CN113951898B - P300 electroencephalogram signal detection method and device for data migration, electronic equipment and medium - Google Patents

Abstract

The invention discloses a data migration P300 EEG signal detection method and device, electronic equipment, and a medium. Multi-domain features are extracted from the preprocessed target EEG signal and migration data set; the migration data set after multi-domain feature extraction is used as a training set, and the target EEG signal training data set after multi-domain feature extraction is used as a test set, using The multi-domain combination classifier based on the confidence coefficient uses the correctly classified target EEG training data set after multi-domain feature extraction as the pre-trained target EEG training set; using the pre-trained target EEG training set for multiple The domain combination classifier performs secondary training to classify the target EEG signal test data set after multi-domain feature extraction, so as to obtain the final target EEG signal P300 classification result.

Description

P300 EEG signal detection method and device for data migration, electronic equipment, medium

technical field

The present application relates to the technical field of EEG signal processing, in particular to a data migration P300 EEG signal detection method and device, electronic equipment, and media.

Background technique

The P300 EEG signal is a positive wave peak that occurs about 300 milliseconds after a person receives a "target stimulus" that occurs with a small probability in the routine irrelevant stimulus. It is a very commonly used event-related potential component. In the brain-computer interface The system is often used to control and communicate the EEG signals of external devices. It is also the main object of lie detection research such as hidden information tests, and is also commonly used to evaluate the cognitive function of the subjects. Detecting P300 components in EEG signals and distinguishing the presence or absence of P300 components are the focus of research on P300 EEG signal processing. Many studies have shown that epilepsy and other brain diseases will affect the shape of P300 components, resulting in a decrease in P300 amplitude and an increase in latency. It may cause adverse effects in brain-computer interface applications. At the same time, the amount of P300 EEG data obtained by epilepsy patients is often small, and the detection effect is limited.

Contents of the invention

The purpose of the embodiments of the present application is to provide a data migration P300 EEG signal detection method and device, electronic equipment, and media to solve the technical problems of abnormal P300 signal shape, small data volume, and difficult detection in epileptic patients.

According to the first aspect of the embodiment of the present application, a P300 EEG signal detection method for data migration is provided, including:

Multi-domain preprocessing of target EEG signal and migration datasets containing P300 components;

Extracting multi-domain features from the target EEG signal and migration data set after the multi-domain preprocessing;

The migration data set through multi-domain feature extraction is used as a training set, and the target EEG signal training data set through multi-domain feature extraction is used as a test set, and a multi-domain combined classifier based on confidence coefficient is used to classify correctly The target EEG signal training data set through multi-domain feature extraction is used as the pre-trained target EEG training set;

Use the pre-trained target EEG training set to perform secondary training on the multi-domain combination classifier, and classify the target EEG signal test data set after multi-domain feature extraction, so as to obtain the final target EEG signal P300 classification results.

Further, multi-domain preprocessing is performed on the target EEG signal and migration data sets containing P300 components, including:

Unify the sampling frequency of the target EEG signal and migration data sets containing P300 components;

Spatial preprocessing is performed on the target EEG signal and migration data sets after uniform sampling frequency by electrode selection and common average reference method;

Time-domain preprocessing is performed on the target EEG signal and migration data set after uniform sampling frequency by extremum adjustment and normalization methods;

The target EEG signal and migration data sets after unified sampling frequency are preprocessed in frequency domain by wavelet packet decomposition method.

Further, extracting multi-domain features from the multi-domain preprocessed target EEG signal and migration data set, including:

Extracting time-domain energy entropy as time-domain features from the multi-domain preprocessed target EEG signals and migration data sets respectively;

Extracting energy information of wavelet coefficients as time-frequency features from the multi-domain preprocessed target EEG signals and migration data sets respectively;

Separately use independent component analysis to extract signal spatial domain features on the target EEG signal and migration data sets that have undergone the multi-domain preprocessing.

