CN113419902A - Multichannel electroencephalogram signal correlation analysis and data recovery method based on long-time and short-time memory network - Google Patents
Multichannel electroencephalogram signal correlation analysis and data recovery method based on long-time and short-time memory network Download PDFInfo
- Publication number
- CN113419902A CN113419902A CN202110723809.7A CN202110723809A CN113419902A CN 113419902 A CN113419902 A CN 113419902A CN 202110723809 A CN202110723809 A CN 202110723809A CN 113419902 A CN113419902 A CN 113419902A
- Authority
- CN
- China
- Prior art keywords
- data
- channel
- time
- correlation
- long
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 49
- 230000015654 memory Effects 0.000 title claims abstract description 34
- 238000011084 recovery Methods 0.000 title claims abstract description 22
- 238000010219 correlation analysis Methods 0.000 title claims abstract description 17
- 230000002159 abnormal effect Effects 0.000 claims abstract description 39
- 238000012360 testing method Methods 0.000 claims abstract description 24
- 238000012549 training Methods 0.000 claims abstract description 23
- 238000013145 classification model Methods 0.000 claims abstract description 22
- 230000009466 transformation Effects 0.000 claims abstract description 17
- 238000000605 extraction Methods 0.000 claims abstract description 16
- 238000001914 filtration Methods 0.000 claims abstract description 16
- 238000012706 support-vector machine Methods 0.000 claims abstract description 14
- 238000007500 overflow downdraw method Methods 0.000 claims abstract description 13
- 238000012937 correction Methods 0.000 claims abstract description 9
- 238000007781 pre-processing Methods 0.000 claims abstract description 8
- 238000005070 sampling Methods 0.000 claims abstract description 7
- 238000013528 artificial neural network Methods 0.000 claims description 18
- 230000006870 function Effects 0.000 claims description 15
- 238000009826 distribution Methods 0.000 claims description 12
- 238000002474 experimental method Methods 0.000 claims description 11
- 238000004458 analytical method Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 8
- 238000004422 calculation algorithm Methods 0.000 claims description 7
- 230000008569 process Effects 0.000 claims description 6
- 230000000306 recurrent effect Effects 0.000 claims description 6
- 230000006403 short-term memory Effects 0.000 claims description 5
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 4
- 238000012935 Averaging Methods 0.000 claims description 4
- 238000012952 Resampling Methods 0.000 claims description 4
- 230000008859 change Effects 0.000 claims description 4
- 238000013135 deep learning Methods 0.000 claims description 4
- 230000003993 interaction Effects 0.000 claims description 4
- 230000002452 interceptive effect Effects 0.000 claims description 4
- 238000012417 linear regression Methods 0.000 claims description 4
- 238000011160 research Methods 0.000 claims description 4
- GNFTZDOKVXKIBK-UHFFFAOYSA-N 3-(2-methoxyethoxy)benzohydrazide Chemical compound COCCOC1=CC=CC(C(=O)NN)=C1 GNFTZDOKVXKIBK-UHFFFAOYSA-N 0.000 claims description 3
- 230000021615 conjugation Effects 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000007619 statistical method Methods 0.000 claims description 3
- FGUUSXIOTUKUDN-IBGZPJMESA-N C1(=CC=CC=C1)N1C2=C(NC([C@H](C1)NC=1OC(=NN=1)C1=CC=CC=C1)=O)C=CC=C2 Chemical compound C1(=CC=CC=C1)N1C2=C(NC([C@H](C1)NC=1OC(=NN=1)C1=CC=CC=C1)=O)C=CC=C2 FGUUSXIOTUKUDN-IBGZPJMESA-N 0.000 claims description 2
- 230000002547 anomalous effect Effects 0.000 claims description 2
- 230000003111 delayed effect Effects 0.000 claims description 2
- 238000012545 processing Methods 0.000 abstract description 9
- 238000005457 optimization Methods 0.000 abstract description 3
- 210000004027 cell Anatomy 0.000 description 4
- 230000007787 long-term memory Effects 0.000 description 3
- 238000003825 pressing Methods 0.000 description 3
- 230000005856 abnormality Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000004904 shortening Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 210000004556 brain Anatomy 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 230000005672 electromagnetic field Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 230000008054 signal transmission Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/14—Error detection or correction of the data by redundancy in operation
- G06F11/1402—Saving, restoring, recovering or retrying
- G06F11/1446—Point-in-time backing up or restoration of persistent data
- G06F11/1448—Management of the data involved in backup or backup restore
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/011—Arrangements for interaction with the human body, e.g. for user immersion in virtual reality
- G06F3/015—Input arrangements based on nervous system activity detection, e.g. brain waves [EEG] detection, electromyograms [EMG] detection, electrodermal response detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computing Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Quality & Reliability (AREA)
- Dermatology (AREA)
- Neurology (AREA)
- Neurosurgery (AREA)
- Human Computer Interaction (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention discloses a multi-channel electroencephalogram signal correlation analysis and data recovery method based on a long-time and short-time memory network, and belongs to the field of brain-computer interface data processing. Collecting required data by using brain-computer interface equipment; preprocessing an electroencephalogram signal: reducing sampling frequency, performing baseline drift correction and band-pass filtering by utilizing wavelet transformation, and positioning data with abnormal positions; acquiring correlation information between channels by using a time-frequency co-fusion method; inputting the channel with the abnormal data and the related channel into a long-time memory network to obtain a data prediction value of the abnormal position; performing feature extraction by using a common space mode; carrying out classification model training by using a support vector machine, inputting a test set into a classification model, and obtaining classification accuracy data; and analyzing results, and obtaining the classification accuracy results of the recovered data and the data which is not recovered according to the result information. The method has important reference significance for the aspects of electroencephalogram signal real-time processing optimization and electroencephalogram signal data recovery.
Description
Technical Field
The invention relates to a multi-channel electroencephalogram signal correlation analysis and data recovery method based on a long-time and short-time memory network, which is applied to the field of brain-computer interface data processing.
Background
The brain-computer interface is used as a new form of control mode, can provide remote control equipment for technicians working in dangerous scenes, and can also detect the working state of workers to avoid accidents caused by fatigue work. In the non-invasive BCI system, because the electrode is far away from the neuron, the proportion of a useful signal reflecting the subjective action consciousness of a user in the obtained electroencephalogram data to EEG background and other interference noise is small, namely the signal to noise ratio is small, so that all information needs to be kept as far as possible, and the signal to noise ratio can be further improved.
