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

A fiber grating wavelength demodulation method and device based on deep learning

A fiber grating and wavelength demodulation technology, applied in biological neural network models, design optimization/simulation, neural architecture, etc., can solve problems such as slow training speed, cumbersome neural network parameter selection, local optimization and overfitting, etc. Achieve the effect of improving accuracy, effective high-speed high-precision demodulation, and high flexibility

Active Publication Date: 2021-08-03
WUHAN UNIV OF TECH
View PDF3 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] Traditional neural networks have problems such as cumbersome parameter selection, slow training speed, easy to fall into local optimum and overfitting, especially in deep neural networks, these problems are more prominent when the number of layers of the network increases

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
  • A fiber grating wavelength demodulation method and device based on deep learning
  • A fiber grating wavelength demodulation method and device based on deep learning
  • A fiber grating wavelength demodulation method and device based on deep learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0071] please see figure 1 , a kind of fiber grating demodulation method based on deep learning provided by the present invention comprises the following steps:

[0072] Step 01: Obtain the wave peak data of the fiber grating wavelength demodulator, fit these data with traditional methods, and obtain their Gaussian function parameters as the output sample Y;

[0073] Step 02: Use the function expression determined by Gaussian parameters to randomly generate L data points as input samples in Generate m training samples according to this method;

[0074] Step 03: Define a neural network with n layers {h 1 , h 2 ,...,h n}, the output function of each layer is expressed as h i (x)=g(θ i x); where g(x) represents a nonlinear activation function, different activation functions are allowed between different neural network layers, θ i Represents the parameter matrix of the i-th layer, and x represents the input vector of neurons in the corresponding layer;

[0075] Step 04:...

Embodiment 2

[0092] This embodiment belongs to the device embodiment, and belongs to the same technical concept as the method embodiment in the above-mentioned embodiment 1. For the content not described in detail in this embodiment, please refer to the method embodiment 1.

[0093] Such as figure 2 As shown, a kind of fiber grating wavelength demodulation device based on deep learning described in the present invention comprises:

[0094] An analysis unit is used to analyze the peak data received by the fiber grating wavelength demodulator and generate training samples;

[0095] an initialization unit for initializing the neural network;

[0096] The first processing unit is used to collect training samples and perform BatchNormalization processing on each layer of neurons;

[0097] The second processing unit is used to perform Dropout processing on each layer of neurons, and introduce the L2 norm for regularization, and use the stochastic gradient descent method to update the network we...

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 fiber grating wavelength demodulation method and device based on deep learning, and specifically proposes a peak finding method for fitting wave peak data obtained by fiber gratings based on a deep neural network. Combining BatchNormalization, Dropout and L2 norm regularization technologies, it solves the problems of slow training speed, cumbersome parameter modification, easy to fall into local optimum and overfitting in deep neural network. The invention realizes a high-speed and high-precision peak-seeking algorithm for the wave peak data of the optical fiber grating wavelength demodulator.

Description

technical field [0001] The invention relates to the field of a fiber grating wavelength demodulation method, in particular to a method and device for analyzing wave peak data in a fiber grating wavelength demodulator using deep learning and performing peak finding. Background technique [0002] With the development of modern industry, various sensing and detection fields have higher requirements for detection accuracy, speed, reliability and equipment cost, but some current related fiber grating sensing and demodulation products cannot fully meet the requirements. Only by improving the demodulation speed and accuracy of the FBG demodulator can the state monitoring of complex mechanical systems be better realized. How to realize high-speed and high-precision FBG demodulation has become one of the bottlenecks affecting its extensive development. [0003] Traditional demodulation methods have problems such as insufficient demodulation speed and low precision, especially when 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 Patents(China)
IPC IPC(8): G06F30/27G06N3/04G06N3/06
CPCG06N3/061G06F30/20G06N3/045
Inventor 邹承明张天柱柳星姜德生
Owner WUHAN UNIV OF 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

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