Further, using the migration data set through multi-domain feature extraction as a training set, using the target EEG signal training data set after multi-domain feature extraction as a test set, using a multi-domain combined classifier based on confidence coefficients, The correctly classified target EEG training data set after multi-domain feature extraction is used as the pre-trained target EEG training set, including:

Use the migration data set to pre-train the multi-domain combination classifier based on the confidence coefficient to obtain a trained multi-domain combination classifier;

Using the migration data set through multi-domain feature extraction as the training set of the once-trained multi-domain combined classifier;

Using the target EEG signal training data set through multi-domain feature extraction as the test set of the multi-domain combination classifier for the one-time training;

Output the target EEG signal training data set that has been correctly classified by the multi-domain combination classifier trained once and has undergone multi-domain feature extraction, as a pre-trained target EEG training set.

Further, the confidence coefficient-based multi-domain combination classifier includes two linear discriminant analysis classifiers with different threshold constants and a naive Bayesian classifier, and the linear discriminant analysis classifier divides the feature vector X from the classification line The distance is used as the confidence coefficient, and the Naive Bayesian classifier uses the difference between the two classification probabilities as its confidence coefficient.

Further, for the two linear discriminant analysis classifiers with different threshold constants, one of the linear discriminant analysis classifiers is used as the basic component classifier, the confidence coefficient threshold is selected, and the classification results with the confidence coefficient higher than the threshold are directly accepted. For For classifying samples with confidence coefficients below the threshold, a Naive Bayesian classifier and another linear discriminant analysis classifier with different threshold constants are used as reference component classifiers.

Further, the multi-domain combination classifier is trained twice using the pre-trained target EEG training set, and the target EEG signal test data set after multi-domain feature extraction is classified to obtain the final target brain. Electrical signal P300 classification results, including:

Using the pre-trained target EEG training set to perform secondary training on the once-trained multi-domain combined classifier to obtain a second-trained multi-domain combined classifier;

Using the pre-trained target EEG training set as the training set of the multi-domain combination classifier of the secondary training;

Using the target EEG test data set through multi-domain feature extraction as the test set of the multi-domain combined classifier of the secondary training;

The classification result of the multi-domain combination classifier trained twice is output as the final target EEG signal P300 classification result.

According to the second aspect of the embodiment of the present application, a P300 EEG signal detection device for data migration is provided, including:

Preprocessing module for multi-domain preprocessing of target EEG signal and migration datasets containing P300 components;

A feature extraction module, configured to extract multi-domain features from the target EEG signal and migration data set after the multi-domain preprocessing;

The pre-training module is used to use the migration data set through multi-domain feature extraction as a training set, and use the target EEG signal training data set through multi-domain feature extraction as a test set, using multi-domain combination based on confidence coefficient A classifier, using the correctly classified target EEG training data set through multi-domain feature extraction as the pre-trained target EEG training set;

The secondary training classification module is used to use the pre-trained target EEG training set to perform secondary training on the multi-domain combination classifier, and classify the target EEG signal test data set through multi-domain feature extraction, Thereby obtaining the final target EEG signal P300 classification result.

According to a third aspect of the embodiments of the present application, there is provided an electronic device, including: one or more processors; a memory for storing one or more programs; when the one or more programs are used by the one or more executed by one or more processors, so that the one or more processors implement the method as described in the first aspect.

According to a fourth aspect of the embodiments of the present application, there is provided a computer-readable storage medium on which computer instructions are stored, wherein, when the instructions are executed by a processor, the steps of the method described in the first aspect are implemented.

