Gas laser absorption spectrum filtering method based on convolutional neural network

A convolutional neural network and absorption spectrum technology, applied in neural learning methods, biological neural network models, neural architectures, etc., can solve problems such as KF deviation and loss of spectral signal processing capabilities, and achieve efficient online filtering, fast calculation speed, The effect of high signal-to-noise ratio

Pending Publication Date: 2022-04-22
SHANDONG UNIV
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, it shows good performance in linear systems, but in nonlinear systems, KF will have serious deviations in the spectral absorption part, and lose the ability to process spectral signals.

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
  • Gas laser absorption spectrum filtering method based on convolutional neural network
  • Gas laser absorption spectrum filtering method based on convolutional neural network
  • Gas laser absorption spectrum filtering method based on convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0024] Embodiment 1 The convolutional neural network-based gas laser absorption spectrum filtering method provided in this embodiment, the process is as follows figure 1 As shown, it specifically includes the following steps:

[0025] 1. Construct a filter convolutional neural network

[0026] Such as figure 2 As shown, the entire filtering network includes three convolutional layers, three BN layers, three pooling layers, two fully connected layers, and one dropout layer.

[0027] The activation function is chosen as Relu:

[0028]

[0029] where X is the input spectral data.

[0030] The loss function selects MSE loss:

[0031]

[0032] where y i is the predicted value of the neural network, is the tag value.

[0033] The optimization method is the momentum gradient descent method. The parameter update formula of the gradient descent method is:

[0034]

[0035]

[0036] where α is the learning rate, is the partial derivative of the cost function wit...

Embodiment 2

[0065] Embodiment 2 utilizes the method of the present invention to carry out the filtering simulation experiment of the absorption spectrum of the gas to be measured

[0066] Calculate the signal-to-noise ratio after filtering, the calculation formula is as follows:

[0067]

[0068] The simulation results are as Figure 4 and 5 as shown, Figure 4 Plots of the unfiltered absorption spectrum and the ideal spectrum. Figure 5 It is the unfiltered spectrum and the predicted spectrogram after filtering by CNN of the present invention. Figure 4 and Figure 5 The abscissa is 1111 sampling points, and the ordinate is the normalized transmitted light intensity. From Figure 5 It can be seen from the filtered output results that the method of the present invention has a good filtering effect, can effectively filter out various noises in the original spectrum, and obtain a high signal-to-noise ratio, which reaches 27.0134.

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 belongs to the technical field of signal processing, and relates to a filtering method of a gas laser absorption spectrum. The invention discloses a gas laser absorption spectrum filtering method based on a convolutional neural network. The method comprises the following steps: constructing the convolutional neural network; using the noisy spectral data as training data, using the noisy spectral data as a label, and training the constructed convolutional neural network; and inputting the acquired absorption spectrum data into the trained convolutional neural network, carrying out filtering processing, and outputting a clean absorption spectrum. Compared with the prior art, the gas laser absorption spectrum filtering method based on the convolutional neural network does not bring inherent noise of hardware and does not limit the frequency of the noise. Compared with the prior art, the method has the advantages that the nonlinear mapping capacity is high, the calculation speed is high, efficient online filtering can be achieved, and compared with an existing method, the filtered spectrum can obtain a higher signal-to-noise ratio.

Description

technical field [0001] The invention belongs to the technical field of signal processing, and relates to a filtering method of gas laser absorption spectrum Background technique [0002] Rapid and high-precision quantitative detection of trace gases is of great significance in many research fields such as biomedical research, hydrological research, atmospheric science, and resource exploration. Laser absorption spectroscopy (LAS), as an effective method for detecting trace gas concentration, has been widely used in isotope abundance analysis. However, the measured gas absorption spectrum data are usually polluted by various noises. In addition to the well-known ambient temperature change, the electrical noise in the system is unstable and can distort the effective absorption spectrum and affect the detection sensitivity. A. Chabuda, P. Durka, and J. "High frequencySSVEP-BCI with hardware stimuli control and phase-synchronized comb filter," IEEE T.Neur.Sys.Reh26(2), 344–35...

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
IPC IPC(8): G01N21/31G06N3/04G06N3/08
CPCG01N21/31G06N3/08G06N3/045
Inventor 张飒飒迟庆金田遴博杨易王韬王昭赵峰榕
Owner SHANDONG UNIV
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