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

Electroencephalogram signal denoising method based on width-depth echo state network

A technology of echo state network and EEG signal, applied in neural learning methods, biological neural network models, pattern recognition in signals, etc., can solve problems such as limited application range and inability to meet noise reduction performance

Pending Publication Date: 2020-10-30
SHAANXI NORMAL UNIV
View PDF0 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

EEG signals are multi-scale, nonlinear, fluctuating and random time-series signals. The traditional echo state network only contains one reservoir, and its application range is limited, especially the data exhibits multi-scale and highly nonlinear dynamic performance. For multivariate time sequence, due to the increase of feature information, the traditional echo state network cannot meet the requirements of noise reduction performance

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
  • Electroencephalogram signal denoising method based on width-depth echo state network
  • Electroencephalogram signal denoising method based on width-depth echo state network
  • Electroencephalogram signal denoising method based on width-depth echo state network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0070] Taking 10 EEG samples with a sampling frequency of 256 Hz and a sampling time of 30 minutes, each EEG sample has 460,800 sampling points and 23 electrode channels as an example, the signal denoising based on the width-depth echo state network of this embodiment The method consists of the following steps (see figure 1 , 2 ):

[0071] (1) Select EEG signal samples

[0072] Select s EEG signal samples from the Physionet database, and s in this embodiment is 10, as the output of the depth-width echo state network;

[0073] Each EEG signal sample is normalized according to the following formula:

[0074]

[0075] where x i is the sample data, where 1≤i≤10.

[0076] (2) Simulate noisy EEG signal samples

[0077] Adding noise with a signal-to-noise ratio of 0dB for the EEG signal sample is simulated into a noisy EEG signal sample. The noise in this embodiment adopts oculoelectric noise, and the noise-containing EEG sample is normalized according to formula (1). Use t...

Embodiment 2

[0125] Taking 10 EEG samples as an example, the sampling frequency is 256 Hz, the sampling time is 30 minutes, each EEG sample has 460,800 sampling points, and 23 electrode channels as an example, the signal denoising based on the width and depth echo state network in this embodiment The method consists of the following steps:

[0126] (1) Select EEG signal samples

[0127] This step is the same as in Example 1.

[0128] (2) Simulate noisy EEG signal samples

[0129] Add noise with a signal-to-noise ratio of -5dB to the EEG signal sample, and simulate it as a noisy EEG signal sample. Other steps in this step are the same as in Example 1.

[0130] (3) Divide network training set and test set

[0131] 70% of the EEG signal samples and noisy EEG samples were used as the network training set and 30% as the network test set by using the leave-out method, and there was no intersection between the test set and the training set.

[0132] (4) Build a network model

[0133] The wi...

Embodiment 3

[0161] Taking 10 EEG samples as an example, the sampling frequency is 256 Hz, the sampling time is 30 minutes, each EEG sample has 460,800 sampling points, and 23 electrode channels as an example, the signal denoising based on the width and depth echo state network in this embodiment The method consists of the following steps:

[0162] (1) Select EEG signal samples

[0163] This step is the same as in Example 1.

[0164] (2) Simulate noisy EEG signal samples

[0165] Add noise with a signal-to-noise ratio of 5dB to the EEG signal sample, and simulate it as a noisy EEG signal sample. Other steps in this step are the same as in Example 1.

[0166] (3) Divide network training set and test set

[0167] 90% of the EEG signal samples and noisy EEG samples are used as the network training set and 10% as the network test set by using the leave-out method, and there is no intersection between the test set and the training set.

[0168] (4) Build a network model

[0169] The width...

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 an electroencephalogram signal denoising method based on a width-depth echo state network. The electroencephalogram signal denoising method comprises the steps of selecting anelectroencephalogram signal sample, simulating a noisy electroencephalogram signal sample, dividing a network training set and a test set, constructing a network model, training the network model andverifying the test set. According to the method, the echo state network is used, only the output weight Wout is calculated in the learning process, training parameters are few, implementation is easy,and the complexity of linear combination and the noise reduction performance of electroencephalogram signals are improved by increasing the number of storage pools; a topological structure of width and depth is adopted, the feature extraction capacity of the reservoir is improved, more useful information is reserved in the feature extraction process, the multi-scale dynamic state of time series data is captured, and more complex features are extracted. The method has the advantages of being high in noise reduction performance, few in training parameters, easy to implement, capable of keepingthe nonlinear characteristics of original electroencephalogram signals and the like, and can be used for preprocessing and denoising signals.

Description

technical field [0001] The invention belongs to the technical field of electroencephalogram signal processing, and in particular relates to an electroencephalogram signal denoising method based on a width-depth echo state network. [0002] technical background [0003] Several electrodes installed on the scalp are used to record the electrical activity of the human brain, and the recording is called an EEG signal. During the acquisition process, it is extremely susceptible to various noise interference such as baseline drift, myoelectric signal, and oculoelectric signal, and eye blinking artifacts. Very common in EEG signals, the low-frequency and high-amplitude signals they produce are much larger than EEG signals. The superposition of these unwanted signals seriously damages EEG signals, resulting in reduced accuracy of EEG feature extraction and affecting subsequent research. Due to the complex time-frequency domain characteristics and unknown distribution of part of the n...

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): G06K9/00G06N3/04G06N3/08
CPCG06N3/08G06N3/045G06F2218/04G06F2218/12
Inventor 吴晓军孙维彤苏玉萍
Owner SHAANXI NORMAL UNIV
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