The technical solutions provided by the embodiments of the present application may include the following beneficial effects:

It can be seen from the above technical solutions that this application overcomes the problem that epilepsy patients have fewer samples of EEG signals and is difficult to learn because of the introduction of the migration data set with a large amount of data, so as to achieve learning and utilization of the data distribution of the migration data set and improve the final The effect of classification accuracy. Because of the introduction of the migration dataset of the P300 signal morphology standard, it overcomes the problem of abnormal EEG signal morphology in epileptic patients, reduces the interference of poor-quality samples in the training data set to the classifier, and improves the final classification accuracy. Because the multi-domain feature extraction is used, the multi-domain features of the P300 signal can be fully extracted, and the P300 signal can be better represented. Because the multi-domain combination classifier is used, the advantages of multiple classifiers can be combined to achieve a higher total classification accuracy in less computing time.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the application.

Description of drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the application and together with the description serve to explain the principles of the application.

Fig. 1 is a flowchart showing a P300 EEG signal detection method for data migration according to an exemplary embodiment.

Fig. 2 is a flow chart of a multi-domain combination classifier based on confidence coefficients according to an exemplary embodiment.

Fig. 3 shows a block diagram of a P300 EEG signal detection device for data migration according to an exemplary embodiment.

Detailed ways

Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, the same numerals in different drawings refer to the same or similar elements unless otherwise indicated. The implementations described in the following exemplary embodiments do not represent all implementations consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with aspects of the present application as recited in the appended claims.

The terminology used in this application is for the purpose of describing particular embodiments only, and is not intended to limit the application. As used in this application and the appended claims, the singular forms "a", "the", and "the" are intended to include the plural forms as well, unless the context clearly dictates otherwise. It should also be understood that the term "and/or" as used herein refers to and includes any and all possible combinations of one or more of the associated listed items.

It should be understood that although the terms first, second, third, etc. may be used in this application to describe various information, the information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of the present application, first information may also be called second information, and similarly, second information may also be called first information. Depending on the context, the word "if" as used herein may be interpreted as "at" or "when" or "in response to a determination."

Fig. 1 is a flow chart of a P300 EEG signal detection method for data migration shown according to an exemplary embodiment. As shown in Fig. 1, the method is applied in a terminal and may include the following steps:

Step S11, performing multi-domain preprocessing on the target EEG signal and migration data set containing the P300 component;

Step S12, extracting multi-domain features from the target EEG signal and migration data set after the multi-domain preprocessing;

Step S13, using the migration data set after multi-domain feature extraction as a training set, using the target EEG signal training data set after multi-domain feature extraction as a test set, using a multi-domain combination classifier based on confidence coefficients, The correctly classified target EEG training data set after multi-domain feature extraction is used as the pre-trained target EEG training set;

Step S14, using the pre-trained target EEG training set to perform secondary training on the multi-domain combination classifier, and classify the target EEG signal test data set after multi-domain feature extraction, so as to obtain the final target brain. Electrical signal P300 classification results.

It can be seen from the above technical solutions that this application overcomes the problem that epilepsy patients have fewer samples of EEG signals and is difficult to learn because of the introduction of the migration data set with a large amount of data, so as to achieve learning and utilization of the data distribution of the migration data set and improve the final The effect of classification accuracy. Because of the introduction of the migration dataset of the P300 signal morphology standard, it overcomes the problem of abnormal EEG signal morphology in epileptic patients, reduces the interference of poor-quality samples in the training data set to the classifier, and improves the final classification accuracy. Because the multi-domain feature extraction is used, the multi-domain features of the P300 signal can be fully extracted, and the P300 signal can be better represented. Because the multi-domain combination classifier is used, the advantages of multiple classifiers can be combined to achieve a higher total classification accuracy in less computing time. The invention can effectively classify P300 electroencephalogram signals with non-standard shape and poor data quality, such as P300 signals of epileptic patients.

In the specific implementation of step S11, multi-domain preprocessing is performed on the target EEG signal and migration data set containing P300 components, and this step may include the following sub-steps:

Step S111, respectively unifying the sampling frequency of the target EEG signal and the migration data set containing the P300 component;

Specifically, use interpolation and downsampling techniques to unify the sampling frequency of the target EEG signal containing P300 components and the EEG signal of the migration data set to 500 Hz, so that the time scale of the subsequent multi-domain preprocessing and multi-domain feature extraction is consistent. Easy to operate.