In the actual production process of an enterprise, a large number of interference sources including an external circuit, weather conditions, an environmental electromagnetic field, man-made interference and the like exist, and the data transmission precision of an electronic instrument and the operation stability inside the instrument are seriously influenced under the condition of not carrying out effective processing. The noise environment in a factory interferes with the acquisition of the electroencephalogram signals, and the data are deviated or even lost. The electroencephalogram signals are typical non-stationary random signals, the time when important electroencephalogram signal characteristics appear is extremely short, and if information loss or interference occurs in individual important channels at the moment, the accuracy of subsequent electroencephalogram signal processing is greatly influenced. Therefore, how to recover the information of the abnormal channel is very important.
Disclosure of Invention
The invention aims to provide a multi-channel electroencephalogram correlation analysis and data recovery method based on a long-time memory network, aiming at the problem of electroencephalogram signal acquisition in complex environments such as factories and vehicles. The method solves the difficulty that the disturbed or lost signal data is recovered without an effective method. In practical application, the number of required relevant channels can be flexibly determined according to the performance of different platforms, the size of data volume and the requirement of calculation time, and the relevant channels are sequentially selected as input data of a long-time memory network from high to low according to the calculated correlation degree.
In order to achieve the purpose, the invention adopts the following technical scheme:
a multi-channel electroencephalogram signal correlation analysis and data recovery method based on a long-time and short-time memory network comprises the following operation steps:
a. collecting required data by utilizing brain-computer interface equipment;
b. preprocessing an electroencephalogram signal: reducing sampling frequency, performing baseline drift correction and band-pass filtering by utilizing wavelet transformation, and positioning abnormal positions of data;
c. acquiring correlation information between channels by using a time-frequency co-fusion method;
d. inputting the channel with the abnormal data and the related channel into a long-time memory network to obtain a data prediction value of the abnormal position;
e. performing feature extraction by using a common space mode;
f. carrying out classification model training by using a support vector machine, inputting a test set into a classification model, and obtaining classification accuracy data;
g. and analyzing results, and respectively obtaining classification accuracy results of the recovered data and the data which is not recovered according to the result information.
Preferably, in the step a, the brain-computer interface device is used for collecting required data, an experimental scheme is formulated, the device is debugged, and the brain-computer interface device is worn by a tester; linking computers to prepare experiments;
designing a motor imagery experiment with an operation numerical control machine tool as a background; the experimental original EEG data is collected by 64-lead equipment, and the sampling rate is 1000 Hz; the experimental contents are that the buttons of the numerical control machine tool operation panel are pressed by the left hand and the right hand, the emergency stop button on the left side of the operation panel is pressed by the left hand, and the cycle start button of the operation panel is pressed by the right hand; one complete experiment is one block, and each experimenter completes 6 blocks in 3 days; the experimental data takes blocks as a unit, each block is continuously acquired EEG data and comprises 20 random motor imagery tasks of left and right hands, each imagination is a trial, the trial is called a trial, and the time length of each trial is 7.5 seconds; in a single block, the left hand and the right hand imagine the tasks 10 times respectively; when the experimental trial begins, firstly entering a target prompting stage, wherein a prompt tone of 'Beep' is set, and meanwhile, one button is highlighted on a screen to prompt a subject that the motor imagery task of the trial is a left hand or a right hand for 1.5 seconds; then, in a motor imagery stage, the button can keep a normally-on state in the motor imagery process, and the subject starts to perform left-hand or right-hand motor imagery; after the motor imagery stage is finished, the subject has a rest time of 2 seconds; a marking signal is recorded at the beginning and the end of the motor imagery of a single trial, and is used as an identification for arranging experimental data.
Preferably, in the step b, the acquired electroencephalogram data is preprocessed: resampling and reducing the frequency of all the acquired electroencephalogram data according to the requirement; then, performing baseline drift correction by using wavelet transformation, and selecting a filtering passband for filtering according to research content; the location of the anomalous data is located.
Preferably, in the step c, the time-frequency co-fusion method obtains the correlation information as follows:
integrating time and frequency components of the electroencephalogram signal; using a time-frequency analysis method based on wavelet transformation; performing wavelet transformation on the preprocessed data to obtain time-frequency power information of each channel and each frequency, and then averaging the power of the frequency band selected by each channel to obtain the change condition of the power on a specific frequency band along with time; the corresponding wavelet transform:
wherein,represents the power density of the ith channel at time t at frequency f; λ is a wavelet parameter; phi is at,f(x) In order to be a basis function of the wavelet,whereinIs smallComplex conjugation of wave basis functions;
calculating linear and nonlinear correlation between any two channels by using an interactive information method; taking the average power of each channel as a random variable, and calculating interaction information through entropy and joint entropy; by random variables FiRepresenting the average power signal of the ith channel by ps(Fi,b) Representing the probability density function of the average power signal of the ith channel at the block b; fiEntropy, i.e. the average amount of information reflecting its uncertainty, in H (F)i) Represents; h (F)i) Expressed as:
where b is 1, …,50 denotes a block index used to construct the approximate probability density function to avoid underestimating the entropy of larger samples and overestimating the entropy of smaller samples; its joint entropy H (F)i,Fj) Expressed as:
in the formula, pc(Fi,b,Fj,b) A joint probability density function representing the average power signal of the ith channel and the average power of the jth channel at block b;
the time-frequency co-fusion method for calculating the two random channels is as follows:
the time-frequency co-fusion value (TFCMI) is an index that evaluates the relationship between two channels based on their average power variation and signal-to-noise ratio over a selected frequency band; and obtaining the relation between each channel through the time-frequency co-fusion value, wherein the higher the value is, the higher the correlation is.
