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

Novel convolutional neural network model and application

A convolutional neural network and model technology, applied in the new convolutional neural network model and application field, can solve the problems of low diagnostic accuracy and difficulty, and achieve good classification ability

Active Publication Date: 2020-12-29
TAIYUAN FORTUCKY LOGISTICS EQUIP TECH CO LTD +1
View PDF4 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In view of the deficiencies in the prior art, the present invention solves the problem of low diagnostic accuracy caused by gradient disappearance by improving the activation function linear correction unit; two layers of residual neuron layers are added to the convolutional neural network to deepen Network depth to facilitate the extraction of potential features that are not easy to be explored; thereby constructing a new convolutional neural network model and applying it to bearing fault diagnosis

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
  • Novel convolutional neural network model and application
  • Novel convolutional neural network model and application
  • Novel convolutional neural network model and application

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0050] Such as Figure 1-3 As shown, a new convolutional neural network model is based on the convolutional neural network, the improved threshold function is used as the activation function tReLU, and residual neurons are introduced into the middle layer of the convolutional neural network. Alternate connections, softmax classification, throughout the entire structure to generate a new convolutional neural network model RLCNN; the activation function tReLU is shown in formula (1):

[0051]

[0052] The residual neuron is represented by formula (2):

[0053] F(x)=W 2 f(W 1 x+b)+b (2)

[0054] In formula (2), x represents the input of the current layer, F(x) represents the input of the next layer, W 1 , W 2 Represents the weights of the current layer and the next layer, f(.) represents the tReLU activation function, and b represents the bias.

[0055] The new convolutional neural network model is a six-layer convolutional neural network, in which the convolution kernel...

Embodiment 2

[0064] Utilize the novel convolutional neural network model of the present invention to diagnose the motor bearing fault, verify the accuracy of the RLCNN model of the present invention to the bearing fault diagnosis, the basic process of the RLCNN model fault diagnosis is as follows Figure 4 shown.

[0065] 1. Dataset

[0066] The RLCNN model of the present invention is verified by using the bearing data set of Case Western Reserve University. 3 types, including outer ring fault (OF), rolling element fault (RF), and inner ring fault (IF), each type has 3 damage degrees, and the damage diameter is 0.18mm, 0.36mm, 0.54mm, a total of 9 One fault type, plus one normal (NO) type, respectively used to represent IF0.18, IF0.36, IF0.54, OF0.18, OF0.36, OF0.54, RF0.18, RF0.36, RF0 9 types of failures of .54 bearings, NO means normal. The data were recorded under 4 load conditions (0hp, 1hp, 2hp, 3hp), with 2000 vibration images for each load in the training dataset and 400 vibrati...

Embodiment 3

[0075] Utilize the novel convolutional neural network model of the present invention to diagnose the bearing fault of the electromechanical transmission system, verify the accuracy of the RLCNN model of the present invention to the bearing fault diagnosis, and the basic flow of the RLCNN model fault diagnosis is as follows Figure 4 shown.

[0076] 1. Dataset

[0077] The application of the RLCNN model of the present invention in bearing fault diagnosis is analyzed by using the bearing data set provided by Paderborn University in Germany. Select a part of the data set for training and testing. In the data set, there are artificial damage and real damage (produced by the accelerated life experiment). In the real damage, select 5 kinds of damage in the inner ring, pitting multiple damage level 1 (KI04), There are 6 fault types including single pitting damage level 3 (KI16), repeated pitting damage level 1 (KI17), single pitting damage level 2 (KI18), single pitting damage level...

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 novel convolutional neural network model and application, and the model takes a convolutional neural network as a model basis, takes an improved threshold function as an activation function tReLU, introduces residual neurons into an intermediate layer of the convolutional neural network, and generates a novel convolutional neural network model RLCNN through employing convolution and pooling alternate connection and softmax classification and penetrating the whole structure. The activation function tReLU is shown as a formula (1). The method for carrying out bearing fault diagnosis by using the model comprises the following steps: firstly, converting a vibration signal from a bearing into a two-dimensional vibration image, processing the image to obtain a pixel intensity matrix of a grayscale image, inputting the pixel intensity matrix into a novel convolutional neural network model, and carrying out convolution by taking an existing public bearing data set asa training set, and carrying out batch normalization processing; and obtaining a diagnosis result of the bearing fault. According to the method, the problem of low bearing fault diagnosis accuracy caused by model gradient disappearance and mean shift due to a traditional activation function is solved.

Description

technical field [0001] The invention belongs to the technical field of computer convolutional neural networks, and in particular relates to a novel convolutional neural network model and its application. Background technique [0002] Modern mechanical equipment is working under complex and high-intensity working conditions. Once a failure occurs, it may be catastrophic. Therefore, mechanical failure hides huge risks and economic losses. Bearings are important parts of mechanical equipment. In order to make some equipment operate normally, the diagnosis of bearings is essential. According to statistics, 30% of them are caused by the damage of bearings. Now that the data of mechanical equipment failures is massive, the detection of mechanical equipment has entered the era of "big data". [0003] In the era of big data, how to automatically perform feature mining in massive data to replace manual feature extraction, real-time detection of bearings, and ensure the accuracy and ...

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): G06N3/04G06N3/08
CPCG06N3/082G06N3/045
Inventor 许习军王巧文董增寿康琳刘鑫
Owner TAIYUAN FORTUCKY LOGISTICS EQUIP TECH CO LTD
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

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com