Step S112, performing spatial preprocessing on the target EEG signal and the migration data set after the uniform sampling frequency by electrode selection and common average reference method;

Specifically, use the public average reference to remove the relevant noise in the EEG signal obtained in S111, and the EEG signal after the public average reference processing is the original EEG signal of the electrode minus the average of the original EEG signals of all electrodes, for n The original EEG signal of electrode channels, let the original signal be _Xi , i∈[1,n], the signal after public average reference is:

Use channel selection to select the signals of Fz, Cz, and Pz electrode channels as the main analysis object to reduce the amount of calculation.

Step S113, performing time-domain preprocessing on the target EEG signal and the migration data set after the unified sampling frequency by extremum adjustment and normalization methods;

Specifically, extreme value adjustment is performed in the time domain, and the largest and smallest 5% amplitude abnormal signal samples in the EEG signal obtained in S111 are replaced with appropriate and acceptable thresholds C1 and C2, which meet the requirements of 5% of the EEG signal. % of the samples are greater than C1, and 5% of the samples are smaller than C2; for normalization, let the EEG signal obtained by S111 be S, and the normalized signal is S', then there is

Among them, U and L are the maximum and minimum values of the projection space after normalization, which are set to 1 and -1 respectively, so that the EEG signal is normalized to the range where the minimum value is -1 and the maximum value is 1, which is convenient for follow-up calculate.

Step S114 , performing frequency-domain preprocessing on the target EEG signal and the migration data set after the uniform sampling frequency by using the wavelet packet decomposition method.

Specifically, the db4 mother wavelet is used for wavelet packet decomposition, and the EEG signal obtained in S111 is filtered to extract 0.25-30 Hz EEG signal.

In step S115 , effective signal extraction is performed on the target EEG signal and the migration data set after uniform sampling frequency through time-domain cutting.

Specifically, the effective signal 200-600 ms after the stimulation was intercepted from the EEG signal obtained at S111 to detect the P300 component and distinguish the target stimulus from the irrelevant stimulus.

In the specific implementation of step S12, multi-domain features are extracted from the target EEG signal and migration data set after the multi-domain preprocessing, and this step may include the following sub-steps:

Step S121, extracting time-domain energy entropy as time-domain features from the multi-domain preprocessed target EEG signals and migration data sets respectively;

Specifically, using the time-domain energy entropy as the time-domain feature, the EEG signal is evenly divided into 10 segments in the time domain. The negative value of the logarithm of the ratio of the energy sum is used as the time-domain feature. Let the signal of each segment be x _i , i∈[1,10], and the time-domain energy entropy E _i of each segment is:

Step S122, extracting energy information of wavelet coefficients as time-frequency features from the multi-domain preprocessed target EEG signals and migration data sets respectively;

Specifically, using the energy information of the wavelet coefficient as the time-frequency feature, using db4 as the wavelet mother function, the EEG signal is decomposed by wavelet, and the wavelet approximation coefficient corresponding to the 0-16Hz frequency band is extracted and squared to obtain its energy information as the time-frequency feature .

Step S123 , using independent component analysis to extract signal spatial domain features on the target EEG signal and migration data sets that have undergone the multi-domain preprocessing.

Specifically, independent component analysis is used to obtain spatial features, and the Infomax algorithm is used to perform independent component analysis on the signals of the three channels of Fz, Cz, and Pz to obtain a 3×3 mixing matrix, and the expanded nine features are used as spatial features.