Preferably, in the step d, a channel with an abnormal data and a related channel are input into a long-term memory network, and a channel through which the data is recovered is acquired:
after the correlation information of the abnormal channel is collected, acquiring data of the channel with high correlation and data of the abnormal channel; taking data in a period of time before the abnormal point and normal data of a channel related to the abnormal point at the same time as the abnormal point as prediction characteristics together; inputting data before the missing point into a Long Short Term memory Network (LSTM) Network for prediction, and inputting a prediction result and normal data of a related channel into a Neural Network (NN) for prediction to obtain a prediction value of the abnormal point;
the LSTM network is one of deep learning networks, and the established LSTM network consists of an input layer, an LSTM unit layer, a Dropout layer, a full connection layer and a linear regression layer; the NN network consists of an input layer, a ReLU layer, a full connection layer and a regression layer; wherein, the LSTM unit layer of the LSTM network consists of 1000 LSTM units, and the discarding rate of the Dropout layer is 50%; the LSTM network trains for 200 rounds, the initial learning rate is 0.01, and the learning rate is reduced by half in every 50 rounds of training; the gradient threshold value is 1, and the optimizer is an adam optimizer;
LSTM is used to process and predict significant events in time series that are very long-spaced and delayed, and generally performs better than recurrent neural networks and hidden markov models HMMs; the LSTM network is one of the recurrent neural networks RNN, which is reliable for long-time sequence prediction; the key of the LSTM network is to Forget a gate Forget gate, then an Input gate, and finally an Output gate; the forgetting gate determines what is discarded from the cell state, the input gate determines to put new information in the cell state, and the output gate determines the output state at this time.
Preferably, in the step e, feature extraction is performed by using a common space mode:
the CSP is a space domain filtering feature extraction algorithm under two classification tasks, and can extract space distribution features of each type from multi-channel brain-computer interface data; the basic principle of the common space mode algorithm is that a group of optimal space filters are found for projection by utilizing diagonalization of a matrix, so that the variance value of two types of signals reaches the maximum difference, and the feature vector with higher discrimination is obtained.
Preferably, in step f, a support vector machine is used to perform classification model training, and the test set is input into a classification model to obtain classification accuracy data:
dividing the recovered channel data into a training set and a test set, inputting the training set features subjected to CSP feature extraction into a support vector machine for classification model training; and inputting the test set into a classification model to obtain classification accuracy.
Preferably, in the step g, result analysis is performed, and according to result information, classification accuracy results of the recovered data and the data that is not recovered are respectively obtained:
according to the experimental result, obtaining the classification accuracy results of the recovered data and the data which are not recovered, specifically comprising a recovered electroencephalogram signal data graph and a classification accuracy comparison table;
if the original correct data exist, in order to evaluate the effectiveness of a multi-channel electroencephalogram correlation analysis and data recovery method based on a long-time and short-time memory network, three performance indexes, namely, correlation of a Root Mean Square Error (RMSE), an absolute error (MAE) and a spearman grade are used; the root mean square error RMSE, absolute error MAE are defined as follows:
wherein m is the number of samples, ytestIn order to test the actual value of the set,predicting values for the test set;
the spearman grade correlation belongs to a non-parameter statistical method, and does not require the distribution of original variables; the method is suitable for the data which do not follow normal distribution, the data which are unknown in overall distribution and represented by the original data by grades, and the electroencephalogram signal calculation formula is as follows:
wherein d isq=(xq-yq),xqAnd yqThe rank of two variables in order according to size is respectively, and n is the sample capacity;
the value range of the Spearman grade correlation coefficient is as follows: [ -1,1]The larger the absolute value is, the stronger the correlation is; r issIf the number is positive, the correlation of positive grades is considered to exist; r issIf the number is negative, the negative rank correlation is considered to exist; r issWhen the value is 1, the two variables are completely same in grade, and complete positive correlation exists; and vice versa.
Compared with the prior art, the invention has the following obvious prominent substantive characteristics and remarkable advantages:
1. the multichannel electroencephalogram signal correlation analysis and data recovery method based on the long-time memory network solves the problem that data in the acquisition phase of the electroencephalogram signal in practical application cannot be recovered due to loss, and fills up the blank of research in related fields;
2. in practical application, the method flexibly determines the number of the required channels according to the performance of different platforms, the size of data volume and the requirement of calculation time, and selects the relevant channels as input data in turn according to the relevance from high to low, thereby achieving the purposes of shortening the calculation time and improving the stability of the system.
Drawings
FIG. 1 is a block flow diagram of the process of the present invention.
FIG. 2 is a process of the correlation analysis, long and short term memory network, of the present invention.
Fig. 3 is an architecture diagram of a long-term memory network according to the present invention.
FIG. 4 is a graph comparing recovered data with correct data according to the present invention.
Detailed Description
The preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings:
the first embodiment is as follows:
in this embodiment, referring to fig. 1, a method for analyzing correlation between multi-channel electroencephalograms and recovering data based on a long-time and short-time memory network includes the following steps:
a. and collecting required data by utilizing a brain-computer interface device.
b. Preprocessing an electroencephalogram signal: and reducing the sampling frequency, and performing baseline drift correction, band-pass filtering and positioning data of abnormal positions by using wavelet transformation.
c. And acquiring the correlation information between channels by using a time-frequency co-fusion method.
d. And inputting the channel with the abnormal data and the related channel into a long-time memory network to obtain a data prediction value of the abnormal position.
e. And performing feature extraction by using a common space mode.
f. And (4) carrying out classification model training by using a support vector machine, inputting the test set into a classification model, and obtaining classification accuracy data.
g. And analyzing results, and respectively obtaining classification accuracy results of the recovered data and the data which is not recovered according to the result information.
The method solves the difficulty that the abnormal channel data is lost and cannot be repaired after being interfered, and simultaneously avoids the influence of interference information contained in the channel irrelevant to the task on the prediction result.
