Eureka AIR delivers breakthrough ideas for toughest innovation challenges, trusted by R&D personnel around the world.

Mechanical fault diagnosis method based on multi-sensor information fusion migration network

A multi-sensor, mechanical failure technology, applied in the testing of mechanical components, biological neural network models, and testing of machine/structural components, etc., can solve the problems of unsatisfactory fusion effect and inability to automatically learn deep feature representation.

Active Publication Date: 2021-01-01
SOUTH CHINA UNIV OF TECH
View PDF11 Cites 14 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the existing fusion methods based on feature layer and decision layer are usually based on shallow network structure, which cannot automatically learn deep feature representation. Facing complex mechanical system diagnosis problems, the effect of fusion is often not ideal, and there are certain limitations.

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
  • Mechanical fault diagnosis method based on multi-sensor information fusion migration network
  • Mechanical fault diagnosis method based on multi-sensor information fusion migration network
  • Mechanical fault diagnosis method based on multi-sensor information fusion migration network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0081] This embodiment discloses a mechanical fault diagnosis method based on a multi-sensor information fusion migration network. The method uses a feature sharing layer and a plurality of convolutional neural networks to perform a diagnosis based on the historical data of multiple source domain sensors in the source domain diagnosis task. learning, and transfer the knowledge learned on the labeled source domain dataset to the unlabeled target domain task, improving the diagnostic performance of the target domain task. like figure 1 As shown, the steps are as follows:

[0082] S1. Multi-sensor data acquisition: According to the actual operating conditions of mechanical equipment, arrange source domain sensors for collecting fault data at different positions of mechanical equipment, and then use source domain sensors to obtain fault data with different sensitivities to faults at each position, Thereby forming multiple source domain datasets Among them, the source domain datas...

Embodiment 2

[0136] This embodiment discloses a mechanical fault diagnosis device based on multi-sensor information fusion migration network, such as Figure 5 shown, including:

[0137] The source domain sensor data acquisition module is used to use the source domain sensors arranged at different positions of the mechanical equipment according to the actual operating conditions of the mechanical equipment to obtain fault data with different sensitivities to faults at each position, thereby forming multiple source domain data sets Among them, the source domain dataset consists of source domain sensor data x s and its corresponding label y s Composition, the label corresponds to the health status of the source domain diagnosis task, M represents the number of source domain sensors, and also represents the number of source domain data sets, j represents the serial number of the source domain sensor, and also represents the serial number of the source domain data set;

[0138] The target d...

Embodiment 3

[0159] This embodiment discloses a storage medium, which stores a program. When the program is executed by a processor, the method for diagnosing mechanical faults based on the multi-sensor information fusion migration network described in Embodiment 1 is implemented, specifically as follows:

[0160] S1. Multi-sensor data acquisition: According to the actual operating conditions of mechanical equipment, arrange source domain sensors for collecting fault data at different positions of mechanical equipment, and then use source domain sensors to obtain fault data with different sensitivities to faults at each position, Thereby forming multiple source domain datasets Among them, the source domain dataset consists of source domain sensor data x s and its corresponding label y s Composition, the label corresponds to the health status of the source domain diagnosis task, M represents the number of source domain sensors, and also represents the number of source domain data sets, j ...

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 mechanical fault diagnosis method based on a multi-sensor information fusion migration network, and the method comprises the steps of firstly collecting the multi-sensor data, obtaining a plurality of source domain data sets and target domain data sets, and then constructing a multi-sensor information fusion migration network diagnosis model, wherein the model is providedwith a feature sharing layer and M convolutional neural networks; constructing a loss function of each convolutional neural network; training the multi-sensor information fusion migration network diagnosis model, and based on the target domain training data of the M source domain data sets and the target domain data sets, in each iteration, sequentially training the first network to the M-th network according to the sequence of the source domain sensors until the number of iterations or the classification precision is reached; and finally, inputting the target domain test data of the target domain data sets into the model, and obtaining a final classification diagnosis result through model and loss function processing and weighted average of M outputs. The method can effectively improve the mechanical fault diagnosis precision.

Description

technical field [0001] The invention relates to the technical field of mechanical fault diagnosis, in particular to a mechanical fault diagnosis method based on multi-sensor information fusion migration network. Background technique [0002] The structure of the mechanical system is complex and there are many parts. In order to monitor the status of the mechanical system, multiple sensors are usually installed at different positions of the key components of the system to obtain information of different sensitivity and complementarity. However, existing methods usually only use the information of a single sensor for fault diagnosis, and the reliability of diagnosis is affected by various factors such as sensor installation location, signal transmission path, and variable working conditions. How to effectively use and fuse multi-sensor information to improve the classification accuracy and diagnostic reliability of mechanical fault diagnosis models is a difficult problem in th...

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): G01M13/00G01M13/021G01M13/028G06N20/00G06N3/04
CPCG01M13/00G01M13/021G01M13/028G06N20/00G06N3/045
Inventor 李巍华陈祝云
Owner SOUTH CHINA UNIV OF 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
Eureka Blog
Learn More
PatSnap group products