In the specific implementation of step S13, the migration data set after multi-domain feature extraction is used as a training set, and the target EEG signal training data set after multi-domain feature extraction is used as a test set, and the confidence coefficient-based multiple The domain combination classifier uses the correctly classified target EEG training data set after multi-domain feature extraction as a pre-trained target EEG training set. Referring to Figure 2, this step may include the following sub-steps:

Step S131, using the migration data set to pre-train the multi-domain combination classifier based on the confidence coefficient to obtain a trained multi-domain combination classifier;

Specifically, the feature-extracted migration data set is used as a training set to pre-train the confidence coefficient-based multi-domain combination classifier to obtain a trained multi-domain combination classifier.

Step S132, using the once-trained multi-domain combined classifier to classify the training data of the target EEG signal, and output the correctly classified training data of the classified target EEG signal after multi-domain feature extraction as the pre-trained target EEG signal training data. electric training set.

Specifically, the target EEG signal training data set through multi-domain feature extraction is used as the test set of the multi-domain combination classifier for the one training, and the multi-domain combination classifier for one training is used to classify the test set, According to the classification result, select the correctly classified target EEG training data output after multi-domain feature extraction as the pre-trained target EEG training set.

Further, the confidence coefficient-based multi-domain combination classifier includes two linear discriminant analysis classifiers with different threshold constants and a naive Bayesian classifier, and the linear discriminant analysis classifier divides the feature vector X from the classification line The distance is used as the confidence coefficient, and the Naive Bayesian classifier uses the difference between the two classification probabilities as its confidence coefficient.

Specifically, in a linear discriminant analysis classifier, for a feature vector X, X=(x ₁ ,x ₂ ,…,x _n ) ^T , there is a linear function

Where W _L =(w ₁ ,w ₂ ,...,w _n ) ^T is an n-dimensional weight vector, and θ is a threshold constant. For binary classification problems, there is a decision mechanism: d(X)>0, X∈C ₁ , d(X)<0, X∈C ₂ .d(x)=0 is the classification boundary. The optimal W _L and θ can be calculated by Fisher's criterion, and they can be substituted into |d(X)|, which is the distance between the feature vector X and the classification line, which is also the confidence coefficient of the linear discriminant analysis classifier.

In the naive Bayesian classifier, for the feature vector X, X=(x ₁ ,x _{2 ,} …,x _n ) ^T , solve the probability P(C _i |X), the difference ΔP between P(C ₁ |X) and P(C ₂ |X) is a classification function, if ΔP is greater than 0, it is classified as C ₁ , and if ΔP is less than 0, it is classified as C ₂ . Therefore |ΔP| is the confidence coefficient of the Naive Bayesian classifier.

Further, for the two linear discriminant analysis classifiers with different threshold constants, one of the linear discriminant analysis classifiers is used as the basic component classifier, the confidence coefficient threshold is selected, and the classification results with the confidence coefficient higher than the threshold are directly accepted. For For classifying samples with confidence coefficients below the threshold, a Naive Bayesian classifier and another linear discriminant analysis classifier with different threshold constants are used as reference component classifiers.

Specifically, in order to eliminate the influence of the variability of the confidence coefficient, the confidence coefficient is normalized, and in order to improve the discrimination of the confidence coefficient, the Logistic curve adjustment is performed on the confidence coefficient.

对置信系数高于阈值的分类结果直接接受。In constructing the combined classifier, the linear discriminant analysis classifier LDA1 with the fastest speed but average classification accuracy is used as the basic component classifier, and the absolute value |d ₁ (X)| of the distance between the feature vector X in LDA1 and the classification line is used as Confidence coefficient, set the confidence coefficient threshold to the mean value of the absolute value of the distance between the feature vector X and the classification line in the training set

Classification results with confidence coefficients higher than the threshold are accepted directly.

For classified samples with confidence coefficients below the threshold, a Naive Bayesian classifier and a linear discriminant analysis classifier LDA2 with different threshold constants are used as reference component classifiers. The confidence coefficient of LDA2 is |d ₂ (X)|, and the two-class difference |ΔP| of the Naive Bayesian classifier is used as its confidence coefficient.

In order to eliminate the variability of confidence coefficients, the confidence coefficients |d ₁ (X)|, |d ₂ (X)| Curve adjustment. Taking the confidence coefficient adjusted by normalization and S-curve as the weight, the classification results of the three component classifiers are weighted and averaged, and the classification of the weighted average is used as the final classification result.

In the specific implementation of step S14, the multi-domain combination classifier is trained twice using the pre-trained target EEG training set to classify the target EEG signal test data set after multi-domain feature extraction, Thereby obtaining the final target EEG signal P300 classification result, this step may include the following sub-steps:

Step S141, using the pre-trained target EEG training set to perform secondary training on the once-trained multi-domain combined classifier to obtain a second-trained multi-domain combined classifier;

Specifically, the pre-trained target EEG training set after multi-domain feature extraction is used as the training set of the multi-domain combination classifier for the second training, and the multi-domain combination classifier trained once is trained twice to obtain A quadratic trained multi-domain ensemble classifier.

Step S142, using the multi-domain combined classifier trained twice to classify the test set of target EEG signals, and output the classification result as the final target EEG signal P300 classification result.

Specifically, the target EEG signal test data set through multi-domain feature extraction is used as the test set of the multi-domain combination classifier of the second training, and the multi-domain combination classifier of the second training is used to classify the test set, The classification result is output as the final target EEG signal P300 classification result.

Due to the introduction of migration data sets, such as data II in BCI Competition III, due to the large sample size of the migration data set, it can overcome the problem of fewer EEG signal samples and difficult learning in epileptic patients. By learning and utilizing the data distribution of the migration data set , to improve the effect of the classification accuracy of the test set in the final target data. Due to the P300 signal shape standard of the migration data set, it can overcome the abnormality of the EEG signal shape of epilepsy patients, filter out the poor quality samples in the training data set in the target data through migration pre-training, and reduce the classification of poor quality samples in the training data set The interference of the machine can achieve the effect of improving the classification accuracy of the test set in the final target data.

Corresponding to the aforementioned embodiment of a data migration P300 EEG signal detection method, the present application also provides an embodiment of a data migration P300 EEG signal detection device.

Fig. 3 is a block diagram of a P300 EEG signal detection device for data migration according to an exemplary embodiment. Referring to Figure 3, the device includes:

A preprocessing module 21, configured to perform multi-domain preprocessing on target EEG signals and migration data sets containing P300 components;

The feature extraction module 22 is used to extract multi-domain features from the target EEG signal and migration data set after the multi-domain preprocessing;

The pre-training module 23 is used to use the migration data set through multi-domain feature extraction as a training set, and use the target EEG signal training data set after multi-domain feature extraction as a test set, using multi-domain based on confidence coefficient Combining classifiers, using the correctly classified target EEG training data set after multi-domain feature extraction as the pre-trained target EEG training set;

The secondary training classification module 24 is used to use the pre-trained target EEG training set to perform secondary training on the multi-domain combination classifier, and classify the target EEG signal test data set after multi-domain feature extraction , so as to obtain the final target EEG signal P300 classification result.

Regarding the apparatus in the foregoing embodiments, the specific manner in which each module executes operations has been described in detail in the embodiments related to the method, and will not be described in detail here.

As for the device embodiment, since it basically corresponds to the method embodiment, for related parts, please refer to the part description of the method embodiment. The device embodiments described above are only illustrative, and the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in One place, or it can be distributed to multiple network elements. Part or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this application. It can be understood and implemented by those skilled in the art without creative effort.

Correspondingly, the present application also provides an electronic device, including: one or more processors; a memory for storing one or more programs; when the one or more programs are executed by the one or more processors , so that the one or more processors implement the P300 EEG signal detection method for data migration as described above.

Correspondingly, the present application also provides a computer-readable storage medium on which computer instructions are stored, which is characterized in that, when the instructions are executed by a processor, the above-mentioned P300 EEG signal detection method for data migration is implemented.

Other embodiments of the present application will readily occur to those skilled in the art from consideration of the specification and practice of the disclosure herein. This application is intended to cover any modification, use or adaptation of the application, these modifications, uses or adaptations follow the general principles of the application and include common knowledge or conventional technical means in the technical field not disclosed in the application . The specification and examples are to be considered exemplary only, with a true scope and spirit of the application indicated by the appended claims.

It should be understood that the present application is not limited to the precise constructions which have been described above and shown in the accompanying drawings, and various modifications and changes may be made without departing from the scope thereof. The scope of the application is limited only by the appended claims.

Claims (7)

1. A P300 EEG signal detection method of data migration, characterized in that, comprising: Perform multi-domain preprocessing on target EEG signals and migration data sets containing P300 components, wherein the target EEG signals are EEG signals of epilepsy patients, and the migration data set is a standard EEG signal data set of P300 signal morphology ; Extracting multi-domain features from the target EEG signal and migration data set after the multi-domain preprocessing; The migration data set through multi-domain feature extraction is used as a training set, and the target EEG signal training data set through multi-domain feature extraction is used as a test set, and a multi-domain combined classifier based on confidence coefficient is used to classify correctly The target EEG signal training data set through multi-domain feature extraction is used as the pre-trained target EEG training set; Use the pre-trained target EEG training set to perform secondary training on the multi-domain combination classifier, and classify the target EEG signal test data set after multi-domain feature extraction, so as to obtain the final target EEG signal P300 classification results; Among them, multi-domain preprocessing is performed on the target EEG signal and migration data sets containing P300 components, including: Unify the sampling frequency of the target EEG signal and migration data sets containing P300 components; Spatial preprocessing is performed on the target EEG signal and migration data sets after uniform sampling frequency by electrode selection and common average reference method; Time-domain preprocessing is performed on the target EEG signal and migration data set after uniform sampling frequency by extremum adjustment and normalization methods; Frequency-domain preprocessing is performed on the target EEG signal and the migration data set after the unified sampling frequency through the wavelet packet decomposition method; Wherein, the multi-domain feature is extracted from the target EEG signal and the migration data set after the multi-domain preprocessing, including: Extracting time-domain energy entropy as time-domain features from the multi-domain preprocessed target EEG signals and migration data sets respectively; Extracting energy information of wavelet coefficients as time-frequency features from the multi-domain preprocessed target EEG signals and migration data sets respectively; Using independent component analysis to extract signal spatial domain features on the target EEG signal and migration data set after the multi-domain preprocessing respectively; Wherein, the multi-domain combination classifier based on the confidence coefficient includes two linear discriminant analysis classifiers with different threshold constants and a naive Bayesian classifier, and the linear discriminant analysis classifier uses the distance of the feature vector X from the classification line As the confidence coefficient, the Naive Bayesian classifier takes the difference of the two classification probabilities as its confidence coefficient, and performs normalization adjustment and S-curve adjustment on each confidence coefficient. 2. The method according to claim 1, characterized in that, the migration data set through multi-domain feature extraction is used as a training set, and the target EEG training data set through multi-domain feature extraction is used as a test set , using a confidence coefficient-based multi-domain combination classifier, the correctly classified target EEG training data set after multi-domain feature extraction is used as a pre-trained target EEG training set, including: Use the migration data set to pre-train the multi-domain combination classifier based on the confidence coefficient to obtain a trained multi-domain combination classifier; Using the migration data set through multi-domain feature extraction as the training set of the once-trained multi-domain combined classifier; Using the target EEG signal training data set through multi-domain feature extraction as the test set of the multi-domain combination classifier for the one-time training; Output the target EEG signal training data set that has been correctly classified by the multi-domain combination classifier trained once and has undergone multi-domain feature extraction, as a pre-trained target EEG training set. 3. method according to claim 1, is characterized in that, for the different linear discriminant analysis classifiers of described two threshold constants, with wherein one linear discriminant analysis classifier is as basic component classifier, selects confidence coefficient threshold value, to Classification results with confidence coefficients higher than the threshold are accepted directly, and for classified samples with confidence coefficients lower than the threshold, a Naive Bayesian classifier and another linear discriminant analysis classifier with different threshold constants are used as reference component classifiers. 4. The method according to claim 2, characterized in that, using the pre-trained target EEG training set to carry out secondary training to the multi-domain combination classifier, the target EEG through multi-domain feature extraction The signal test data set is classified to obtain the final target EEG signal P300 classification results, including: Using the pre-trained target EEG training set to perform secondary training on the once-trained multi-domain combined classifier to obtain a second-trained multi-domain combined classifier; Using the pre-trained target EEG training set as the training set of the multi-domain combination classifier of the secondary training; Using the target EEG test data set through multi-domain feature extraction as the test set of the multi-domain combined classifier of the secondary training; The classification result of the multi-domain combination classifier trained twice is output as the final target EEG signal P300 classification result. 5. A P300 EEG signal detection device for data migration, characterized in that it comprises: The preprocessing module is used to perform multi-domain preprocessing on target EEG signals and migration data sets containing P300 components, wherein the target EEG signals are EEG signals of epilepsy patients, and the migration data sets are P300 signal morphology standards EEG signal data set; A feature extraction module, configured to extract multi-domain features from the target EEG signal and migration data set after the multi-domain preprocessing; The pre-training module is used to use the migration data set through multi-domain feature extraction as a training set, and use the target EEG signal training data set through multi-domain feature extraction as a test set, using multi-domain combination based on confidence coefficient A classifier, using the correctly classified target EEG training data set through multi-domain feature extraction as the pre-trained target EEG training set; The secondary training classification module is used to use the pre-trained target EEG training set to perform secondary training on the multi-domain combination classifier, and classify the target EEG signal test data set through multi-domain feature extraction, Thereby obtaining the final target EEG signal P300 classification result; Among them, multi-domain preprocessing is performed on the target EEG signal and migration data sets containing P300 components, including: Unify the sampling frequency of the target EEG signal and migration data sets containing P300 components; Spatial preprocessing is performed on the target EEG signal and migration data sets after uniform sampling frequency by electrode selection and common average reference method; Time-domain preprocessing is performed on the target EEG signal and migration data set after uniform sampling frequency by extremum adjustment and normalization methods; Frequency-domain preprocessing is performed on the target EEG signal and the migration data set after the unified sampling frequency through the wavelet packet decomposition method; Wherein, the multi-domain feature is extracted from the target EEG signal and the migration data set after the multi-domain preprocessing, including: Extracting time-domain energy entropy as time-domain features from the multi-domain preprocessed target EEG signals and migration data sets respectively; Extracting energy information of wavelet coefficients as time-frequency features from the multi-domain preprocessed target EEG signals and migration data sets respectively; Using independent component analysis to extract signal spatial domain features on the target EEG signal and migration data set after the multi-domain preprocessing respectively; Wherein, the multi-domain combination classifier based on the confidence coefficient includes two linear discriminant analysis classifiers with different threshold constants and a naive Bayesian classifier, and the linear discriminant analysis classifier uses the distance of the feature vector X from the classification line As the confidence coefficient, the Naive Bayesian classifier takes the difference of the two classification probabilities as its confidence coefficient, and performs normalization adjustment and S-curve adjustment on each confidence coefficient. 6. An electronic device, characterized in that it comprises: one or more processors; memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors are made to implement the method according to any one of claims 1-4. 7. A computer-readable storage medium, on which computer instructions are stored, wherein the steps of the method according to any one of claims 1-4 are implemented when the instructions are executed by a processor.

Images