Example two:
this embodiment is substantially the same as the first embodiment, and is characterized in that:
in this embodiment, referring to fig. 1, a method for analyzing correlation between multi-channel electroencephalogram signals and recovering data based on a long-time and short-time memory network includes the following operation steps:
a. acquiring required data by utilizing a brain-computer interface device:
and (5) establishing an experimental scheme. And the debugging equipment is used for wearing brain-computer interface equipment for a tester. Linking computers, and preparing experiments.
b. Preprocessing an electroencephalogram signal:
preprocessing the acquired electroencephalogram data: resampling and reducing the frequency of all the acquired electroencephalogram data according to the requirement; then, performing baseline drift correction by using wavelet transformation, and selecting a filtering passband for filtering according to research content; locating the position of the abnormal data;
c. obtaining relevant information by a time-frequency co-fusion method:
integrating time and frequency components of the electroencephalogram signal; using a time-frequency analysis method based on wavelet transformation; performing wavelet transformation on the preprocessed data to obtain time-frequency power information of each channel and each frequency, and then averaging the power of the frequency band selected by each channel to obtain the change condition of the power on a specific frequency band along with time; corresponding wavelet transform
Wherein,represents the power density of the ith channel at time t at frequency f; λ is a wavelet parameter; phi is at,f(x) In order to be a basis function of the wavelet,whereinIs the complex conjugation of wavelet basis functions;
calculating linear and nonlinear correlation between any two channels by using an interactive information method; taking the average power of each channel as a random variable, and calculating interaction information through entropy and joint entropy; by random variables FiRepresenting the average power signal of the ith channel by ps(Fi,b) Representing the probability density function of the average power signal of the ith channel at the block b; fiEntropy, i.e. the average amount of information reflecting its uncertainty, ofH(Fi) Represents; h (F)i) Expressed as:
where b is 1, …,50 denotes a block index used to construct the approximate probability density function to avoid underestimating the entropy of larger samples and overestimating the entropy of smaller samples; its joint entropy H (F)i,Fj) Expressed as:
in the formula, pc(Fi,b,Fj,b) A joint probability density function representing the average power signal of the ith channel and the average power of the jth channel at block b;
the time-frequency co-fusion method for calculating the two random channels is as follows:
the time-frequency co-fusion value (TFCMI) is an index that evaluates the relationship between two channels based on their average power variation and signal-to-noise ratio over a selected frequency band; obtaining the relation between each channel through the time-frequency co-fusion value, wherein the higher the value is, the higher the correlation is;
d. inputting the channel with the abnormality of the related channel and the data into a long-time memory network, and acquiring the channel of the data after recovery:
after the correlation information of the abnormal channel is collected, the data of the channel with high correlation and the data of the abnormal channel are obtained. And taking data in a period of time before the missing point and normal data of channels related to the missing point at the same time as the predicting characteristic together. Inputting data before the missing point into a Long Short Term memory Network (LSTM) Network for prediction, and inputting a prediction result and normal data of a related channel into a Neural Network (NN) for prediction to obtain a prediction value of the missing point.
The LSTM network is one type of deep learning network. The established LSTM network consists of an input layer, an LSTM unit layer, a Dropout layer, a full connection layer and a linear regression layer. The NN network consists of an input layer, a ReLU layer, a full connection layer and a regression layer. Wherein, the LSTM unit layer of the LSTM network consists of 1000 LSTM units, and the discarding rate of the Dropout layer is 50%. The LSTM network is trained for 200 rounds, the initial learning rate is 0.01, and the learning rate is reduced by half every 50 rounds of training. The gradient threshold is 1 and the optimizer is an adam optimizer.
LSTM is suitable for processing and predicting significant events of very long intervals and delays in time series, and generally performs better than recurrent neural networks and Hidden Markov Models (HMMs). The LSTM network is one of the Recurrent Neural Networks (RNNs), which has proven reliable for long-term sequence prediction in many studies. The key to the LSTM network is the forgetting gate (Forget gate), followed by the Input gate (Input gate), and the Output gate (Output gate) the first time. The forgetting gate determines what is discarded from the cell state, the input gate determines to put new information in the cell state, and the output gate determines the output state at this time.
e. And (3) common spatial mode feature extraction:
the common space mode (CSP) is an algorithm for extracting spatial filtering features under two classification tasks, and can extract spatial distribution features of each class from multi-channel brain-computer interface data; the common space mode algorithm has the basic principle that a group of optimal space filters are found for projection by utilizing diagonalization of a matrix, so that the variance value of two types of signals reaches the maximum difference, and the characteristic vector with higher discrimination is obtained;
f. and (3) support vector machine feature classification:
dividing the preprocessed data into a training set and a test set, inputting the training set features after CSP feature extraction into a support vector machine for classification model training; inputting the test set into a classification model to obtain classification accuracy;
g. and (4) analyzing results:
and obtaining the classification accuracy results of the recovered data and the data which is not recovered according to the experimental result. The method specifically comprises a recovered electroencephalogram data graph and a classification accuracy comparison table.
In the embodiment, the method for analyzing the correlation of the multi-channel electroencephalogram signals and recovering the data based on the long-time and short-time memory network selects the required data channel in a targeted manner, and inputs the related data channel and the abnormal data channel into the long-time and short-time memory network together to obtain the recovered data. Performing feature extraction by adopting a common space mode, and performing feature classification by using a support vector machine; in practical application, the method flexibly determines the number of the required channels according to the performance of different platforms, the size of data volume and the requirement of calculation time, and selects the relevant channels as input data in sequence according to the relevance from high to low, thereby achieving the purposes of shortening the calculation time and improving the stability of the system.
Example three:
this embodiment is substantially the same as the above embodiment, and is characterized in that:
in this embodiment, referring to fig. 1, a method for analyzing correlation between multi-channel electroencephalogram signals and recovering data based on a long-time and short-time memory network includes the following steps:
a. collecting required data by brain-computer interface equipment
And (5) establishing an experimental scheme. And the debugging equipment is used for wearing brain-computer interface equipment for a tester. Linking computers, and preparing experiments.
