratio of each sample as a corresponding weight factor, establishing a hypersphere modelby using a weighted SVDD algorithm, and performing optimization calculation to obtain a Lagrange factor alpha i, a center a and a radius R of a hypersphere, wherein the local reachable density ratio is a value after local reachable density normalization; calculating the statistical magnitude of to-be-detected detection window data Xon to serve as a to-be-detected sample; calculating the distance DIST from the to-be-detected sample to a according to alpha i; if DIST is larger than R, judging that the system breaks down; and if DIST is less than or equal to R, judging that the system is normal.According to the invention, the accuracy and sensitivity of the multi-modal process fault detection process can be improved.

Description

technical field [0001] The invention belongs to the technical field of multimodal monitoring of industrial manufacturing processes, and more particularly relates to a multimodal process fault detection method based on weighted SVDD. Background technique [0002] Due to factors such as changes in the external environment, changes in the production plan, or inherent characteristics of the process, the production process has multiple stable working conditions. The multimodality of multimodal data distribution makes traditional multivariate statistical process monitoring methods unable to be directly applied to multimodal process monitoring. [0003] At present, most monitoring algorithms are directly oriented to multi-modal raw data, but some characteristics between modes are difficult to reflect in the raw data. If the two modalities have great differences in the direction of change, but have a large number of overlapping parts in the spatial position, in many existing algori...

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 Applications(China)
IPC IPC(8): G06Q10/06G06N20/10
CPCG06Q10/0639G06N20/10
Inventor 王兆静郑英张洪王彦伟
Owner HUAZHONG UNIV OF SCI & TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products