A motor imagery experiment with the background of operating a numerical control machine tool is designed. Experimental raw EEG data was collected by a 64-lead device with a sample rate of 1000 Hz. The experimental contents are that the left hand and the right hand are used for pressing the buttons of the numerical control machine tool operation panel, the left hand is used for pressing the emergency stop button on the left side of the operation panel, and the right hand is used for pressing the cycle start button of the operation panel. One complete experiment is one block, and each experimenter completes 6 blocks in 3 days. The experimental data takes blocks as units, each block is continuously acquired EEG data and comprises 20 left-hand and right-hand random motor imagery tasks, each imagery task is a trial, the trial is called a trial, and the time length of each trial is 7.5 seconds. In a single block, the left and right hands imagine the tasks 10 times each. When the experimental trial begins, a target prompting stage is firstly entered, in the stage, a prompt tone of 'Beep' is generated, meanwhile, one button is highlighted on a screen, the motor imagery task of the trial is prompted to be the left hand or the right hand, and the duration is 1.5 seconds. Then, in the motor imagery stage, the button is kept in a normally-on state during the motor imagery, and the subject starts to perform left-hand or right-hand motor imagery. After the motor imagery phase is over, the subject will have a 2 second rest period. Recording a marking signal at the beginning and the end of the motor imagery of a single trial time to be used as an identification for arranging experimental data;
b. electroencephalogram signal preprocessing
According to the content of the invention content a, preprocessing the acquired motor imagery electroencephalogram data; resampling the acquired electroencephalogram data and reducing the frequency to 250 Hz; performing baseline drift correction by using wavelet transformation, and performing band-pass filtering with a passband of 8-30Hz by using a second-order IIR filter based on Butterworth; carrying out 8-30Hz band-pass filtering to reserve the motor imagery signal characteristics contained in alpha wave and beta wave frequency bands and simultaneously removing most electromyographic noise interference; observing the oscillogram or finding the position of abnormal data by using a full-automatic algorithm for identifying the artifact;
c. method for acquiring correlation information by time-frequency co-fusion method
According to the content of the invention content b, analyzing the correlation among the channels by using a time-frequency co-fusion method; firstly, a time-frequency analysis method based on wavelet transformation is applied; firstly, performing wavelet transformation on preprocessed data to obtain time-frequency power information of each channel and each frequency, and then averaging the power of a frequency band selected by each channel to obtain the change condition of the power on a specific frequency band along with time;
calculating linear and nonlinear correlation between any two channels by using an interactive information method; taking the average power of each channel as a random variable, and calculating interaction information through entropy and joint entropy;
and calculating the time-frequency co-fusion value of the two random channels, wherein the higher the numerical value is, the higher the correlation degree is.
d. Inputting the channel with the abnormality of the related channel and the data into a long-time memory network, and acquiring the channel of the data after recovery:
after the correlation information of the abnormal channel is collected, the data of the channel with high correlation and the data of the abnormal channel are obtained. And taking data in a period of time before the missing point and normal data of channels related to the missing point at the same time as the predicting characteristic together. Inputting data before the missing point into a Long Short Term memory Network (LSTM) Network for prediction, and inputting a prediction result and normal data of a related channel into a Neural Network (NN) for prediction to obtain a prediction value of the missing point.
The LSTM network is one type of deep learning network. The established LSTM network consists of an input layer, an LSTM unit layer, a Dropout layer, a full connection layer and a linear regression layer. The NN network consists of an input layer, a ReLU layer, a full connection layer and a regression layer. Wherein, the LSTM unit layer of the LSTM network consists of 1000 LSTM units, and the discarding rate of the Dropout layer is 50%. The LSTM network is trained for 200 rounds, the initial learning rate is 0.01, and the learning rate is reduced by half every 50 rounds of training. The gradient threshold is 1 and the optimizer is an adam optimizer.
e. Common spatial mode feature extraction
According to the content of the invention content d, a common space mode is adopted for feature extraction; if the original correct data exist, dividing the preprocessed data into two types of data sets, wherein the two types of data sets are respectively as follows:
(1) data with no exception is generated;
(2) data which are recovered through the method are abnormal;
dividing the two types of data sets into a training set and a test set respectively, and extracting features by using a common space mode;
f. support vector machine feature classification
According to the content of the invention content e, inputting the training set characteristics of the two types of data sets into a support vector machine for carrying out classification model training; inputting the test set into a classification model to obtain the classification accuracy of two data sets;
g. analysis of results
Three performance indicators, Root Mean Square Error (RMSE), absolute error (MAE), and Spearman Rank Correlation (Spearman Rank Correlation), were used to evaluate the performance of the data recovery model in the presence of the original correct data. The results are shown in FIG. 4. In the embodiment, a motor imagery electroencephalogram data processing experiment is designed by utilizing a multi-channel electroencephalogram signal correlation analysis and data recovery method based on a long-time and short-time memory network, and according to the experimental result, the data recovery method provided by the invention is applied to the environment in which wireless signal transmission is influenced, such as an industrial workshop or a vehicle-mounted system. The channel number and the channels can be selected according to the actual system performance and the real-time requirement, and the method has reference and guidance significance for real-time processing and optimization of the electroencephalogram signals.
Example four:
this embodiment is substantially the same as the above embodiment, and is characterized in that:
in this embodiment, referring to fig. 1, in the step g, result analysis is performed, and according to result information, classification accuracy results of recovered data and unrecoverable data are obtained respectively:
according to the experimental result, obtaining the classification accuracy results of the recovered data and the data which are not recovered, specifically comprising a recovered electroencephalogram signal data graph and a classification accuracy comparison table;
if the original correct data exist, in order to evaluate the effectiveness of a multi-channel electroencephalogram correlation analysis and data recovery method based on a long-time and short-time memory network, three performance indexes, namely, correlation of a Root Mean Square Error (RMSE), an absolute error (MAE) and a spearman grade are used; the root mean square error RMSE, absolute error MAE are defined as follows:
wherein m is the number of samples, ytestIn order to test the actual value of the set,predicting values for the test set;
the spearman grade correlation belongs to a non-parameter statistical method, and does not require the distribution of original variables; the method is suitable for the data which do not follow normal distribution, the data which are unknown in overall distribution and represented by the original data by grades, and the electroencephalogram signal calculation formula is as follows:
wherein d isq=(xq-yq),xqAnd yqThe rank of two variables in order according to size is respectively, and n is the sample capacity;
the value range of the Spearman grade correlation coefficient is as follows: [ -1,1]The larger the absolute value is, the stronger the correlation is; r issIf the number is positive, the correlation of positive grades is considered to exist; r issIf the number is negative, the negative rank correlation is considered to exist; r issWhen the value is 1, the two variables are completely same in grade, and complete positive correlation exists; and vice versa.
To sum up, the above embodiment is based on a multi-channel electroencephalogram correlation analysis and data recovery method of a long-and-short-term memory network, and utilizes brain-computer interface equipment to acquire required data; then brain electrical signal pretreatment is carried out: reducing sampling frequency, performing baseline drift correction and band-pass filtering by utilizing wavelet transformation, and positioning abnormal positions of data; acquiring correlation information between channels by using a time-frequency co-fusion method; inputting the channel with the abnormal data and the related channel into a long-time memory network to obtain a data prediction value of the abnormal position; performing feature extraction by using a common space mode; carrying out classification model training by using a support vector machine, inputting a test set into a classification model, and obtaining classification accuracy data; and analyzing results, and respectively obtaining classification accuracy results of the recovered data and the data which is not recovered according to the result information. The embodiment has remarkable innovativeness and feasibility, and has important reference significance for real-time processing optimization of the electroencephalogram signals and data recovery of the electroencephalogram signals.
The above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.
Claims (8)
1. A multi-channel electroencephalogram signal correlation analysis and data recovery method based on a long-time and short-time memory network is characterized by comprising the following operation steps:
a. collecting required data by utilizing brain-computer interface equipment;
b. preprocessing an electroencephalogram signal: reducing sampling frequency, performing baseline drift correction and band-pass filtering by utilizing wavelet transformation, and positioning abnormal positions of data;
c. acquiring correlation information between channels by using a time-frequency co-fusion method;
d. inputting the channel with the abnormal data and the related channel into a long-time memory network to obtain a data prediction value of the abnormal position;
e. performing feature extraction by using a common space mode;
f. carrying out classification model training by using a support vector machine, inputting a test set into a classification model, and obtaining classification accuracy data;
g. and analyzing results, and respectively obtaining classification accuracy results of the recovered data and the data which is not recovered according to the result information.
2. The multi-channel electroencephalogram signal correlation analysis and data recovery method based on the long-time and short-time memory network as claimed in claim 1, which is characterized in that: in the step a, the brain-computer interface equipment is used for collecting required data, an experimental scheme is formulated, the equipment is debugged, and the brain-computer interface equipment is worn by a tester; linking computers to prepare experiments;
designing a motor imagery experiment with an operation numerical control machine tool as a background; the experimental original EEG data is collected by 64-lead equipment, and the sampling rate is 1000 Hz; the experimental contents are that the buttons of the numerical control machine tool operation panel are pressed by the left hand and the right hand, the emergency stop button on the left side of the operation panel is pressed by the left hand, and the cycle start button of the operation panel is pressed by the right hand; one complete experiment is one block, and each experimenter completes 6 blocks in 3 days; the experimental data takes blocks as a unit, each block is continuously acquired EEG data and comprises 20 random motor imagery tasks of left and right hands, each imagination is a trial, the trial is called a trial, and the time length of each trial is 7.5 seconds; in a single block, the left hand and the right hand imagine the tasks 10 times respectively; when the experimental trial begins, firstly entering a target prompting stage, wherein a prompt tone of 'Beep' is set, and meanwhile, one button is highlighted on a screen to prompt a subject that the motor imagery task of the trial is a left hand or a right hand for 1.5 seconds; then, in a motor imagery stage, the button can keep a normally-on state in the motor imagery process, and the subject starts to perform left-hand or right-hand motor imagery; after the motor imagery stage is finished, the subject has a rest time of 2 seconds; a marking signal is recorded at the beginning and the end of the motor imagery of a single trial, and is used as an identification for arranging experimental data.
3. The multi-channel electroencephalogram signal correlation analysis and data recovery method based on the long-time memory network as claimed in claim 1, wherein in the step b, the acquired electroencephalogram data are preprocessed: resampling and reducing the frequency of all the acquired electroencephalogram data according to the requirement; then, performing baseline drift correction by using wavelet transformation, and selecting a filtering passband for filtering according to research content; the location of the anomalous data is located.
4. The method for analyzing correlation and recovering data of multi-channel electroencephalogram signals based on the long-and-short-term memory network as claimed in claim 1, wherein in the step c, the method for acquiring correlation information by the time-frequency co-fusion method is as follows:
integrating time and frequency components of the electroencephalogram signal; using a time-frequency analysis method based on wavelet transformation; performing wavelet transformation on the preprocessed data to obtain time-frequency power information of each channel and each frequency, and then averaging the power of the frequency band selected by each channel to obtain the change condition of the power on a specific frequency band along with time; the corresponding wavelet transform:
wherein,represents the power density of the ith channel at time t at frequency f; λ is a wavelet parameter; phi is at,f(x) In order to be a basis function of the wavelet,wherein Is the complex conjugation of wavelet basis functions;
calculating linear and nonlinear correlation between any two channels by using an interactive information method; taking the average power of each channel as a random variable, and calculating interaction information through entropy and joint entropy; by random variables FiRepresenting the average power signal of the ith channel by ps(Fi,b) Representing the probability density function of the average power signal of the ith channel at the block b; fiEntropy, i.e. the average amount of information reflecting its uncertainty, in H (F)i) Represents; h (F)i) Expressed as:
where b is 1, …,50 denotes a block index used to construct the approximate probability density function to avoid underestimating the entropy of larger samples and overestimating the entropy of smaller samples; its joint entropy H (F)i,Fj) Expressed as:
in the formula, pc(Fi,b,Fj,b) A joint probability density function representing the average power signal of the ith channel and the average power of the jth channel at block b;
the time-frequency co-fusion method for calculating the two random channels is as follows:
the time-frequency co-fusion value (TFCMI) is an index that evaluates the relationship between two channels based on their average power variation and signal-to-noise ratio over a selected frequency band; and obtaining the relation between each channel through the time-frequency co-fusion value, wherein the higher the value is, the higher the correlation is.
5. The method for analyzing correlation of multi-channel electroencephalogram signals and recovering data based on the long-time memory network as claimed in claim 1, wherein in the step d, the channel with the abnormal correlation channel and data is input into the long-time memory network, and the recovered channel of the data is obtained:
after the correlation information of the abnormal channel is collected, acquiring data of the channel with high correlation and data of the abnormal channel; taking data in a period of time before the abnormal point and normal data of a channel related to the abnormal point at the same time as the abnormal point as prediction characteristics together; inputting data before the missing point into a Long Short Term memory Network (LSTM) Network for prediction, and inputting a prediction result and normal data of a related channel into a Neural Network (NN) for prediction to obtain a prediction value of the abnormal point;
the LSTM network is one of deep learning networks, and the established LSTM network consists of an input layer, an LSTM unit layer, a Dropout layer, a full connection layer and a linear regression layer; the NN network consists of an input layer, a ReLU layer, a full connection layer and a regression layer; wherein, the LSTM unit layer of the LSTM network consists of 1000 LSTM units, and the discarding rate of the Dropout layer is 50%; the LSTM network trains for 200 rounds, the initial learning rate is 0.01, and the learning rate is reduced by half in every 50 rounds of training; the gradient threshold value is 1, and the optimizer is an adam optimizer;
LSTM is used to process and predict significant events in time series that are very long-spaced and delayed, and generally performs better than recurrent neural networks and hidden markov models HMMs; the LSTM network is one of the recurrent neural networks RNN, which is reliable for long-time sequence prediction; the key of the LSTM network is that a forgetting gate Forgetgate is used, an input gate is used next, and an output gate is used last; the forgetting gate determines what is discarded from the cell state, the input gate determines to put new information in the cell state, and the output gate determines the output state at this time.
6. The method for analyzing correlation and recovering data of multi-channel electroencephalogram signals based on a long-and-short time memory network as claimed in claim 1, wherein in the step e, a common space mode is utilized for feature extraction:
the CSP is a space domain filtering feature extraction algorithm under two classification tasks, and can extract space distribution features of each type from multi-channel brain-computer interface data; the basic principle of the common space mode algorithm is that a group of optimal space filters are found for projection by utilizing diagonalization of a matrix, so that the variance value of two types of signals reaches the maximum difference, and the feature vector with higher discrimination is obtained.
7. The method for analyzing correlation and recovering data of multi-channel electroencephalogram signals based on the long-and-short-term memory network as claimed in claim 1, wherein in the step f, a support vector machine is utilized to carry out classification model training, a test set is input into a classification model, and classification accuracy data are obtained:
dividing the recovered channel data into a training set and a test set, inputting the training set features subjected to CSP feature extraction into a support vector machine for classification model training; and inputting the test set into a classification model to obtain classification accuracy.
8. The multi-channel electroencephalogram signal correlation analysis and data recovery method based on the long-and-short-term memory network as claimed in claim 1, wherein in the step g, result analysis is performed, and according to result information, classification accuracy results of recovered data and data which are not recovered are respectively obtained:
according to the experimental result, obtaining the classification accuracy results of the recovered data and the data which are not recovered, specifically comprising a recovered electroencephalogram signal data graph and a classification accuracy comparison table;
if the original correct data exist, in order to evaluate the effectiveness of a multi-channel electroencephalogram correlation analysis and data recovery method based on a long-time and short-time memory network, three performance indexes, namely, correlation of a Root Mean Square Error (RMSE), an absolute error (MAE) and a spearman grade are used; the root mean square error RMSE, absolute error MAE are defined as follows:
wherein m is the number of samples, ytestIn order to test the actual value of the set,predicting values for the test set;
the spearman grade correlation belongs to a non-parameter statistical method, and does not require the distribution of original variables; the method is suitable for the data which do not follow normal distribution, the data which are unknown in overall distribution and represented by the original data by grades, and the electroencephalogram signal calculation formula is as follows:
wherein d isq=(xq-yq),xqAnd yqThe rank of two variables in order according to size is respectively, and n is the sample capacity;
the value range of the Spearman grade correlation coefficient is as follows: [ -1,1]The larger the absolute value is, the stronger the correlation is; r issIf the number is positive, the correlation of positive grades is considered to exist; r issIf the number is negative, the negative rank correlation is considered to exist; r issWhen the value is 1, the two variables are completely same in grade, and complete positive correlation exists; and vice versa.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110723809.7A CN113419902A (en) | 2021-06-29 | 2021-06-29 | Multichannel electroencephalogram signal correlation analysis and data recovery method based on long-time and short-time memory network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202110723809.7A CN113419902A (en) | 2021-06-29 | 2021-06-29 | Multichannel electroencephalogram signal correlation analysis and data recovery method based on long-time and short-time memory network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113419902A true CN113419902A (en) | 2021-09-21 |
Family
ID=77717006
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202110723809.7A Pending CN113419902A (en) | 2021-06-29 | 2021-06-29 | Multichannel electroencephalogram signal correlation analysis and data recovery method based on long-time and short-time memory network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113419902A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113988400A (en) * | 2021-10-22 | 2022-01-28 | 重庆工商大学 | AQI index prediction method based on PEARSON-LSTM multi-step fusion network |
CN114533066A (en) * | 2022-04-28 | 2022-05-27 | 之江实验室 | Social anxiety assessment method and system based on composite expression processing brain network |
CN115356694A (en) * | 2022-08-26 | 2022-11-18 | 哈尔滨工业大学(威海) | Anti-impact interference method and system for high-frequency ground wave radar, computer equipment and application |
CN116269450A (en) * | 2023-03-21 | 2023-06-23 | 苏州海臻医疗器械有限公司 | Patient limb rehabilitation state evaluation system and method based on electromyographic signals |
CN116541766A (en) * | 2023-07-04 | 2023-08-04 | 中国民用航空飞行学院 | Training method of electroencephalogram data restoration model, electroencephalogram data restoration method and device |
CN118070002A (en) * | 2024-03-21 | 2024-05-24 | 北京交通大学 | Method and system for recovering track traffic multichannel vibration signal lost data |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104035563A (en) * | 2014-06-19 | 2014-09-10 | 北京工业大学 | W-PCA (wavelet transform-principal component analysis) and non-supervision GHSOM (growing hierarchical self-organizing map) based electrocardiographic signal identification method |
CN109959122A (en) * | 2019-03-11 | 2019-07-02 | 浙江工业大学 | A kind of Energy consumption forecast for air conditioning method based on shot and long term memory Recognition with Recurrent Neural Network |
CN110464334A (en) * | 2019-07-23 | 2019-11-19 | 苏州国科视清医疗科技有限公司 | A kind of electrocardiographic abnormality detection method based on composite depth learning network |
KR102096565B1 (en) * | 2018-11-08 | 2020-04-02 | 광운대학교 산학협력단 | Analysis method of convolutional neural network based on Wavelet transform for identifying motor imagery brain waves |
CN111053549A (en) * | 2019-12-23 | 2020-04-24 | 威海北洋电气集团股份有限公司 | Intelligent biological signal abnormality detection method and system |
CN112515685A (en) * | 2020-11-10 | 2021-03-19 | 上海大学 | Multi-channel electroencephalogram signal channel selection method based on time-frequency co-fusion |
-
2021
- 2021-06-29 CN CN202110723809.7A patent/CN113419902A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104035563A (en) * | 2014-06-19 | 2014-09-10 | 北京工业大学 | W-PCA (wavelet transform-principal component analysis) and non-supervision GHSOM (growing hierarchical self-organizing map) based electrocardiographic signal identification method |
KR102096565B1 (en) * | 2018-11-08 | 2020-04-02 | 광운대학교 산학협력단 | Analysis method of convolutional neural network based on Wavelet transform for identifying motor imagery brain waves |
CN109959122A (en) * | 2019-03-11 | 2019-07-02 | 浙江工业大学 | A kind of Energy consumption forecast for air conditioning method based on shot and long term memory Recognition with Recurrent Neural Network |
CN110464334A (en) * | 2019-07-23 | 2019-11-19 | 苏州国科视清医疗科技有限公司 | A kind of electrocardiographic abnormality detection method based on composite depth learning network |
CN111053549A (en) * | 2019-12-23 | 2020-04-24 | 威海北洋电气集团股份有限公司 | Intelligent biological signal abnormality detection method and system |
CN112515685A (en) * | 2020-11-10 | 2021-03-19 | 上海大学 | Multi-channel electroencephalogram signal channel selection method based on time-frequency co-fusion |
Non-Patent Citations (1)
Title |
---|
任彬: "Research on time-frequency cross mutual of motor imagination data based on multichannel EEG signal", 《HIGH TECHNOLOGY LETTERS》, no. 28, 31 March 2022 (2022-03-31), pages 21 - 29 * |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113988400A (en) * | 2021-10-22 | 2022-01-28 | 重庆工商大学 | AQI index prediction method based on PEARSON-LSTM multi-step fusion network |
CN113988400B (en) * | 2021-10-22 | 2024-04-09 | 重庆工商大学 | AQI index prediction method based on PEARSON-LSTM multi-step fusion network |
CN114533066A (en) * | 2022-04-28 | 2022-05-27 | 之江实验室 | Social anxiety assessment method and system based on composite expression processing brain network |
CN114533066B (en) * | 2022-04-28 | 2022-08-19 | 之江实验室 | Social anxiety assessment method and system based on composite expression processing brain network |
CN115356694A (en) * | 2022-08-26 | 2022-11-18 | 哈尔滨工业大学(威海) | Anti-impact interference method and system for high-frequency ground wave radar, computer equipment and application |
CN115356694B (en) * | 2022-08-26 | 2023-08-22 | 哈尔滨工业大学(威海) | High-frequency ground wave radar anti-impact interference method, system, computer equipment and application |
CN116269450A (en) * | 2023-03-21 | 2023-06-23 | 苏州海臻医疗器械有限公司 | Patient limb rehabilitation state evaluation system and method based on electromyographic signals |
CN116269450B (en) * | 2023-03-21 | 2023-12-19 | 苏州海臻医疗器械有限公司 | Patient limb rehabilitation state evaluation system and method based on electromyographic signals |
CN116541766A (en) * | 2023-07-04 | 2023-08-04 | 中国民用航空飞行学院 | Training method of electroencephalogram data restoration model, electroencephalogram data restoration method and device |
CN116541766B (en) * | 2023-07-04 | 2023-09-22 | 中国民用航空飞行学院 | Training method of electroencephalogram data restoration model, electroencephalogram data restoration method and device |
CN118070002A (en) * | 2024-03-21 | 2024-05-24 | 北京交通大学 | Method and system for recovering track traffic multichannel vibration signal lost data |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113419902A (en) | Multichannel electroencephalogram signal correlation analysis and data recovery method based on long-time and short-time memory network | |
CN112515685B (en) | Multi-channel electroencephalogram signal channel selection method based on time-frequency co-fusion | |
CN101259015B (en) | Electroencephalogram signal analyzing monitoring method and device thereof | |
CN103310113B (en) | A kind ofly to be separated and the general blood glucose prediction method of data modeling based on frequency band | |
CN113128552B (en) | Electroencephalogram emotion recognition method based on depth separable causal graph convolution network | |
CN112244873A (en) | Electroencephalogram time-space feature learning and emotion classification method based on hybrid neural network | |
CN114533086B (en) | Motor imagery brain electrolysis code method based on airspace characteristic time-frequency transformation | |
CN101515200B (en) | Target selecting method based on transient visual evoked electroencephalogram | |
CN113536882B (en) | Multi-class motor imagery electroencephalogram signal feature extraction and classification method | |
US20140025715A1 (en) | Neural Signal Processing and/or Interface Methods, Architectures, Apparatuses, and Devices | |
CN112036354A (en) | Natural action electroencephalogram recognition method based on Riemann geometry | |
CN102488517A (en) | Method and device for detecting burst suppression state in brain signal | |
CN112957055B (en) | EEMD-PCA-based method and device for removing motion artifacts in EEG signals | |
CN113017627B (en) | Depression and bipolar disorder brain network analysis method based on two-channel phase synchronization feature fusion | |
CN104490391A (en) | Combatant state monitoring system based on electroencephalogram signals | |
CN111832452B (en) | Characteristic optimization and recognition method of special gesture instruction based on brain electricity | |
CN111067513B (en) | Sleep quality detection key brain area judgment method based on characteristic weight self-learning | |
CN115964655A (en) | Method for monitoring error-related potential in brain-computer interface based on mutual information quantity | |
Shi et al. | A Novel Method of sEMG Signal Segmentation | |
CN117918863A (en) | Method and system for processing brain electrical signal real-time artifacts and extracting features | |
CN115067878A (en) | EEGNet-based resting state electroencephalogram consciousness disorder classification method and system | |
CN115270847A (en) | Design decision electroencephalogram recognition method based on wavelet packet decomposition and convolutional neural network | |
CN110516711B (en) | Training set quality evaluation method of MI-BCI system and optimization method of single training sample | |
CN114841192A (en) | Electroencephalogram signal feature enhancement method based on reinforcement learning combined denoising and space-time relation modeling | |
Murthy et al. | Design and implementation of hybrid techniques and DA-based reconfigurable FIR filter design for noise removal in EEG signals on FPGA |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |