Physiological electrical signal classification and processing method, device, computer equipment and storage medium

Abstract

The present application relates to a physiological electrical signal classification and processing method, apparatus, computer equipment and storage medium, the method comprising: performing data alignment on the initial physiological electrical signal to be classified corresponding to the target user identifier based on the target signal spatial information corresponding to the target user identifier to obtain the target physiological electrical signal to be classified; performing spatial feature extraction on the target physiological electrical signal to be classified based on the target spatial filter matrix to obtain the spatial features to be classified, the target spatial filter matrix being generated based on target training physiological electrical signals corresponding to multiple training user identifiers and training labels corresponding to each target training physiological electrical signal, the target training physiological electrical signal being obtained by performing data alignment on the initial training physiological electrical signal based on the training signal spatial information corresponding to the training user identifier; and obtaining the classification result corresponding to the initial physiological electrical signal to be classified based on the spatial features to be classified. The use of this method can improve the efficiency and classification accuracy of physiological electrical signals.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method, apparatus, computer equipment, and storage medium for classifying and processing physiological electrical signals. Background Art

[0002] With the development of computer technology, deep learning technology has shown obvious advantages in fields such as computer vision, speech recognition and natural language processing. Therefore, deep learning technology has gradually been introduced by researchers into the classification task of physiological electrical signals.

[0003] Physiological electrical signals can reflect human physiological activity, but due to differences in physiological activity responses between individuals, physiological electrical signals can vary significantly between subjects. Traditionally, physiological electrical signal classification has been based on transfer learning. However, transfer learning is a method that leverages source domain information to improve target domain learning performance. Model training requires the use of physiological electrical signals from both source and target domain subjects to develop a classification model suitable for target domain subjects, which is cumbersome. Summary of the Invention

[0004] Based on this, it is necessary to provide a physiological electrical signal classification and processing method, device, computer equipment and storage medium to address the above technical problems. This method can classify the physiological electrical signals of the target user without obtaining the physiological electrical signals of the target user in advance, which is more convenient and efficient.

[0005] A method for classifying and processing physiological electrical signals, the method comprising:

[0006] Obtaining an initial to-be-classified physiological electrical signal corresponding to the target user identifier;

[0007] Performing data alignment on the initial physiological electrical signal to be classified based on the target signal spatial information corresponding to the target user identifier to obtain the target physiological electrical signal to be classified;

[0008] Extracting spatial features of target physiological electrical signals to be classified based on a target spatial filter matrix to obtain spatial features to be classified, wherein the target spatial filter matrix is ​​generated based on target training physiological electrical signals corresponding to multiple training user identifiers and training labels corresponding to each target training physiological electrical signal, and the target training physiological electrical signal is obtained by aligning the initial training physiological electrical signal with the spatial information of the training signal corresponding to the training user identifier;

[0009] The classification result corresponding to the initial physiological electrical signal to be classified is obtained based on the spatial features to be classified.

[0010] In one embodiment, obtaining an initial to-be-classified physiological electrical signal corresponding to the target user identifier includes:

[0011] Obtain candidate physiological electrical signals to be classified corresponding to the target user identifier;

[0012] Performing signal extraction of at least one target frequency band on the candidate physiological electrical signal to be classified to obtain initial sub-signals to be classified corresponding to each target frequency band of the candidate physiological electrical signal to be classified;

[0013] An initial physiological electrical signal to be classified is obtained based on each initial sub-signal to be classified.

[0014] In one embodiment, before performing data alignment on the initial physiological electrical signal to be classified based on the target signal spatial information corresponding to the target user identifier to obtain the target physiological electrical signal to be classified, the method further includes:

[0015] Obtaining a starting reference matrix corresponding to the initial physiological electrical signal to be classified;

[0016] Correcting the initial reference matrix based on the initial physiological electrical signal to be classified to obtain a corrected reference matrix corresponding to the initial physiological electrical signal to be classified;

[0017] The modified reference matrix corresponding to the initial physiological electrical signal to be classified is used as the target signal spatial information.

[0018] In one embodiment, the starting reference matrix is ​​corrected based on the initial physiological electrical signal to be classified to obtain a corrected reference matrix corresponding to the initial physiological electrical signal to be classified, including:

[0019] Obtaining statistical results of the number of classified physiological electrical signals corresponding to the target user identifier;

[0020] Calculating the covariance matrix to be classified corresponding to the initial physiological electrical signal to be classified;

[0021] The initial reference matrix is ​​modified based on the number statistics result and the covariance matrix to be classified, so as to obtain the modified reference matrix corresponding to the initial physiological electrical signal to be classified.

[0022] In one embodiment, the starting reference matrix includes at least one starting reference submatrix corresponding to each target frequency band, and the initial physiological electrical signal to be classified includes at least one initial sub-signal to be classified corresponding to each target frequency band. The starting reference matrix is ​​corrected based on the statistical result of the number and the covariance matrix to be classified to obtain a corrected reference matrix corresponding to the initial physiological electrical signal to be classified, including:

[0023] Based on the initial sub-signals to be classified corresponding to the same target frequency band and the number statistics result, the corresponding starting reference sub-matrix is ​​corrected to obtain the corrected reference sub-matrices corresponding to the respective target frequency bands;

[0024] A modified reference matrix is ​​obtained based on each modified reference sub-matrix.

[0025] In one embodiment, the generation of the target spatial filter matrix includes the following steps:

[0026] Acquire initial training physiological electrical signals corresponding to a plurality of training user identifiers; the initial training physiological electrical signals carry training labels;

[0027] Performing data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier to obtain target training physiological electrical signals corresponding to each training user identifier;

[0028] A target spatial filter matrix is ​​generated based on the signal difference between target training physiological electrical signals corresponding to different training labels.

[0029] In one embodiment, before performing data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier to obtain the target training physiological electrical signals corresponding to the respective training user identifiers, the method further includes:

[0030] Generate corresponding initial reference matrices based on the initial training physiological electrical signals corresponding to the same training user identifier, and obtain initial reference matrices corresponding to the respective training user identifiers;

[0031] The initial reference matrix corresponding to the same training user identity is used as the corresponding training signal space information.

[0032] In one embodiment, the initial training physiological electrical signal includes at least one initial training sub-signal corresponding to each target frequency band, and the initial reference matrix corresponding to each initial training physiological electrical signal corresponding to the same training user identifier is generated based on the initial training physiological electrical signal. The initial reference matrix corresponding to each training user identifier is obtained, including:

[0033] Calculate the initial covariance matrix corresponding to each initial training sub-signal;

[0034] Calculate the corresponding initial reference submatrix based on each initial covariance matrix corresponding to the same training user identifier and the same target frequency band, and obtain the initial reference submatrix corresponding to each training user identifier in each target frequency band;

[0035] Based on each initial reference sub-matrix, an initial reference matrix corresponding to each training user identifier is obtained.

[0036] In one embodiment, the initial training physiological electrical signal includes channel signals corresponding to multiple acquisition channels on the physiological electrical signal acquisition device, and the initial training sub-signal includes channel sub-signals corresponding to each acquisition channel.

[0037] Calculate the initial covariance matrix corresponding to each initial training sub-signal, including:

[0038] In the current initial training sub-signal, the covariance between the sub-signals of each channel is calculated;

[0039] An initial covariance matrix corresponding to the current initial training sub-signal is generated based on the covariance between the sub-signals of each channel.

[0040] In one embodiment, the initial reference matrix includes at least one initial reference sub-matrix corresponding to each target frequency band, and the initial training physiological electrical signal includes at least one initial training sub-signal corresponding to each target frequency band. Data alignment is performed on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier to obtain target training physiological electrical signals corresponding to each training user identifier, including:

[0041] The initial reference sub-matrix and the initial training sub-signal corresponding to the same training user identifier and the same target frequency band are fused to obtain the target training sub-signals corresponding to each training user identifier in each target frequency band;

[0042] Based on the target training sub-signals corresponding to the training user identifiers in the target frequency bands, target training physiological electrical signals corresponding to the training user identifiers are obtained.

[0043] In one embodiment, the target training physiological electrical signal includes at least one target training sub-signal corresponding to each target frequency band, and generating the target spatial filter matrix based on the signal difference between the target training physiological electrical signals corresponding to different training labels includes:

[0044] In the same target frequency band, the corresponding target spatial filter submatrix is ​​generated based on the signal difference between the target training sub-signals corresponding to different training labels, thereby obtaining the target spatial filter submatrix corresponding to each target frequency band;

[0045] A target spatial filter matrix is ​​generated based on each target spatial filter sub-matrix.

[0046] In one embodiment, in the same target frequency band, a corresponding target spatial filter submatrix is ​​generated based on the signal difference between target training sub-signals corresponding to different training labels, thereby obtaining target spatial filter submatrices corresponding to each target frequency band, including:

[0047] In the current target frequency band, calculate the target covariance matrix corresponding to each target training sub-signal;

[0048] Calculate the corresponding target reference matrix based on the target covariance matrix corresponding to the same training label, and obtain the target reference matrix corresponding to each training label;

[0049] The target reference matrices are fused to obtain a fused reference matrix, and the fused reference matrix is ​​subjected to eigenvalue decomposition to obtain an initial eigenvalue matrix and an initial eigenvector matrix corresponding to the fused reference matrix;

[0050] Obtain a whitening matrix based on the initial eigenvalue matrix and the initial eigenvector matrix;

[0051] Perform whitening transformation on each target reference matrix based on the whitening matrix to obtain the transformation reference matrix corresponding to each target reference matrix;

[0052] Perform eigenvalue decomposition on any transformation reference matrix to obtain an eigenvalue decomposition result, and obtain a target eigenvector matrix based on the eigenvalue decomposition result;

[0053] A target spatial filter submatrix corresponding to the current target frequency band is generated based on the whitening matrix and the target eigenvector matrix.

[0054] In one embodiment, generating a target spatial filter submatrix corresponding to the current target frequency band based on the whitening matrix and the target eigenvector matrix includes:

[0055] Fuse the whitening matrix and the target eigenvector matrix to obtain the initial spatial filter matrix;

[0056] Extracting at least one initial spatial filter submatrix from the initial spatial filter matrix to obtain at least one initial spatial filter submatrix;

[0057] A target spatial filter submatrix is ​​obtained based on each initial spatial filter submatrix.

[0058] In one embodiment, obtaining a classification result corresponding to the initial physiological electrical signal to be classified based on the spatial features to be classified includes:

[0059] The spatial features to be classified are input into the target physiological electrical signal classification model to obtain the classification results.

[0060] In one embodiment, the training process of the target physiological electrical signal classification model includes the following steps:

[0061] Based on the target spatial filter matrix, spatial features of each target training physiological electrical signal are extracted to obtain the training spatial features corresponding to each target training physiological electrical signal;

[0062] Input each training spatial feature into the initial physiological electrical signal classification model to obtain the prediction label corresponding to each target training physiological electrical signal;

[0063] Based on the predicted labels and training labels corresponding to the same target training physiological electrical signals, the model parameters of the initial physiological electrical signal classification model are adjusted until the convergence conditions are met, thereby obtaining the target physiological electrical signal classification model.

[0064] A physiological electrical signal classification and processing device, comprising:

[0065] A signal acquisition module is used to acquire an initial physiological electrical signal to be classified corresponding to the target user identification;

[0066] A data alignment module is used to align the initial physiological electrical signal to be classified based on the target signal space information corresponding to the target user identifier to obtain the target physiological electrical signal to be classified;

[0067] A feature extraction module is used to extract spatial features of the target physiological electrical signals to be classified based on a target spatial filter matrix to obtain spatial features to be classified. The target spatial filter matrix is ​​generated based on target training physiological electrical signals corresponding to multiple training user identifiers and training labels corresponding to each target training physiological electrical signal. The target training physiological electrical signals are obtained by aligning the initial training physiological electrical signals with the spatial information of the training signals corresponding to the training user identifiers.

[0068] The signal classification module is used to obtain the classification result corresponding to the initial physiological electrical signal to be classified based on the spatial features to be classified.

[0069] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0070] Obtaining an initial to-be-classified physiological electrical signal corresponding to the target user identifier;

[0071] Performing data alignment on the initial physiological electrical signal to be classified based on the target signal spatial information corresponding to the target user identifier to obtain the target physiological electrical signal to be classified;

[0072] Extracting spatial features of target physiological electrical signals to be classified based on a target spatial filter matrix to obtain spatial features to be classified, wherein the target spatial filter matrix is ​​generated based on target training physiological electrical signals corresponding to multiple training user identifiers and training labels corresponding to each target training physiological electrical signal, and the target training physiological electrical signal is obtained by aligning the initial training physiological electrical signal with the spatial information of the training signal corresponding to the training user identifier;

[0073] The classification result corresponding to the initial physiological electrical signal to be classified is obtained based on the spatial features to be classified.

[0074] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

[0075] Obtaining an initial to-be-classified physiological electrical signal corresponding to the target user identifier;

[0076] Performing data alignment on the initial physiological electrical signal to be classified based on the target signal spatial information corresponding to the target user identifier to obtain the target physiological electrical signal to be classified;

[0077] Extracting spatial features of target physiological electrical signals to be classified based on a target spatial filter matrix to obtain spatial features to be classified, wherein the target spatial filter matrix is ​​generated based on target training physiological electrical signals corresponding to multiple training user identifiers and training labels corresponding to each target training physiological electrical signal, and the target training physiological electrical signal is obtained by aligning the initial training physiological electrical signal with the spatial information of the training signal corresponding to the training user identifier;

[0078] The classification result corresponding to the initial physiological electrical signal to be classified is obtained based on the spatial features to be classified.

[0079] The above-mentioned physiological electrical signal classification and processing method, device, computer equipment and storage medium first perform data alignment on the corresponding initial physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier, which can reduce the distribution differences between the physiological electrical signals of different training users. Then, based on the target physiological electrical signal obtained by data alignment and the corresponding training label, a universal target spatial filter matrix can be generated. The target spatial filter matrix can be used to extract spatial features in the physiological electrical signal that can be used to distinguish the categories of physiological electrical signals. Then, when classifying the physiological electrical signals of unknown users, the initial physiological electrical signals to be classified corresponding to the target user identifier are first data aligned based on the target signal spatial information corresponding to the target user identifier to reduce the distribution differences between the physiological electrical signals of the target user and the training user. Then, based on the universal target spatial filter matrix, the spatial features of the target physiological electrical signals to be classified obtained by data alignment are extracted, so that the classification results corresponding to the initial physiological electrical signals to be classified can be obtained based on the extracted spatial features to be classified. In this way, the classification of the physiological electrical signals of the target user can be achieved without pre-acquiring the physiological electrical signals of the target user, which is more convenient and efficient.

[0080] A method for classifying and processing physiological electrical signals, the method comprising:

[0081] Acquire initial training physiological electrical signals corresponding to a plurality of training user identifiers; the initial training physiological electrical signals carry training labels;

[0082] Performing data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier to obtain target training physiological electrical signals corresponding to each training user identifier;

[0083] Generate a target spatial filter matrix based on the signal difference between target training physiological electrical signals corresponding to different training labels;

[0084] Based on the target spatial filter matrix, spatial features of each target training physiological electrical signal are extracted to obtain the training spatial features corresponding to each target training physiological electrical signal;

[0085] The initial physiological electrical signal classification model is trained based on the training space features and training labels corresponding to each target training physiological electrical signal until the convergence condition is met, thereby obtaining the target physiological electrical signal classification model.

[0086] A physiological electrical signal classification and processing device, comprising:

[0087] A signal acquisition module is used to acquire initial training physiological electrical signals corresponding to multiple training user identifiers; the initial training physiological electrical signals carry training labels;

[0088] A data alignment module is used to align the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier, so as to obtain the target training physiological electrical signals corresponding to each training user identifier;

[0089] A spatial filter matrix generation module is used to generate a target spatial filter matrix based on the signal difference between target training physiological electrical signals corresponding to different training labels;

[0090] A feature extraction module is used to extract spatial features of each target training physiological electrical signal based on the target spatial filter matrix to obtain the training spatial features corresponding to each target training physiological electrical signal;

[0091] The model training module is used to train the initial physiological electrical signal classification model based on the training space features and training labels corresponding to each target training physiological electrical signal until the convergence condition is met to obtain the target physiological electrical signal classification model.

[0092] A computer device includes a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0093] Acquire initial training physiological electrical signals corresponding to a plurality of training user identifiers; the initial training physiological electrical signals carry training labels;

[0094] Performing data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier to obtain target training physiological electrical signals corresponding to each training user identifier;

[0095] Generate a target spatial filter matrix based on the signal difference between target training physiological electrical signals corresponding to different training labels;

[0096] Based on the target spatial filter matrix, spatial features of each target training physiological electrical signal are extracted to obtain the training spatial features corresponding to each target training physiological electrical signal;

[0097] The initial physiological electrical signal classification model is trained based on the training space features and training labels corresponding to each target training physiological electrical signal until the convergence condition is met, thereby obtaining the target physiological electrical signal classification model.

[0098] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the following steps:

[0099] Acquire initial training physiological electrical signals corresponding to a plurality of training user identifiers; the initial training physiological electrical signals carry training labels;

[0100] Performing data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier to obtain target training physiological electrical signals corresponding to each training user identifier;

[0101] Generate a target spatial filter matrix based on the signal difference between target training physiological electrical signals corresponding to different training labels;

[0102] Based on the target spatial filter matrix, spatial features of each target training physiological electrical signal are extracted to obtain the training spatial features corresponding to each target training physiological electrical signal;

[0103] The initial physiological electrical signal classification model is trained based on the training space features and training labels corresponding to each target training physiological electrical signal until the convergence condition is met, thereby obtaining the target physiological electrical signal classification model.

[0104] The above-mentioned physiological electrical signal classification and processing method, device, computer equipment and storage medium perform data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier, which can reduce the distribution differences between the physiological electrical signals of different training users. Then, based on the target physiological electrical signals obtained by data alignment and the corresponding training labels, a universal target spatial filter matrix can be generated. The target spatial filter matrix can be used to extract spatial features in the physiological electrical signals that can be used to distinguish the categories of physiological electrical signals. In this way, it is possible to train and obtain a target spatial filter matrix and a target physiological electrical signal classification model that can be used to classify the physiological electrical signals of the target user without pre-acquiring the physiological electrical signals of the target user. The target spatial filter matrix and the target physiological electrical signal classification model can be used to classify the physiological electrical signals of the target user, which is more convenient and efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0105] Figure 1A diagram showing an application environment of a physiological electrical signal classification and processing method according to an embodiment;

[0106] Figure 2 1 is a flow chart of a method for classifying and processing physiological electrical signals in one embodiment;

[0107] Figure 3 A schematic diagram of a process for determining target signal spatial information in one embodiment;

[0108] Figure 4 A schematic diagram of a process for determining spatial information of a target signal in another embodiment;

[0109] Figure 5 A schematic diagram of a process for generating a target spatial filter matrix in one embodiment;

[0110] Figure 6 A schematic diagram of a process for generating a target spatial filter matrix in another embodiment;

[0111] Figure 7 A schematic diagram of a process for generating a target spatial filter matrix in another embodiment;

[0112] Figure 8 is a flow chart of a method for classifying and processing physiological electrical signals in another embodiment;

[0113] Figure 9A is a schematic diagram of the structure of an EEG signal in one embodiment;

[0114] Figure 9B 1 is a schematic diagram of a process for classifying EEG signals in one embodiment;

[0115] Figure 10 is a structural block diagram of a physiological electrical signal classification and processing device in one embodiment;

[0116] Figure 11 is a structural block diagram of a physiological electrical signal classification and processing device in another embodiment;

[0117] Figure 12 is a structural block diagram of a physiological electrical signal classification and processing device in yet another embodiment;

[0118] Figure 13 A structural block diagram of a physiological electrical signal classification and processing device in yet another embodiment;

[0119] Figure 14 is a diagram of the internal structure of a computer device in one embodiment;

[0120] Figure 15 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0121] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0122] Artificial intelligence (AI) technology is a comprehensive discipline encompassing a wide range of fields, encompassing both hardware and software technologies. Foundational AI technologies generally include sensors, specialized AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technologies primarily encompass computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0123] Machine learning (ML) is a multidisciplinary field that encompasses probability theory, statistics, approximation theory, convex analysis, and algorithmic complexity theory. It specifically studies how computers can simulate or implement human learning behaviors to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is at the core of artificial intelligence and the fundamental way to make computers intelligent. Its applications span all areas of AI. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and self-learning.

[0124] The solutions provided in the embodiments of this application involve artificial intelligence machine learning, big data processing and other technologies, which are specifically illustrated by the following embodiments:

[0125] The physiological electrical signal classification and processing method provided in this application can be applied to Figure 1 In the application environment shown, the terminal 102 communicates with the server 104 via a network. The terminal 102 can be, but is not limited to, various personal computers, laptops, smart phones, tablet computers, portable wearable devices, and physiological electrical signal acquisition devices. The server 104 can be implemented as an independent server or a server cluster consisting of multiple servers or a cloud server.

[0126] The terminal 102 and the server 104 can both be used independently to execute the physiological electrical signal classification and processing method provided in the embodiments of the present application.

[0127] For example, the terminal first obtains the initial physiological electrical signal to be classified corresponding to the target user identifier, and performs data alignment on the initial physiological electrical signal to be classified based on the target signal spatial information corresponding to the target user identifier to obtain the target physiological electrical signal to be classified. The terminal then obtains the target spatial filter matrix, and extracts spatial features of the target physiological electrical signal to be classified based on the target spatial filter matrix to obtain the spatial features to be classified. Among them, the target spatial filter matrix is ​​generated based on the target training physiological electrical signals corresponding to multiple training user identifiers and the training labels corresponding to each target training physiological electrical signal, and the target training physiological electrical signal is obtained by performing data alignment on the initial training physiological electrical signal based on the training signal spatial information corresponding to the training user identifier. Finally, the terminal can obtain the classification result corresponding to the initial physiological electrical signal to be classified based on the spatial features to be classified.

[0128] The server obtains initial training physiological electrical signals corresponding to multiple training user identifiers, wherein each initial training physiological electrical signal carries a corresponding training label. The server performs data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier, and can obtain target training physiological electrical signals corresponding to each training user identifier. The target spatial filter matrix can be generated based on the signal difference between the target training physiological electrical signals corresponding to different training labels. Then, the server extracts spatial features of each target training physiological electrical signal based on the target spatial filter matrix, obtains the training spatial features corresponding to each target training physiological electrical signal, and performs model training on the initial physiological electrical signal classification model based on the training spatial features and training labels corresponding to each target training physiological electrical signal, until the training is completed, and the target physiological electrical signal classification model is obtained.

[0129] The terminal 102 and the server 104 can also be used in conjunction to execute the physiological electrical signal classification and processing method provided in the embodiments of the present application.

[0130] For example, the server generates a target spatial filter matrix based on the target training physiological electrical signals corresponding to the training user identifiers and the training labels corresponding to the target training physiological electrical signals. The terminal obtains the target spatial filter matrix from the server and classifies the initial physiological electrical signals to be classified corresponding to the target user identifier based on the target spatial filter matrix.

[0131] The terminal collects initial training physiological electrical signals corresponding to multiple training user identifiers and determines the training labels corresponding to each initial training physiological electrical signal. The server obtains the initial training physiological electrical signals corresponding to multiple training user identifiers from the terminal and trains a target spatial filter matrix and a target physiological electrical signal classification model based on each initial training physiological electrical signal and the corresponding training labels.

[0132] In one embodiment, Figure 2As shown, a physiological electrical signal classification and processing method is provided, which is applied to Figure 1 The computer device in the example is used to illustrate, and the computer device can be the above Figure 1 The terminal 102 or the server 104 in FIG. Figure 2 , the physiological electrical signal classification and processing method includes the following steps:

[0133] Step S202: obtaining an initial physiological electrical signal to be classified corresponding to the target user identifier.

[0134] Among them, the user identifier is an identifier used to uniquely identify a user, and specifically may include a character string of at least one of letters, numbers, and symbols. The target user identifier refers to the user identifier corresponding to the target user. The physiological electrical signal refers to a physiological signal presented as current or voltage, which is used to reflect the electrophysiological activity of nerve cells. The physiological electrical signal can specifically be an electroencephalogram signal, an electromyogram signal, an electrocardiogram signal, etc. The physiological electrical signal to be classified refers to the physiological electrical signal to be classified. The initial physiological electrical signal to be classified refers to the physiological electrical signal to be classified that has not undergone data alignment.

[0135] Specifically, the computer device can obtain the initial to-be-classified physiological electrical signals corresponding to the target user identifier locally or from another terminal or server. It is understood that when collecting physiological electrical signals, the physiological electrical signal acquisition device can associate the collected physiological electrical signals with the corresponding user. Specifically, the physiological electrical signals can be associated with the user identifier of the corresponding user, thereby effectively distinguishing the physiological electrical signals of different users based on the user identifier corresponding to the physiological electrical signals.

[0136] In one embodiment, the physiological electrical signals to be classified may be physiological electrical signals collected in real time. The physiological electrical signal collection device may collect physiological electrical signals in real time, and the computer device may classify the latest physiological electrical signals in real time to obtain corresponding classification results. Of course, the physiological electrical signals to be classified may also be physiological electrical signals collected at a historical time. The physiological electrical signals collected in real time by the physiological electrical signal collection device may be stored in a database of a terminal or server, and the computer device may obtain physiological electrical signals collected at a historical time from the database, classify the physiological electrical signals, and obtain corresponding classification results.

[0137] In one embodiment, the initial physiological electrical signal to be classified can be an original physiological electrical signal, that is, the physiological electrical signal collected by the physiological electrical signal acquisition device is directly used as the initial physiological electrical signal to be classified. The initial physiological electrical signal to be classified can also be a pre-processed physiological electrical signal. For example, the physiological electrical signal collected by the physiological electrical signal acquisition device is band-pass filtered, and the filtered physiological electrical signal is used as the initial physiological electrical signal to be classified. First band-pass filtering the physiological electrical signal to filter out noise, and then classifying the band-pass filtered physiological electrical signal can effectively improve the classification accuracy.

[0138] Step S204 , performing data alignment on the initial physiological electrical signal to be classified based on the target signal space information corresponding to the target user identifier to obtain the target physiological electrical signal to be classified.

[0139] Among them, the target signal space information refers to the signal space information corresponding to the target user. Signal space information refers to the signal distribution information generated based on multiple physiological electrical signals of a user in Euclidean space, which is used to characterize the overall distribution of multiple physiological electrical signals of a user. Different users correspond to different signal space information. Data alignment refers to mapping the initial physiological electrical signals to be classified into the target range to obtain the target physiological electrical signals to be classified, so that the target physiological electrical signals to be classified corresponding to each initial physiological electrical signal to be classified can be located in the same target range, thereby achieving data alignment of each initial physiological electrical signal to be classified.

[0140] Specifically, the computer device can obtain multiple physiological electrical signals corresponding to the same user identifier and generate signal space information corresponding to the user identifier based on the multiple physiological electrical signals. After obtaining the initial physiological electrical signal to be classified corresponding to the target user identifier, the computer device can obtain the target signal space information corresponding to the target user identifier and perform data alignment on the initial physiological electrical signal to be classified based on the target signal space information, thereby obtaining the target physiological electrical signal to be classified.

[0141] In one embodiment, the signal space information is updated in real time. For example, once the computer device acquires the physiological electrical signals corresponding to the target user identifier, it can update the signal space information corresponding to the target user identifier. It will be understood that the more physiological electrical signals there are, the more accurate and reliable the generated signal space information.

[0142] In one embodiment, the computer device can generate signal space information corresponding to the user identifier based on the covariance matrix corresponding to at least one physiological electrical signal corresponding to the same user identifier. Specifically, the mean of the covariance matrices corresponding to the various physiological electrical signals can be used as the signal space information. The physiological electrical signals include channel signals collected by multiple acquisition channels. The covariance matrix corresponding to the physiological electrical signals can reflect the correlation between the channel signals and the distribution of the channel signals. The mean of the covariance matrix corresponding to the various physiological electrical signals can reflect the average correlation between the channel signals and the average distribution of the channel signals. The average distribution of the channel signals is used as the overall distribution of the physiological electrical signals.

[0143] In one embodiment, the initial physiological electrical signal to be classified may be a preprocessed physiological electrical signal, so the initial physiological electrical signal to be classified may include an initial sub-signal to be classified corresponding to at least one target frequency band. Correspondingly, the target signal spatial information may include target signal spatial sub-information corresponding to at least one target frequency band. Then, when data alignment is performed on the initial physiological electrical signal to be classified based on the target signal spatial information, data alignment can be performed on the corresponding initial sub-signals to be classified based on the target signal spatial sub-information corresponding to the same target frequency band to obtain the target sub-information to be classified corresponding to each target frequency band, and the target physiological electrical signal to be classified is obtained based on each target sub-information to be classified.

[0144] Step S206, spatial features of the target physiological electrical signals to be classified are extracted based on the target spatial filter matrix to obtain spatial features to be classified. The target spatial filter matrix is ​​generated based on the target training physiological electrical signals corresponding to multiple training user identifiers and the training labels corresponding to each target training physiological electrical signal. The target training physiological electrical signals are obtained by data alignment of the initial training physiological electrical signals based on the training signal spatial information corresponding to the training user identifier.

[0145] Among them, the target spatial filter matrix is ​​a spatial filter used to extract the spatial features of physiological electrical signals. It can extract spatial features with high discrimination from physiological electrical signals, so that the classification results of physiological electrical signals can be obtained based on the extracted spatial features. The target spatial filter matrix is ​​generated based on the target training physiological electrical signals corresponding to multiple training user identifiers and obtained after data alignment, as well as the training labels corresponding to each target training physiological electrical signal. The generated target spatial filter matrix can maximize the difference in spatial features corresponding to physiological electrical signals of different categories, so that the spatial features extracted from physiological electrical signals based on the target spatial filter matrix can have high discrimination, which is beneficial to the classification of physiological electrical signals.

[0146] The training user ID refers to the user ID corresponding to the training user. The training user and the target user are different users. The training signal space information refers to the signal space information corresponding to the training user. The training electrophysiological signal refers to the electrophysiological signal corresponding to the training user, which is a known electrophysiological signal. The training label refers to the classification result corresponding to the training electrophysiological signal. The initial training electrophysiological signal refers to the training electrophysiological signal before data alignment. The target training electrophysiological signal refers to the training electrophysiological signal after data alignment.

[0147] Specifically, the computer device obtains the initial training physiological electrical signals corresponding to multiple training user identifiers and the training labels corresponding to each initial training physiological electrical signal, and performs data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier, thereby obtaining the target training physiological electrical signals corresponding to each initial training physiological electrical signal. For example, the initial training physiological electrical signals corresponding to training user A are data aligned based on the training signal spatial information corresponding to training user A, and the target training physiological electrical signals corresponding to the initial training physiological electrical signals of training user A are obtained. Then, the computer device generates a target spatial filter matrix based on the target training physiological electrical signals corresponding to each training user identifier and the training labels corresponding to each target training physiological electrical signal. Then, after obtaining the target physiological electrical signal to be classified, the computer device can perform spatial feature extraction on the target physiological electrical signal to be classified based on the target spatial filter matrix to obtain the spatial features to be classified, and thus the classification result corresponding to the initial physiological electrical signal to be classified can be obtained based on the spatial features to be classified.

[0148] It can be understood that the target spatial filter matrix is ​​a universal spatial filter matrix, which can be applied to both the physiological electrical signals corresponding to the target user identifier and the physiological electrical signals corresponding to the training user identifier.

[0149] In one embodiment, the target spatial filter matrix may include at least one target spatial filter sub-matrix. Based on each target spatial filter sub-matrix, spatial features of the target physiological electrical signals to be classified are extracted respectively to obtain each spatial sub-feature to be classified, and the spatial features to be classified are obtained based on each spatial sub-feature to be classified.

[0150] In one embodiment, extracting spatial features from target physiological electrical signals to be classified based on a target spatial filter matrix can specifically include performing signal projection on the target physiological electrical signals to be classified based on the target spatial filter matrix, and obtaining the spatial features to be classified based on the signal projection results. Accordingly, when the target spatial filter matrix includes at least one target spatial filter submatrix, signal projection is performed on the target physiological electrical signals to be classified based on each target spatial filter submatrix, and the spatial features to be classified are obtained based on each signal projection result.

[0151] Step S208 : obtaining a classification result corresponding to the initial physiological electrical signal to be classified based on the spatial feature to be classified.

[0152] Specifically, since the spatial features to be classified have a certain degree of discrimination, the computer device can obtain the classification result corresponding to the initial physiological electrical signal to be classified based on the spatial features to be classified.

[0153] In one embodiment, the computer device may perform data processing on the spatial features to be classified based on a custom formula to obtain a classification result.

[0154] In one embodiment, the classification processing of physiological electrical signals can be performed with the help of a machine learning model, specifically, a classifier for classifying spatial features can be trained. The computer device can extract spatial features of each target training physiological electrical signal based on the target spatial filter matrix, obtain the training spatial features corresponding to each target training physiological electrical signal, and obtain the target physiological electrical signal classification model based on the training spatial features and training labels corresponding to each target training physiological electrical signal. Then, when the classification result corresponding to the initial physiological electrical signal to be classified is obtained based on the spatial features to be classified, the computer device can obtain the trained target physiological electrical signal classification model, input the spatial features to be classified into the target physiological electrical signal classification model, and the target physiological electrical signal classification model outputs the classification result. Among them, the classifier can be a logistic regression classifier, or an SVM (Support Vector Machine), etc.

[0155] In one embodiment, different target spatial filter matrices can be generated for different classification tasks, and different target physiological electrical signal classification models can also be generated. For example, when the physiological electrical signal is an EEG signal, the classification tasks for the EEG signal may include emotion classification, motor imagery classification, attention classification, etc. Then, a spatial filter matrix and an EEG signal classification model specifically used for emotion classification of EEG signals can be trained based on the EEG signal whose training label is the result of emotion classification, a spatial filter matrix and an EEG signal classification model specifically used for motor imagery classification of EEG signals can be trained based on the EEG signal whose training label is the result of motor imagery classification, and a spatial filter matrix and an EEG signal classification model specifically used for attention classification of EEG signals can be trained based on the EEG signal whose training label is the result of attention classification. When the physiological electrical signal is an electromyographic signal, the classification tasks for the electromyographic signal include emotion classification, muscle state classification, etc. Then, a spatial filter matrix and an electromyographic signal classification model specifically used for emotion classification of electromyographic signals can be trained based on the electromyographic signals whose training labels are emotion classification results, and a spatial filter matrix and an electromyographic signal classification model specifically used for muscle state classification of electromyographic signals can be trained based on the electromyographic signals whose training labels are muscle state classification results.

[0156] In one embodiment, the classification results of EEG signals in a motor imagery classification task can be used to help people with disabilities achieve functions such as object grasping and prosthetic limb control. When performing motor imagery, users generate EEG signals with certain characteristics. For example, the EEG signals of users imagining left-hand movement and right-hand movement are different. The physiological electrical signal classification and processing method of the present application can be applied to an online brain-computer interface system to accurately classify EEG signals for motor imagery, thereby helping people with disabilities achieve functions such as object grasping and prosthetic limb control.

[0157] In the above-mentioned physiological electrical signal classification and processing method, data alignment is first performed on the corresponding initial physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier, which can reduce the distribution differences between the physiological electrical signals of different training users. Then, a universal target spatial filter matrix can be generated based on the target physiological electrical signal obtained by data alignment and the corresponding training label. The spatial features in the physiological electrical signal that can be used to distinguish the categories of physiological electrical signals can be extracted through the target spatial filter matrix. Then, when classifying the physiological electrical signals of unknown users, data alignment is first performed on the initial physiological electrical signals to be classified corresponding to the target user identifier based on the target signal spatial information corresponding to the target user identifier to reduce the distribution differences between the physiological electrical signals of the target user and the training user. Then, the spatial features of the target physiological electrical signals to be classified obtained by data alignment are extracted based on the universal target spatial filter matrix, so that the classification results corresponding to the initial physiological electrical signals to be classified can be obtained based on the extracted spatial features to be classified. In this way, the classification of the physiological electrical signals of the target user can be achieved without pre-acquiring the physiological electrical signals of the target user, which is more convenient and efficient.

[0158] In one embodiment, Figure 3 As shown, obtaining the initial physiological electrical signal to be classified corresponding to the target user identifier includes:

[0159] Acquire a candidate physiological electrical signal to be classified corresponding to the target user identifier; perform signal extraction of at least one target frequency band on the candidate physiological electrical signal to be classified to obtain initial sub-signals to be classified corresponding to the candidate physiological electrical signal to be classified in each target frequency band; and obtain an initial physiological electrical signal to be classified based on each initial sub-signal to be classified.

[0160] Specifically, a candidate physiological electrical signal to be classified refers to a physiological electrical signal to be classified that has not undergone any data processing. The computer device can obtain the candidate physiological electrical signal to be classified corresponding to the target user identifier and perform bandpass filtering on the candidate physiological electrical signal to be classified. That is, the computer device can extract the signal of at least one target frequency band from the candidate physiological electrical signal to be classified, thereby obtaining the initial sub-signals to be classified corresponding to the candidate physiological electrical signal to be classified in each target frequency band. Then, the initial physiological electrical signal to be classified is composed of each initial sub-signal to be classified.

[0161] In one embodiment, different target frequency bands may or may not overlap. For example, the target frequency bands may be divided into 4-8 Hz, 8-12 Hz, 12-16 Hz, and 16-20 Hz, with no overlap between the target frequency bands. Alternatively, the target frequency bands may be divided into 4-8 Hz, 6-10 Hz, 8-12 Hz, 10-14 Hz, 12-16 Hz, 14-18 Hz, and 16-20 Hz, with overlap between the target frequency bands.

[0162] In this embodiment, by extracting the signal of the target frequency band, not only the noise and some invalid signals in the physiological electrical signals to be classified can be filtered out, but also the physiological electrical signals to be classified can be subdivided, and the big data can be subdivided into small data for subsequent processing, which helps to improve the accuracy of physiological electrical signal classification.

[0163] In one embodiment, Figure 3 As shown, before performing data alignment on the initial physiological electrical signal to be classified based on the target signal space information corresponding to the target user identifier to obtain the target physiological electrical signal to be classified, the method further includes:

[0164] Step S302: obtaining a starting reference matrix corresponding to the initial physiological electrical signal to be classified.

[0165] Step S304 : correcting the initial reference matrix based on the initial physiological electrical signal to be classified to obtain a corrected reference matrix corresponding to the initial physiological electrical signal to be classified.

[0166] Step S306: Using the modified reference matrix corresponding to the initial physiological electrical signal to be classified as target signal spatial information.

[0167] Among them, the reference matrix is ​​used to align the data of the physiological electrical signals. The starting reference matrix corresponding to the initial physiological electrical signal to be classified is generated based on the classified physiological electrical signal corresponding to the target user identifier, and is used to align the data of the previous physiological electrical signal to be classified corresponding to the target user identifier. The corrected reference matrix corresponding to the initial physiological electrical signal to be classified is generated based on the initial physiological electrical signal to be classified and the starting reference matrix, and is used to align the data of the initial physiological electrical signal to be classified.

[0168] Specifically, the computer device can perform data alignment on the physiological electrical signals based on the reference matrix to reduce the distribution differences between the physiological electrical signals of different users. The computer device can obtain a starting reference matrix corresponding to the initial physiological electrical signals to be classified, correct the starting reference matrix based on the initial physiological electrical signals to be classified, obtain a corrected reference matrix corresponding to the initial physiological electrical signals to be classified, use the corrected reference matrix as the target signal spatial information corresponding to the target user identifier, and then perform data alignment on the initial physiological electrical signals to be classified based on the corrected reference matrix to obtain the target physiological electrical signals to be classified.

[0169] In one embodiment, the initial reference matrix is ​​a modified reference matrix corresponding to the last physiological electrical signal to be classified corresponding to the target user identifier.

[0170] Specifically, during the physiological electrical signal classification process, the reference matrix is ​​gradually revised. Each time the computer device obtains a physiological electrical signal to be classified corresponding to the target user's identifier, it revises the reference matrix. Therefore, the starting reference matrix corresponding to the target user's current physiological electrical signal to be classified is the revised reference matrix corresponding to the target user's previous physiological electrical signal to be classified. In this way, by continuously revising the reference matrix based on new data, the reference matrix used for data alignment will become increasingly accurate and reliable, thereby helping to improve the accuracy of physiological electrical signal classification.

[0171] For example, when the target user's unclassified physiological electrical signal is obtained for the first time, the computer device can initialize the reference matrix corresponding to the target user to obtain the starting reference matrix A1, correct the starting reference matrix A1 based on the physiological electrical signal to be classified, obtain the corrected reference matrix B1, and use the corrected reference matrix B1 as the target signal space information corresponding to the physiological electrical signal to be classified. When a new physiological electrical signal to be classified from the target user is obtained, the computer device can use the corrected reference matrix B1 as the starting reference matrix A2, correct the starting reference matrix A2 based on the physiological electrical signal to be classified, obtain the corrected reference matrix B2, and use the corrected reference matrix B2 as the target signal space information corresponding to the physiological electrical signal to be classified. Similarly, in the process of physiological electrical signal classification, the reference matrix is ​​gradually corrected. Specifically, initializing the reference matrix corresponding to the target user can be to initialize the reference matrix corresponding to the target user to 0.

[0172] In this embodiment, the starting reference matrix is ​​corrected based on the initial unclassified physiological electrical signals to obtain a corrected reference matrix corresponding to the initial unclassified physiological electrical signals. This corrected reference matrix corresponding to the initial unclassified physiological electrical signals is then used as the target signal spatial information. In this way, the reference matrix used for data alignment continuously incorporates relevant information about the current physiological electrical signals, which can more accurately reflect the overall distribution of multiple physiological electrical signals of the target user.

[0173] In one embodiment, Figure 4 As shown, the starting reference matrix is ​​corrected based on the initial physiological electrical signal to be classified to obtain a corrected reference matrix corresponding to the initial physiological electrical signal to be classified, including:

[0174] Step S402: Obtain statistical results of the number of classified physiological electrical signals corresponding to the target user identifier.

[0175] Step S404 , calculating the covariance matrix to be classified corresponding to the initial physiological electrical signal to be classified.

[0176] Step S406 , correcting the initial reference matrix based on the number statistics result and the covariance matrix to be classified, to obtain a corrected reference matrix corresponding to the initial physiological electrical signal to be classified.

[0177] The physiological electrical signal is a multi-channel signal, including channel signals corresponding to each acquisition channel. The physiological electrical signal acquisition device includes multiple acquisition channels (electrodes), and different acquisition channels are used to acquire physiological electrical signals at different locations. The covariance matrix corresponding to the physiological electrical signal is a matrix composed of the covariances between the signals of each channel in the physiological electrical signal. The covariance matrix to be classified refers to the covariance matrix corresponding to the initial physiological electrical signal to be classified.

[0178] Specifically, the reference matrix can be a covariance matrix, and the covariance matrix of a data can reflect the correlation between data elements. Therefore, when correcting the starting reference matrix, the covariance matrix to be classified corresponding to the initial physiological electrical signal to be classified can be referred to. In addition, the reference matrix is ​​continuously updated based on the new physiological electrical signal. Therefore, when correcting the starting reference matrix, the statistical results of the number of classified physiological electrical signals corresponding to the target user identification can be further referred to. Each time the computer device classifies a physiological electrical signal of the target user, the statistical results of the number of classified physiological electrical signals corresponding to the target user identification are updated in a timely manner.

[0179] When processing the current initial physiological electrical signal to be classified, the computer device can calculate the covariance matrix to be classified corresponding to the initial physiological electrical signal to be classified, obtain the number statistics of the classified physiological electrical signals corresponding to the target user identification, and correct the starting reference matrix based on the number statistics and the covariance matrix to be classified, thereby obtaining the corrected reference matrix corresponding to the initial physiological electrical signal to be classified.

[0180] In one embodiment, the starting reference matrix includes at least one starting reference submatrix corresponding to each target frequency band, and the initial to-be-classified physiological electrical signal includes at least one initial to-be-classified sub-signal corresponding to each target frequency band. Therefore, when modifying the starting reference matrix, the starting reference submatrix corresponding to each target frequency band is modified independently.

[0181] In this embodiment, the starting reference matrix is ​​corrected based on the number statistics result and the covariance matrix to be classified, which can accurately and effectively correct the starting reference matrix, so that data alignment based on the accurate corrected reference matrix helps to improve the classification accuracy of physiological electrical signals.

[0182] In one embodiment, the starting reference matrix includes at least one starting reference sub-matrix corresponding to each target frequency band, and the initial physiological electrical signal to be classified includes at least one initial sub-signal to be classified corresponding to each target frequency band. The starting reference matrix is ​​corrected based on the number statistics and the covariance matrix to be classified to obtain a corrected reference matrix corresponding to the initial physiological electrical signal to be classified, including:

[0183] Based on the initial sub-signals to be classified and the number statistics results corresponding to the same target frequency band, the corresponding starting reference sub-matrix is ​​corrected to obtain the corrected reference sub-matrices corresponding to each target frequency band; and a corrected reference matrix is ​​obtained based on each corrected reference sub-matrix.

[0184] Specifically, if the starting reference matrix includes at least one starting reference submatrix corresponding to each target frequency band, the starting reference submatrix corresponding to each target frequency band needs to be independently corrected. Therefore, the computer device can correct the corresponding starting reference submatrix based on the initial to-be-classified sub-signals and the number statistics corresponding to the same target frequency band, thereby obtaining a corrected reference submatrix corresponding to each target frequency band. The corrected reference submatrices are then combined to form a corrected reference matrix.

[0185] In one embodiment, the starting reference matrix can be corrected using the following formula:

[0186]

[0187] Among them, R j Represents the modified reference submatrix corresponding to the target frequency band j, R′ j represents the starting reference submatrix corresponding to the target frequency band j, N represents the number of statistical results, x j represents the initial sub-signal to be classified corresponding to the target frequency band j, Represents x j The transpose of . Can represent x j The corresponding covariance matrix.

[0188] In this embodiment, the starting reference submatrix corresponding to each target frequency band is corrected independently, which can improve the correction accuracy. Therefore, data alignment and subsequent processing based on the accurate corrected reference matrix can help improve the classification accuracy of physiological electrical signals.

[0189] In one embodiment, the modified reference matrix corresponding to the initial physiological electrical signal to be classified includes at least one modified reference sub-matrix corresponding to each target frequency band, and the initial physiological electrical signal to be classified includes at least one initial sub-signal to be classified corresponding to each target frequency band. Data alignment of the initial physiological electrical signal to be classified is performed based on target signal spatial information corresponding to the target user identifier to obtain the target physiological electrical signal to be classified, including:

[0190] The modified reference sub-matrix corresponding to the same target frequency band and the initial sub-signal to be classified are fused to obtain the target sub-signals to be classified corresponding to each target frequency band; and the target physiological electrical signal to be classified is obtained based on each target sub-signal to be classified.

[0191] Specifically, if the physiological electrical signals are not band-pass filtered during the classification process of the physiological electrical signals, the corrected reference matrix and the initial physiological electrical signals to be classified can be directly fused, so that the initial physiological electrical signals to be classified are data aligned based on the corrected reference matrix to obtain the target physiological electrical signals to be classified. If the physiological electrical signals are band-pass filtered during the classification process of the physiological electrical signals, then when performing data alignment, each target frequency band needs to be independently aligned. If the physiological electrical signals are band-pass filtered, the corrected reference matrix corresponding to the initial physiological electrical signals to be classified includes a corrected reference sub-matrix corresponding to at least one target frequency band, and the initial physiological electrical signals to be classified include at least one initial sub-signal to be classified corresponding to each target frequency band. The computer device can fuse the corrected reference sub-matrix corresponding to the same target frequency band and the initial sub-signal to be classified to obtain the target sub-signals to be classified corresponding to each target frequency band. Then, the target physiological electrical signals to be classified are composed of each target sub-signal.

[0192] In one embodiment, data alignment can be performed using the following formula:

[0193]

[0194] in, represents the target sub-signal to be classified corresponding to the target frequency band j, R j represents the modified reference submatrix corresponding to the target frequency band j, x j Represents the initial sub-signal to be classified corresponding to the target frequency band j.

[0195] In this embodiment, the initial sub-signals to be classified corresponding to each target frequency band are independently data aligned, which can improve the accuracy of data alignment, and thus subsequent processing based on the target physiological electrical signals to be classified composed of each target sub-signal to be classified can help improve the classification accuracy of the physiological electrical signals.

[0196] In one embodiment, Figure 5As shown, the generation of the target spatial filter matrix includes the following steps:

[0197] Step S502: Acquire initial training physiological electrical signals corresponding to a plurality of training user identifiers; the initial training physiological electrical signals carry training labels.

[0198] Step S504 , performing data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier, to obtain target training physiological electrical signals corresponding to each training user identifier.

[0199] Step S506 : generating a target spatial filter matrix based on the signal differences between target training physiological electrical signals corresponding to different training labels.

[0200] Specifically, a target spatial filter matrix applicable to all users can be generated based on the physiological electrical signals of known classification results corresponding to multiple training users. First, the computer device needs to obtain the initial training physiological electrical signals corresponding to multiple training user identifiers, and perform data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier to obtain the target training physiological electrical signals corresponding to each training user identifier, that is, each training user performs data alignment independently. Then, the computer device can generate a target spatial filter matrix based on the signal difference between the target training physiological electrical signals corresponding to different training labels. The target spatial filter matrix can maximize the spatial feature difference between physiological electrical signals of different categories.

[0201] In one embodiment, when the classification task is a binary classification task, the computer device can generate a target spatial filter matrix based on the signal difference between the target training physiological electrical signal corresponding to the training label A and the target training physiological electrical signal corresponding to the training label B. When the classification task is a multi-classification task, the computer device can first convert the multi-classification task into a binary classification task in a one-to-many or many-to-many manner, first generate a corresponding first target spatial filter matrix based on the binary classification task, and then subdivide the binary classification task to generate a corresponding second target spatial filter matrix until it can no longer be subdivided, and finally obtain multiple target spatial filter matrices. Converting a multi-classification task into a binary classification task in a one-to-many manner means selecting one category from all categories as one category and the other categories as another category. Converting a multi-classification task into a binary classification task in a many-to-many manner means selecting a part of all categories as one category and the other part as another category.

[0202] For example, when the classification task is a three-category task, the computer device can generate a first target spatial filter matrix based on the signal difference between the target training physiological electrical signal corresponding to training label A and the target training physiological electrical signal corresponding to other training labels (training label B and training label C). The spatial features obtained by performing spatial feature extraction on the physiological electrical signal based on the first target spatial filter matrix can be used to distinguish whether the category of the physiological electrical signal is training label A. The computer device can generate a second target spatial filter matrix based on the signal difference between the target training physiological electrical signal corresponding to training label B and the target training physiological electrical signal corresponding to training label C. The spatial features obtained by performing spatial feature extraction on the physiological electrical signal based on the second target spatial filter matrix can be used to distinguish whether the category of the physiological electrical signal is training label B or training label C. In specific applications, the spatial features of the physiological electrical signal to be classified are first performed based on the first target spatial filter matrix to obtain the first spatial features to be classified. If the classification result corresponding to the physiological electrical signal to be classified is training label A based on the first spatial features to be classified, then the classification result corresponding to the physiological electrical signal to be classified is the category corresponding to training label A. If the classification result corresponding to the physiological electrical signal to be classified based on the first spatial feature to be classified is not training label A, the physiological electrical signal to be classified is spatially characterized based on the second target spatial filter matrix to obtain a second spatial feature to be classified. Finally, based on the second spatial feature to be classified, it is determined whether the classification result corresponding to the physiological electrical signal to be classified is training label B or training label C.

[0203] In this embodiment, data alignment can be used to reduce the distribution differences of training samples between different training users, and then based on the signal differences between target training physiological electrical signals corresponding to different training labels, a target spatial filter matrix can be generated that maximizes the spatial feature differences between physiological electrical signals of different categories, so that physiological electrical signals can be classified based on the spatial features to be classified extracted by the target spatial filter matrix.

[0204] In one embodiment, before performing data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier to obtain the target training physiological electrical signals corresponding to the respective training user identifiers, the method further includes:

[0205] Based on each initial training physiological electrical signal corresponding to the same training user identifier, a corresponding initial reference matrix is ​​generated to obtain the initial reference matrix corresponding to each training user identifier; and the initial reference matrix corresponding to the same training user identifier is used as the corresponding training signal space information.

[0206] Specifically, before performing data alignment, it is necessary to independently calculate the training signal space information corresponding to each training user. The computer device can generate a corresponding initial reference matrix based on each initial training physiological electrical signal corresponding to the same training user identifier, obtain the initial reference matrix corresponding to each training user identifier, and then use the initial reference matrix corresponding to the same training user identifier as the corresponding training signal space information. For example, the computer device generates an initial reference matrix a corresponding to training user A based on each initial training physiological electrical signal corresponding to training user A, uses the initial reference matrix a as the training signal space information corresponding to training user A, and generates an initial reference matrix b corresponding to training user B based on each initial training physiological electrical signal corresponding to training user B, and uses the initial reference matrix b as the training signal space information corresponding to training user B.

[0207] In one embodiment, the initial reference matrix can be calculated using the following formula:

[0208]

[0209] Where R represents the initial reference matrix corresponding to a training user ID, m represents the total number of initial training physiological electrical signals corresponding to a training user ID, and x i represents the i-th initial training physiological electrical signal, Represents x i The transpose of represents the covariance matrix corresponding to the i-th initial training physiological electrical signal. The initial reference matrix can be the mean of the covariance matrices of all training samples of a training user.

[0210] In this embodiment, an initial reference matrix corresponding to each training user ID is generated based on the initial training electrophysiological signals corresponding to the same training user ID. This initial reference matrix corresponding to the same training user ID is then used as the corresponding training signal spatial information. This reference matrix used for data alignment incorporates relevant information about a large number of electrophysiological signals from the same user, more accurately reflecting the overall distribution of multiple electrophysiological signals corresponding to a single user.

[0211] In one embodiment, the initial training physiological electrical signal includes at least one initial training sub-signal corresponding to each target frequency band, and the initial reference matrix corresponding to each initial training physiological electrical signal corresponding to the same training user identifier is generated based on the initial training physiological electrical signal. The initial reference matrix corresponding to each training user identifier is obtained, including:

[0212] Calculate the initial covariance matrix corresponding to each initial training sub-signal; calculate the corresponding initial reference sub-matrix based on the initial covariance matrices corresponding to the same training user identifier and the same target frequency band, and obtain the initial reference sub-matrix corresponding to each training user identifier in each target frequency band; obtain the initial reference matrix corresponding to each training user identifier based on each initial reference sub-matrix.

[0213] Specifically, when calculating the reference matrix, each target frequency band of each subject is calculated independently. The computer device can perform band-pass filtering on the training physiological electrical signal, and obtain an initial training physiological electrical signal composed of initial training sub-signals corresponding to at least one target frequency band based on the band-pass filtering result. Furthermore, when generating the initial reference matrix, the computer device first calculates the initial covariance matrix corresponding to each initial training sub-signal, and then calculates the corresponding initial reference sub-matrix based on each initial covariance matrix corresponding to the same training user identifier and the same target frequency band, and obtains the initial reference sub-matrix corresponding to each training user identifier in each target frequency band, and then, the initial reference matrix corresponding to each training user identifier is composed of each initial reference sub-matrix. Among them, an initial reference sub-matrix can be the mean of the covariance matrix of all training samples of a training user in a target frequency band, and accordingly, an initial reference matrix can be a combination of the mean of the covariance matrix of all training samples of a training user in each target frequency band.

[0214] In this embodiment, when calculating the reference matrix, each target frequency band of each subject is calculated independently, which can improve the accuracy of the reference matrix, thereby helping to improve the accuracy of subsequent classification of physiological electrical signals.

[0215] In one embodiment, the initial training physiological electrical signal includes channel signals corresponding to multiple acquisition channels on the physiological electrical signal acquisition device, and the initial training sub-signals include channel sub-signals corresponding to each acquisition channel. Calculating the initial covariance matrix corresponding to each initial training sub-signal includes:

[0216] In the current initial training sub-signal, the covariance between each channel sub-signal is calculated; and based on the covariance between each channel sub-signal, an initial covariance matrix corresponding to the current initial training sub-signal is generated.

[0217] The physiological electrical signal acquisition device is used to acquire physiological electrical signals. The physiological electrical signal acquisition device includes multiple electrodes, with different electrodes being used to acquire electrical signals from different locations. Each electrode corresponds to a single acquisition channel. The physiological electrical signals are multi-channel signals, including channel signals corresponding to each acquisition channel. The initial training sub-signals include channel sub-signals corresponding to each acquisition channel.

[0218] Specifically, the physiological electrical signal is a multi-channel signal. The initial training physiological electrical signal includes channel signals corresponding to multiple acquisition channels on the physiological electrical signal acquisition device. The initial covariance matrix corresponding to the initial training physiological electrical signal is a matrix composed of the covariances between the various channel signals. In the current initial training sub-signal, the covariances between the various channel sub-signals are calculated. Then, the initial covariance matrix corresponding to the current initial training sub-signal is composed of the covariances between the various channel sub-signals. By analogy, the initial covariance matrices corresponding to the various initial training sub-signals can be finally obtained.

[0219] In one embodiment, Figure 6 As shown, the initial reference matrix includes at least one initial reference sub-matrix corresponding to each target frequency band, and the initial training physiological electrical signal includes at least one initial training sub-signal corresponding to each target frequency band. Based on the training signal spatial information corresponding to the same training user identifier, data alignment is performed on the corresponding initial training physiological electrical signals to obtain the target training physiological electrical signals corresponding to each training user identifier, including:

[0220] Step S602 : The initial reference sub-matrix and the initial training sub-signal corresponding to the same training user identifier and the same target frequency band are fused to obtain target training sub-signals corresponding to respective training user identifiers in respective target frequency bands.

[0221] Step S604 : obtaining target training physiological electrical signals corresponding to the respective training user identifiers based on the target training sub-signals corresponding to the respective training user identifiers in the respective target frequency bands.

[0222] Specifically, when performing data alignment, each target frequency band of each training user is independently aligned. If the physiological electrical signals are not bandpass filtered during the training process, the initial reference matrix corresponding to the same training user identifier and the initial training physiological electrical signals can be directly fused, thereby performing data alignment on the initial training physiological electrical signals based on the initial reference matrix to obtain the target training physiological electrical signals. If the physiological electrical signals are bandpass filtered during the training process, then when performing data alignment, each target frequency band of each training user needs to be independently aligned. If the physiological electrical signals are bandpass filtered, the initial reference matrix includes at least one initial reference sub-matrix corresponding to each target frequency band, and the initial training physiological electrical signals include at least one initial training sub-signal corresponding to each target frequency band. The computer device can fuse the initial reference sub-matrix and the initial training sub-signal corresponding to the same training user identifier and the same target frequency band to obtain the target training sub-signals corresponding to each training user identifier in each target frequency band. Then, the target training sub-signals corresponding to the same training user identifier are combined to obtain the corresponding target training physiological electrical signals. Ultimately, the target training physiological electrical signals corresponding to each training user identifier can be obtained.

[0223] In this embodiment, when performing data alignment, each target frequency band of each training user is independently aligned, which can improve the accuracy of data alignment and thus help improve the accuracy of subsequent classification of physiological electrical signals.

[0224] In one embodiment, Figure 7 As shown, the target training physiological electrical signal includes at least one target training sub-signal corresponding to each target frequency band, and a target spatial filter matrix is ​​generated based on the signal difference between the target training physiological electrical signals corresponding to different training labels, including:

[0225] Step S702 , in the same target frequency band, generating corresponding target spatial filter submatrices based on signal differences between target training sub-signals corresponding to different training labels, thereby obtaining target spatial filter submatrices corresponding to respective target frequency bands;

[0226] Step S704: Generate a target spatial filter matrix based on each target spatial filter sub-matrix.

[0227] Specifically, a corresponding spatial filter can be generated for each target frequency band. Within the same target frequency band, the computer device can generate a corresponding target spatial filter submatrix based on the signal differences between target training sub-signals corresponding to different training labels, thereby obtaining target spatial filter submatrices corresponding to each target frequency band. The target spatial filter matrix is ​​then formed from the target spatial filter submatrices.

[0228] For example, the target frequency band includes frequency band 1, frequency band 2 and frequency band 3, and each target training physiological electrical signal includes target training sub-signals corresponding to frequency band 1, frequency band 2 and frequency band 3, respectively. In each target training sub-signal corresponding to frequency band 1, the target spatial filter sub-matrix corresponding to frequency band 1 is generated based on the signal difference between the target training sub-signal corresponding to training label A and the target training sub-signal corresponding to training label B. In each target training sub-signal corresponding to frequency band 2, the target spatial filter sub-matrix corresponding to frequency band 2 is generated based on the signal difference between the target training sub-signal corresponding to training label A and the target training sub-signal corresponding to training label B. In each target training sub-signal corresponding to frequency band 3, the target spatial filter sub-matrix corresponding to frequency band 3 is generated based on the signal difference between the target training sub-signal corresponding to training label A and the target training sub-signal corresponding to training label B. Finally, the target spatial filter sub-matrices corresponding to frequency band 1, frequency band 2 and frequency band 3, respectively, constitute the target spatial filter matrix.

[0229] In this embodiment, each target frequency band independently generates a corresponding target spatial filter sub-matrix, and the target spatial filter matrix is ​​composed of each target spatial filter sub-matrix. Then, when applied, spatial feature extraction can be performed by frequency band, thereby improving the accuracy of feature extraction and further improving the classification accuracy of physiological electrical signals.

[0230] In one embodiment, in the same target frequency band, a corresponding target spatial filter submatrix is ​​generated based on the signal difference between target training sub-signals corresponding to different training labels, thereby obtaining target spatial filter submatrices corresponding to each target frequency band, including:

[0231] In the current target frequency band, the target covariance matrix corresponding to each target training sub-signal is calculated; based on the target covariance matrices corresponding to the same training label, the corresponding target reference matrix is ​​calculated to obtain the target reference matrix corresponding to each training label; the target reference matrices are fused to obtain a fused reference matrix, and the fused reference matrix is ​​eigenvalue decomposition is performed on the fused reference matrix to obtain an initial eigenvalue matrix and an initial eigenvector matrix corresponding to the fused reference matrix; a whitening matrix is ​​obtained based on the initial eigenvalue matrix and the initial eigenvector matrix; based on the whitening matrix, a whitening transformation is performed on each target reference matrix to obtain a transformation reference matrix corresponding to each target reference matrix; eigenvalue decomposition is performed on any transformation reference matrix to obtain an eigenvalue decomposition result, and a target eigenvector matrix is ​​obtained based on the eigenvalue decomposition result; based on the whitening matrix and the target eigenvector matrix, a target spatial filter submatrix corresponding to the current target frequency band is generated.

[0232] Eigenvalue decomposition involves decomposing a matrix into multiple eigenvectors. Each eigenvector can be understood as a direction, and the eigenvalue corresponding to that eigenvector is the projection of the matrix in that direction. Eigenvectors with larger eigenvalues ​​are dominant. Whitening is used to remove redundant information from the input data, reducing the correlation between features.

[0233] Specifically, the computer device can generate a spatial filter based on a common spatial pattern. In the current target frequency band, the computer device can first calculate the target covariance matrix corresponding to each target training sub-signal, and calculate the corresponding target reference matrix based on each target covariance matrix corresponding to the same training label, to obtain the target reference matrix corresponding to each training label. When calculating the target reference matrix, the mean of each target covariance matrix corresponding to training label A can be used as the target reference matrix corresponding to training label A, and the mean of each target covariance matrix corresponding to training label B can be used as the target reference matrix corresponding to training label B. Then, the computer device fuses each target reference matrix to obtain a fused reference matrix, and performs eigenvalue decomposition on the fused reference matrix to obtain an initial eigenvalue matrix and an initial eigenvector matrix corresponding to the fused reference matrix. When fusing each target reference matrix, the computer device can specifically add each target reference matrix to obtain a fused reference matrix. Then, the computer device obtains a whitening matrix based on the initial eigenvalue matrix and the initial eigenvector matrix, and performs whitening transformation on each target reference matrix based on the whitening matrix to obtain a transformation reference matrix corresponding to each target reference matrix. Because performing eigenvalue decomposition on each transformation reference matrix separately can obtain an eigenvalue decomposition result including the corresponding data, the computer device can perform eigenvalue decomposition on any transformation reference matrix to obtain an eigenvalue decomposition result, and obtain the target eigenvector matrix based on the eigenvalue decomposition result. Finally, the computer device can generate a target spatial filter submatrix corresponding to the current target frequency band based on the whitening matrix and the target eigenvector matrix. Specifically, the whitening matrix and the target eigenvector matrix can be fused, and the fused matrix can be directly used as the target spatial filter submatrix corresponding to the current target frequency band, or part of the data of the fused matrix can be extracted as the target spatial filter submatrix corresponding to the current target frequency band.

[0234] In one embodiment, a binary classification task is taken as an example to illustrate the process of generating the target spatial filter submatrix corresponding to the current target frequency band.

[0235] 1. Calculate the average covariance matrix of the two types of signals separately and

[0236]

[0237]

[0238] in, represents the average covariance matrix corresponding to training label 1 (i.e., the target reference matrix corresponding to training label 1), represents the average covariance matrix corresponding to training label 2 (i.e., the target reference matrix corresponding to training label 2). M represents the number of training user identifiers, i.e., the number of training users, and m represents the number of target training physiological electrical signals corresponding to each training user identifier, i.e., the number of training samples corresponding to each training user. represents the i-th target training sub-signal, the training label of the target training sub-signal is 1, represents the target covariance matrix corresponding to the i-th target training sub-signal, whose training label is 1. trace(Y) represents the trace of matrix Y, that is, the sum of the diagonal elements of matrix Y. represents the i-th target training sub-signal, the training label of which is 2. represents the target covariance matrix corresponding to the i-th target training sub-signal, and the training label of the target training sub-signal is 2.

[0239] 2. Calculate the composite covariance matrix R and perform eigenvalue decomposition on it

[0240]

[0241] Where R represents the fusion reference matrix, U represents the eigenvector matrix corresponding to the matrix R (that is, the initial eigenvector matrix corresponding to the fusion reference matrix), λ represents the eigenvalue matrix corresponding to the matrix R (that is, the initial eigenvalue matrix corresponding to the fusion reference matrix), λ is a diagonal matrix composed of the eigenvalues ​​corresponding to each eigenvector in the eigenvector matrix, and U T represents the transpose of the matrix U.

[0242] 3. Calculate the whitening matrix P

[0243]

[0244] Wherein, λ′ represents the eigenvalue matrix obtained by arranging the eigenvalues ​​in descending order, i.e., the rearranged λ, and U′ represents the eigenvector matrix corresponding to ′, i.e., the rearranged U.

[0245] 4. Average covariance matrix and Perform whitening transformation and eigenvalue decomposition

[0246]

[0247]

[0248] B1=B2=B

[0249] Among them, S1 represents The transformation reference matrix obtained by whitening transformation, S2 represents the The transformation reference matrix obtained by whitening transformation. P T represents the transpose of matrix P. B1 and λ1 are the eigenvalue decomposition results of S1, and B2 and λ2 are the eigenvalue decomposition results of S2. B is the target eigenvector matrix.

[0250] 5. Calculate the spatial filter, that is, calculate the target spatial filter matrix

[0251] W=B T P

[0252] Among them, W represents the target space filter matrix, B T Represents the transpose of matrix B.

[0253] In this embodiment, a spatial filter is generated based on the common spatial pattern. The spatial filter can maximize the variance between different categories of the mapped samples, thereby achieving the purpose of classification and recognition.

[0254] In one embodiment, generating a target spatial filter submatrix corresponding to the current target frequency band based on the whitening matrix and the target eigenvector matrix includes:

[0255] The whitening matrix and the target eigenvector matrix are fused to obtain an initial spatial filter matrix; at least one initial spatial filter submatrix is ​​extracted from the initial spatial filter matrix to obtain at least one initial spatial filter submatrix; and a target spatial filter submatrix is ​​obtained based on each initial spatial filter submatrix.

[0256] Specifically, to reduce the amount of computation, the computer device can extract partial data from a fusion matrix of a whitening matrix and a target eigenvector matrix as an initial spatial filter submatrix. The computer device first fuses the whitening matrix and the target eigenvector matrix to obtain an initial spatial filter matrix, extracts at least one initial spatial filter submatrix from the initial spatial filter matrix, obtains at least one initial spatial filter submatrix, and then forms a target spatial filter submatrix from each initial spatial filter submatrix. Specifically, a row of data in the initial spatial filter matrix can be used as an initial spatial filter submatrix, or several rows of data in the initial spatial filter matrix can be used as an initial spatial filter submatrix.

[0257] In one embodiment, the first two rows and the last two rows of the matrix W can be selected as initial spatial filter submatrices, and a row of the matrix W can be used as an initial spatial filter submatrix to obtain four initial spatial filter submatrices, and the target spatial filter submatrix is ​​composed of the four initial spatial filter submatrices. That is, a row of the matrix W can be used as a spatial filter, and the first two rows and the last two rows of the matrix W can be selected to obtain four spatial filters, and the spatial filter group is composed of the four spatial filters. The difference between the first several rows of data and the last several rows of data in the matrix W is relatively large, and spatial features of different angles can be extracted from the physiological electrical signals.

[0258] In this embodiment, part of the data is extracted from the fusion matrix of the whitening matrix and the target eigenvector matrix as the initial spatial filter submatrix, and the target spatial filter submatrix is ​​composed of each initial spatial filter submatrix. This can reduce the amount of data in the target spatial filter submatrix, thereby reducing the subsequent calculation amount and improving the classification efficiency of physiological electrical signals.

[0259] In one embodiment, the target spatial filter matrix includes at least one target spatial filter submatrix corresponding to each target frequency band, and the target physiological electrical signal to be classified includes at least one target sub-signal to be classified corresponding to each target frequency band. Spatial features of the target physiological electrical signal to be classified are extracted based on the target spatial filter matrix to obtain spatial features to be classified, including:

[0260] The spatial features of the corresponding target sub-signals to be classified are extracted based on the target spatial filter sub-matrix corresponding to the same target frequency band to obtain the spatial sub-features to be classified corresponding to each target frequency band; and the spatial features to be classified are generated based on each spatial sub-feature to be classified.

[0261] Specifically, based on the training samples, spatial filters corresponding to each target frequency band can be generated. Then, when performing spatial feature extraction, each target frequency band is independently processed. The target spatial filter matrix includes at least one target spatial filter sub-matrix corresponding to each target frequency band, and the target physiological electrical signal to be classified includes at least one target sub-signal to be classified corresponding to each target frequency band. The computer device can extract the spatial features of the corresponding target sub-signals to be classified based on the target spatial filter sub-matrix corresponding to the same target frequency band, thereby obtaining the spatial sub-features to be classified corresponding to each target frequency band. Then, the spatial features to be classified are composed of the various spatial sub-features to be classified. For example, spatial feature extraction is performed on the target sub-signal to be classified corresponding to frequency band 1 based on the target spatial filter sub-matrix corresponding to frequency band 1 to obtain spatial sub-feature 1 to be classified. Spatial feature extraction is performed on the target sub-signal to be classified corresponding to frequency band 2 based on the target spatial filter sub-matrix corresponding to frequency band 2 to obtain spatial sub-feature 2 to be classified. The spatial sub-feature 1 to be classified and the spatial sub-feature 2 to be classified are then concatenated to obtain the spatial feature to be classified.

[0262] In this embodiment, when performing spatial feature extraction, independent calculation of each target frequency band can improve the accuracy and reliability of the spatial features to be classified.

[0263] In one embodiment, the target spatial filter submatrix includes at least one initial spatial filter submatrix, and the spatial features of the corresponding target sub-signals to be classified are extracted based on the target spatial filter submatrix corresponding to the same target frequency band to obtain the spatial sub-features to be classified corresponding to each target frequency band, including:

[0264] In the current target frequency band, signal projection is performed on the corresponding target sub-signals to be classified based on each initial spatial filter sub-matrix to obtain target projection sub-signals corresponding to each target sub-signal to be classified; initial variance data corresponding to each target projection sub-signal is calculated; each initial variance data is normalized to obtain corresponding target variance data; and based on each target variance data, the spatial sub-features to be classified corresponding to the current target frequency band are obtained.

[0265] Specifically, the spatial filter corresponding to a target frequency band can be a spatial filter group. Then, when performing spatial feature extraction, it is necessary to perform spatial feature extraction on the physiological electrical signal based on each spatial filter in the spatial filter group, and obtain the spatial sub-features to be classified based on each spatial feature extraction result. In the current target frequency band, the computer device can perform signal projection on the corresponding target sub-signals to be classified based on each initial spatial filter sub-matrix to obtain the target projection sub-signals corresponding to each target sub-signal to be classified. For example, the target spatial filter sub-matrix corresponding to the current target frequency band includes four initial spatial filter sub-matrices. Based on the initial spatial filter sub-matrix a, the target sub-signals to be classified corresponding to the current target frequency band are projected to obtain target projection sub-signal a. Based on the initial spatial filter sub-matrix b, the target sub-signals to be classified corresponding to the current target frequency band are projected to obtain target projection sub-signal b. Based on the initial spatial filter sub-matrix c, the target sub-signals to be classified corresponding to the current target frequency band are projected to obtain target projection sub-signal c. Based on the initial spatial filter sub-matrix d, the target sub-signals to be classified corresponding to the current target frequency band are projected to obtain target projection sub-signal d. Next, the computer equipment calculates the initial variance data corresponding to each target projection sub-signal, and then normalizes each initial variance data to obtain the target variance data corresponding to each initial variance data. Finally, each target variance data is spliced ​​to obtain the spatial sub-feature to be classified corresponding to the current target frequency band.

[0266] In one embodiment, performing spatial feature extraction on physiological electrical signals based on a spatial filter bank in a target frequency band includes the following steps:

[0267] 1. Calculate sample projection

[0268] Z=W f X

[0269] Among them, Z represents the sample projection result, X represents the training sample, and W f Represents a spatial filter. For example, when X is the target sub-signal to be classified, W f is the initial spatial filter sub-matrix, and Z is the target projection sub-signal.

[0270] 2. Calculate the variance of the sample projection results corresponding to each spatial filter and normalize them

[0271]

[0272] Where F represents the normalized result of the variance, var(Z) represents the variance corresponding to Z, and sum(var(Z)) represents the sum of the variances.

[0273] 3. Splice the normalized results to obtain spatial features

[0274] For example, assuming that the spatial filter group corresponding to a target frequency band includes four spatial filters, sample A can obtain four sample projection results Z1, Z2, Z3, and Z4 through the four spatial filters, and the variance data corresponding to each sample projection result is calculated to obtain the initial variance data V1, V2, V3, and V4. The four initial variances are added to obtain variance statistics, and the ratios of the four initial variances to the variance statistics are calculated respectively to obtain sample features F1, F2, F3, and F4. The four sample features are spliced ​​to finally obtain the spatial sub-features to be classified corresponding to the target frequency band.

[0275] In this embodiment, the spatial filter can maximize the variance between different categories of the mapped samples. Therefore, the sample projection of the physiological electrical signal to be classified is first performed, and then the variance data is calculated, and then normalization and splicing are performed. The spatial features to be classified obtained based on the above processing can be used for classification and recognition to determine the classification results of the physiological electrical signal to be classified.

[0276] In one embodiment, obtaining a classification result corresponding to the initial physiological electrical signal to be classified based on the spatial features to be classified includes:

[0277] The spatial features to be classified are input into the target physiological electrical signal classification model to obtain the classification results.

[0278] The physiological electrical signal classification model is a machine learning model used to classify physiological electrical signals, and the target physiological electrical signal classification model refers to a trained physiological electrical signal classification model.

[0279] Specifically, the computer device can classify the spatial features to be classified based on the machine learning model to obtain a classification result. The computer device can obtain a target physiological electrical signal classification model, input the spatial features to be classified into the target physiological electrical signal classification model, and predict the classification result corresponding to the initial physiological electrical signal to be classified using the target physiological electrical signal classification model.

[0280] In this embodiment, classification processing is performed on the to-be-classified spatial features based on the target physiological electrical signal classification model, so that a relatively accurate classification result can be obtained quickly.

[0281] In one embodiment, the training process of the target physiological electrical signal classification model includes the following steps:

[0282] Based on the target spatial filter matrix, spatial features of each target training physiological electrical signal are extracted to obtain the training spatial features corresponding to each target training physiological electrical signal; each training spatial feature is input into the initial physiological electrical signal classification model to obtain the prediction label corresponding to each target training physiological electrical signal; based on the prediction label and training label corresponding to the same target training physiological electrical signal, the model parameters of the initial physiological electrical signal classification model are adjusted until the convergence condition is met to obtain the target physiological electrical signal classification model.

[0283] The initial physiological electrical signal classification model refers to the physiological electrical signal classification model to be trained, and the target physiological electrical signal classification model refers to the trained physiological electrical signal classification model.

[0284] Specifically, when training the target physiological electrical signal classification model, the computer device can extract spatial features of each target training physiological electrical signal based on the target spatial filter matrix, obtain the training spatial features corresponding to each target training physiological electrical signal, use the training spatial features corresponding to the target training physiological electrical signal as the input of the model, use the training labels corresponding to the target training physiological electrical signal as the expected output of the model, and obtain the trained physiological electrical signal classification model through supervised training. The computer device can specifically input the training spatial features corresponding to the target training physiological electrical signal into the initial physiological electrical signal classification model, obtain the predicted labels corresponding to the target training physiological electrical signals, adjust the model parameters of the initial physiological electrical signal classification model based on the predicted labels and training labels corresponding to the same target training physiological electrical signal, until the convergence conditions are met, and obtain the target physiological electrical signal classification model. The convergence conditions can be customized, such as the number of iterations reaching the iteration threshold, the difference between the training label and the predicted label reaching the minimum value, etc. Adjusting model parameters can specifically involve calculating the difference between the training labels and the predicted labels, adjusting the model parameters of the initial electrophysiological signal classification model through backpropagation of the difference, and continuing training until the updated difference or the number of iterations meets the convergence condition. Training is completed, and a trained electrophysiological signal classification model is obtained. The target electrophysiological signal classification model can be used to classify the spatial features corresponding to the electrophysiological signal to be classified, and obtain a classification result corresponding to the electrophysiological signal to be classified.

[0285] In this embodiment, the training spatial features and training labels corresponding to each target training physiological electrical signal can be trained to obtain a target physiological electrical signal classification model. The target physiological electrical signal classification model can be used to classify the target spatial features corresponding to the physiological electrical signals to be classified, thereby improving the classification efficiency and accuracy of the physiological electrical signals.

[0286] In one embodiment, Figure 8 As shown, a physiological electrical signal classification and processing method is provided, which is applied to Figure 1 The computer device in the example is used to illustrate, and the computer device can be the above Figure 1 The terminal 102 or the server 104 in FIG. Figure 8 , the physiological electrical signal classification and processing method includes the following steps:

[0287] Step S802: Acquire initial training physiological electrical signals corresponding to a plurality of training user identifiers; the initial training physiological electrical signals carry training labels.

[0288] Specifically, the computer device can obtain training samples locally or from other terminals or servers to train the physiological electrical signal classification model. The training samples are multiple initial training physiological electrical signals corresponding to multiple training user identifiers, and each initial training physiological electrical signal carries a corresponding training label.

[0289] Step S804 , performing data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier, to obtain target training physiological electrical signals corresponding to each training user identifier.

[0290] Specifically, the computer device can generate corresponding training signal spatial information based on each initial training electrophysiological signal corresponding to the same training user identifier, thereby obtaining training signal spatial information corresponding to each training user identifier. The computer device can then perform data alignment on the corresponding initial training electrophysiological signals based on the training signal spatial information corresponding to the same training user identifier, thereby obtaining target training electrophysiological signals corresponding to each training user identifier. In other words, data alignment is performed independently for each training user.

[0291] The specific process of generating signal space information and performing data alignment can refer to the methods described in the various related embodiments of the aforementioned physiological electrical signal classification and processing method, and will not be repeated here.

[0292] Step S806 : generating a target spatial filter matrix based on the signal differences between target training physiological electrical signals corresponding to different training labels.

[0293] Specifically, the computer device can generate a target spatial filter matrix based on the signal difference between the target training physiological electrical signals corresponding to different training labels. The target spatial filter matrix can maximize the spatial feature difference between physiological electrical signals of different categories.

[0294] The specific process of generating the target spatial filter matrix may refer to the methods described in the various related embodiments of the aforementioned physiological electrical signal classification and processing method, and will not be repeated here.

[0295] Step S808 : performing spatial feature extraction on each target training physiological electrical signal based on the target spatial filter matrix to obtain training spatial features corresponding to each target training physiological electrical signal.

[0296] Specifically, the computer device can extract spatial features of each target training physiological electrical signal based on the target spatial filter matrix to obtain the training spatial features corresponding to each target training physiological electrical signal. Therefore, a classifier can be trained based on the training spatial features corresponding to different training labels to perform classification processing on the spatial features to be classified corresponding to the physiological electrical signals to be classified and output the classification results.

[0297] The specific process of spatial feature extraction can refer to the methods described in the various related embodiments of the aforementioned physiological electrical signal classification and processing method, and will not be repeated here.

[0298] Step S810 , performing model training on the initial physiological electrical signal classification model based on the training space features and training labels corresponding to each target training physiological electrical signal until a convergence condition is met, thereby obtaining a target physiological electrical signal classification model.

[0299] Specifically, the computer device can use the training space features as the input of the model and the corresponding training labels as the expected output, and obtain the trained physiological electrical signal classification model through supervised training. The computer device can specifically input the training space features corresponding to the target training physiological electrical signals into the initial physiological electrical signal classification model to obtain the predicted labels corresponding to the target training physiological electrical signals, and then adjust the model parameters of the initial physiological electrical signal classification model based on the predicted labels and training labels corresponding to the same target training physiological electrical signals until the convergence conditions are met to obtain the target physiological electrical signal classification model. Among them, the convergence conditions can be customized, such as the number of iterations reaches the iteration threshold, the difference between the training label and the predicted label reaches the minimum value, etc. Adjusting the model parameters can specifically be calculating the difference between the training label and the predicted label, adjusting the model parameters of the initial physiological electrical signal classification model through difference back propagation and continuing training until the updated difference or number of iterations meets the convergence conditions, then the training is completed, and the trained physiological electrical signal classification model is obtained.

[0300] During application, the computer device can obtain the initial physiological electrical signal to be classified corresponding to the target user identifier, perform data alignment on the initial physiological electrical signal to be classified based on the target signal spatial information corresponding to the target user identifier, obtain the target physiological electrical signal to be classified, extract spatial features of the target physiological electrical signal to be classified based on the target spatial filter matrix, obtain the spatial features to be classified, and finally input the spatial features to be classified into the target physiological electrical signal classification model to obtain the classification result corresponding to the initial physiological electrical signal to be classified.

[0301] The specific application process of the target physiological electrical signal classification model may refer to the methods described in the various related embodiments of the aforementioned physiological electrical signal classification processing method, and will not be repeated here.

[0302] The above-mentioned physiological electrical signal classification and processing method performs data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier, which can reduce the distribution differences between the physiological electrical signals of different training users. Then, based on the target physiological electrical signals obtained through data alignment and the corresponding training labels, a universal target spatial filter matrix can be generated. The target spatial filter matrix can be used to extract spatial features in the physiological electrical signals that can be used to distinguish the categories of physiological electrical signals. In this way, without pre-acquiring the physiological electrical signals of the target user, a target spatial filter matrix and a target physiological electrical signal classification model that can be used to classify the physiological electrical signals of the target user can be trained. The target spatial filter matrix and the target physiological electrical signal classification model can be used to classify the physiological electrical signals of the target user, which is more convenient and efficient.

[0303] This application also provides an application scenario, which applies the above-mentioned physiological electrical signal classification and processing method. Specifically, the application of the physiological electrical signal classification and processing method in this application scenario is as follows:

[0304] The physiological electrical signal classification processing method of the present application can be applied to the task of EEG signal classification. Electroencephalogram (EEG) signals are physiological electrical signals obtained by amplifying and recording scalp electrical signals through electronic instruments (i.e., acquisition equipment), which are multi-channel time series. Figure 9A The acquisition device includes multiple electrodes, one electrode corresponds to one acquisition channel, and a complete EEG signal is composed of EEG signals corresponding to multiple acquisition channels. 902 may represent an electrode.

[0305] refer to Figure 9B Explain the specific process of EEG signal classification:

[0306] 1. Offline training

[0307] Offline training is mainly based on a large amount of data from different training users to train a robust classification model so that it has high generalization for the EEG signals of unknown users (i.e., target users). Assume that the training data is X i ∈R m*c*n , where X i represents the training data of the i-th training user, M represents the number of training users, m is the number of training samples for each training user, c is the number of EEG signal acquisition channels, and n is the number of EEG signal sampling points.

[0308] 1-1. Bandpass filtering

[0309] For all training data, the computer device first performs bandpass filtering on the original EEG signal training samples. The filtering frequency bands include multiple, which may or may not overlap. Then, the computer device can obtain EEG signals in multiple target frequency bands. N is the total number of filtering frequency bands.

[0310] 1-2. Data alignment

[0311] For the filtered training samples, data alignment is used to reduce the difference in the covariance matrix of the training samples between different training users. Specifically, Euclidean distance alignment can be used. Euclidean distance alignment is a method based on a reference matrix, and each target frequency band of each training user is calculated independently. Let the specific frequency band sample of a training user be x, x∈R m*c*n Assuming the reference matrix is ​​R, then for each training sample x i ∈R c*n , data alignment can be performed using the following formula:

[0312]

[0313] The reference matrix R is the mean of the covariance matrices of all training samples in each frequency band for each training user and can be calculated using the following formula:

[0314]

[0315] Through Euclidean distance alignment, the average covariance matrix of all training users can be converted into a unit matrix, which is believed to reduce the distribution difference of the covariance matrix between different training users.

[0316] 1-3. Spatial feature extraction

[0317] Through bandpass filtering and data alignment, the computer equipment can obtain training samples of each training user with similar covariance matrix distribution. M is the total number of training users, and N is the total number of target frequency bands. The computer then mixes all the aligned training samples and uses cospatial patterns to extract spatial features for each target frequency band. Cospatial patterns are a spatial feature extraction method based on the covariance matrix. It aims to find an optimal spatial filter that maximizes the variance between different classes of the mapped samples, thereby achieving classification and recognition.

[0318] 1-3-1. Calculate the spatial filter bank (i.e., target spatial filter matrix)

[0319] (1) Calculate the average covariance matrix of the two types of signals respectively and

[0320] (2) Calculate the composite covariance matrix R and perform eigenvalue decomposition on it

[0321] (3) Calculate the whitening matrix P

[0322] (4) Average covariance matrix and Perform whitening transformation and eigenvalue decomposition

[0323] (5) Calculate the spatial filter group, that is, calculate the target spatial filter submatrix

[0324] 1-3-2. Spatial feature extraction based on spatial filter bank

[0325] (1) Calculate the sample projection corresponding to the training sample after data alignment

[0326] (2) Calculate the variance of the sample projection results corresponding to each spatial filter and normalize it

[0327] (3) Concatenate the normalized results to obtain the training space features corresponding to the training samples

[0328] 1-4. Training the classifier

[0329] The logistic regression classifier is trained based on the training space features and training labels corresponding to each training sample.

[0330] 2. Online prediction

[0331] The spatial filter banks and logistic regression classifiers for each frequency band, calculated using offline training, can be applied to online brain-computer interface systems for signal recognition. However, during online prediction, signal samples from unknown users appear individually, making it impossible to calculate a reference matrix. Therefore, a solution can be adopted to gradually modify the reference matrix during system operation to adapt to the data distribution of unknown users.

[0332] 2-1. Bandpass filtering

[0333] First initialize the reference matrix R i =0, i=1, 2, 3, ..., F, the number of samples of the target user at the beginning is N=0, where F is the total number of target frequency bands.

[0334] Assume that the EEG signal to be classified is x∈R c*n , the computer device performs bandpass filtering on the EEG signal to be classified, and obtains the filtered EEG signal to be classified That is, after bandpass filtering the EEG signal to be classified, we can obtain the initial sub-signal to be classified corresponding to the target frequency band 1, the initial sub-signal to be classified corresponding to the target frequency band 2, the initial sub-signal to be classified corresponding to the target frequency band 3, ..., the initial sub-signal to be classified corresponding to the target frequency band f.

[0335] 2-2. Data alignment

[0336] First update the reference matrix And the number of samples of the target user is N = N + 1. Then, the updated reference matrix is ​​used to perform Euclidean distance alignment on the filtered EEG signals to be classified: That is, after data alignment of the EEG signals to be classified, the target sub-signal to be classified corresponding to the target frequency band 1, the target sub-signal to be classified corresponding to the target frequency band 2, the target sub-signal to be classified corresponding to the target frequency band 3, ..., the target sub-signal to be classified corresponding to the target frequency band f can be obtained.

[0337] 2-3. Spatial feature extraction

[0338] For each target frequency band, the EEG signal is extracted based on the trained spatial filter bank to obtain the corresponding spatial sub-features to be classified. The spatial sub-features to be classified for all target frequency bands are then concatenated to obtain the final spatial features to be classified. In other words, after spatially characterizing the EEG signal to be classified, the spatial sub-features to be classified corresponding to target frequency band 1, target frequency band 2, ..., and target frequency band f are obtained. The spatial sub-features to be classified corresponding to target frequency bands 1, 2, ..., and f are concatenated to obtain the spatial features to be classified.

[0339] 2-4. Feature Classification

[0340] Use the trained logistic regression classifier to perform feature classification, input the spatial features to be classified into the trained classifier, and obtain the classification results corresponding to the EEG signals to be classified.

[0341] Repeat steps 2-1 to 2-4 to perform online classification on each EEG signal of the target user to achieve online classification of EEG signals of unknown users.

[0342] In this embodiment, the distribution differences between EEG signals of different users can be reduced, thereby enabling cross-user EEG signal classification. Furthermore, after offline training, the trained parameters can be embedded in an online brain-computer interface system. As signal samples from unknown users are collected, the signal distribution can be adaptively adjusted, thereby enabling online classification of EEG signals from unknown users.

[0343] It is understood that, in addition to being applied to EEG signal classification tasks, the physiological electrical signal classification and processing method of the present application can also be applied to other physiological electrical signal classification tasks, such as ECG signal classification tasks, myoelectric signal classification tasks, etc. For example, when a user is exercising, the user's myoelectric signal can be classified into muscle status. When the muscle status is muscle fatigue, a prompt message is generated to remind the user to take a break in time.

[0344] It should be understood that although Figure 2-8 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. In addition, Figure 2-8 At least part of the steps may include multiple steps or multiple stages. These steps or stages are not necessarily performed at the same time, but can be performed at different times. The order of execution of these steps or stages is not necessarily one by one, but can be performed in turn or alternately with other steps or at least part of the steps or stages in other steps.

[0345] In one embodiment, Figure 10 As shown, a physiological electrical signal classification and processing device is provided. The device can be a software module or a hardware module, or a combination of the two to form a part of a computer device. The device specifically includes: a signal acquisition module 1002, a data alignment module 1004, a feature extraction module 1006 and a signal classification module 1008, wherein:

[0346] The signal acquisition module 1002 is used to acquire the initial physiological electrical signal to be classified corresponding to the target user identification;

[0347] The data alignment module 1004 is configured to perform data alignment on the initial physiological electrical signal to be classified based on the target signal space information corresponding to the target user identifier to obtain the target physiological electrical signal to be classified;

[0348] A feature extraction module 1006 is configured to extract spatial features of the target physiological electrical signal to be classified based on a target spatial filter matrix to obtain spatial features to be classified, wherein the target spatial filter matrix is ​​generated based on target training physiological electrical signals corresponding to multiple training user identifiers and training labels corresponding to each target training physiological electrical signal, and the target training physiological electrical signal is obtained by aligning the initial training physiological electrical signal with the spatial information of the training signal corresponding to the training user identifier;

[0349] The signal classification module 1008 is configured to obtain a classification result corresponding to the initial physiological electrical signal to be classified based on the spatial features to be classified.

[0350] In one embodiment, the signal acquisition module is also used to obtain candidate physiological electrical signals to be classified corresponding to the target user identifier; perform signal extraction of at least one target frequency band on the candidate physiological electrical signals to be classified to obtain initial sub-signals to be classified corresponding to the candidate physiological electrical signals to be classified in each target frequency band; and obtain an initial physiological electrical signal to be classified based on each initial sub-signal to be classified.

[0351] In one embodiment, the data alignment module is also used to obtain a starting reference matrix corresponding to the initial physiological electrical signal to be classified; correct the starting reference matrix based on the initial physiological electrical signal to be classified to obtain a corrected reference matrix corresponding to the initial physiological electrical signal to be classified; and use the corrected reference matrix corresponding to the initial physiological electrical signal to be classified as the target signal spatial information.

[0352] In one embodiment, the initial reference matrix is ​​a modified reference matrix corresponding to the last physiological electrical signal to be classified corresponding to the target user identifier.

[0353] In one embodiment, the data alignment module is also used to obtain the number statistics of the classified physiological electrical signals corresponding to the target user identification; calculate the covariance matrix to be classified corresponding to the initial physiological electrical signals to be classified; correct the starting reference matrix based on the number statistics and the covariance matrix to be classified to obtain the corrected reference matrix corresponding to the initial physiological electrical signals to be classified.

[0354] In one embodiment, the starting reference matrix includes at least one starting reference submatrix corresponding to each target frequency band, and the initial physiological electrical signal to be classified includes at least one initial sub-signal to be classified corresponding to each target frequency band. The data alignment module is further configured to modify the corresponding starting reference submatrix based on the initial sub-signals to be classified corresponding to the same target frequency band and the number statistics result, thereby obtaining modified reference submatrices corresponding to each target frequency band; and to obtain a modified reference matrix based on each modified reference submatrix.

[0355] In one embodiment, the corrected reference matrix corresponding to the initial physiological electrical signal to be classified includes at least one corrected reference sub-matrix corresponding to each target frequency band, and the initial physiological electrical signal to be classified includes at least one initial sub-signal to be classified corresponding to each target frequency band. The data alignment module is further configured to fuse the corrected reference sub-matrix corresponding to the same target frequency band with the initial sub-signal to be classified to obtain target sub-signals to be classified corresponding to each target frequency band; and to obtain a target physiological electrical signal to be classified based on each target sub-signal to be classified.

[0356] In one embodiment, Figure 11 As shown, the device also includes:

[0357] The spatial filter matrix generation module 1000 is used to obtain initial training physiological electrical signals corresponding to multiple training user identifiers; the initial training physiological electrical signals carry training labels; based on the training signal spatial information corresponding to the same training user identifier, the corresponding initial training physiological electrical signals are data aligned to obtain target training physiological electrical signals corresponding to each training user identifier; based on the signal difference between the target training physiological electrical signals corresponding to different training labels, a target spatial filter matrix is ​​generated.

[0358] In one embodiment, the spatial filter matrix generation module is also used to generate corresponding initial reference matrices based on the initial training physiological electrical signals corresponding to the same training user identifier, and obtain the initial reference matrices corresponding to the respective training user identifiers; and use the initial reference matrices corresponding to the same training user identifier as the corresponding training signal spatial information.

[0359] In one embodiment, the initial training physiological electrical signal includes at least one initial training sub-signal corresponding to each target frequency band. The spatial filter matrix generation module is further configured to calculate an initial covariance matrix corresponding to each initial training sub-signal; calculate a corresponding initial reference sub-matrix based on each initial covariance matrix corresponding to the same training user identifier and the same target frequency band, thereby obtaining an initial reference sub-matrix corresponding to each training user identifier in each target frequency band; and obtain an initial reference matrix corresponding to each training user identifier based on each initial reference sub-matrix.

[0360] In one embodiment, the initial training physiological electrical signal includes channel signals corresponding to multiple acquisition channels on the physiological electrical signal acquisition device, and the initial training sub-signals include channel sub-signals corresponding to each acquisition channel. The spatial filter matrix generation module is further configured to calculate the covariance between each channel sub-signal in the current initial training sub-signal; and generate an initial covariance matrix corresponding to the current initial training sub-signal based on the covariance between each channel sub-signal.

[0361] In one embodiment, the initial reference matrix includes at least one initial reference sub-matrix corresponding to each target frequency band, and the initial training physiological electrical signal includes at least one initial training sub-signal corresponding to each target frequency band. The spatial filter matrix generation module is further configured to fuse the initial reference sub-matrix and the initial training sub-signal corresponding to the same training user identifier and the same target frequency band to obtain target training sub-signals corresponding to each training user identifier in each target frequency band; and based on the target training sub-signals corresponding to each training user identifier in each target frequency band, obtain target training physiological electrical signals corresponding to each training user identifier.

[0362] In one embodiment, the target training physiological electrical signal includes at least one target training sub-signal corresponding to each target frequency band. The spatial filter matrix generation module is further configured to generate corresponding target spatial filter sub-matrices within the same target frequency band based on signal differences between target training sub-signals corresponding to different training labels, thereby obtaining target spatial filter sub-matrices corresponding to each target frequency band; and generate a target spatial filter matrix based on each target spatial filter sub-matrix.

[0363] In one embodiment, the spatial filter matrix generation module is also used to calculate the target covariance matrix corresponding to each target training sub-signal in the current target frequency band; calculate the corresponding target reference matrix based on each target covariance matrix corresponding to the same training label to obtain the target reference matrix corresponding to each training label; fuse the target reference matrices to obtain a fused reference matrix, perform eigenvalue decomposition on the fused reference matrix to obtain an initial eigenvalue matrix and an initial eigenvector matrix corresponding to the fused reference matrix; obtain a whitening matrix based on the initial eigenvalue matrix and the initial eigenvector matrix; perform a whitening transformation on each target reference matrix based on the whitening matrix to obtain a transformation reference matrix corresponding to each target reference matrix; perform eigenvalue decomposition on any one of the transformation reference matrices to obtain an eigenvalue decomposition result, and obtain a target eigenvector matrix based on the eigenvalue decomposition result; generate a target spatial filter submatrix corresponding to the current target frequency band based on the whitening matrix and the target eigenvector matrix.

[0364] In one embodiment, the spatial filter matrix generation module is further used to fuse the whitening matrix and the target eigenvector matrix to obtain an initial spatial filter matrix; extract at least one initial spatial filter submatrix from the initial spatial filter matrix to obtain at least one initial spatial filter submatrix; and obtain a target spatial filter submatrix based on each initial spatial filter submatrix.

[0365] In one embodiment, the target spatial filter matrix includes at least one target spatial filter submatrix corresponding to each target frequency band, and the target physiological electrical signal to be classified includes at least one target sub-signal to be classified corresponding to each target frequency band. The feature extraction module is further configured to extract spatial features of the corresponding target sub-signals to be classified based on the target spatial filter submatrix corresponding to the same target frequency band, thereby obtaining spatial sub-features to be classified corresponding to each target frequency band; and generate spatial features to be classified based on each spatial sub-feature to be classified.

[0366] In one embodiment, the target spatial filter submatrix includes at least one initial spatial filter submatrix. The feature extraction module is further configured to perform signal projection on the corresponding target sub-signals to be classified in the current target frequency band based on each initial spatial filter submatrix to obtain target projection sub-signals corresponding to each target sub-signal to be classified; calculate initial variance data corresponding to each target projection sub-signal; normalize each initial variance data to obtain corresponding target variance data; and obtain the spatial sub-features to be classified corresponding to the current target frequency band based on each target variance data.

[0367] In one embodiment, the signal classification module is further configured to input the spatial features to be classified into the target physiological electrical signal classification model to obtain a classification result.

[0368] In one embodiment, Figure 12 As shown, the device also includes:

[0369] The model training module 1001 is used to extract spatial features of each target training physiological electrical signal based on the target spatial filter matrix to obtain the training spatial features corresponding to each target training physiological electrical signal; input each training spatial feature into the initial physiological electrical signal classification model to obtain the prediction labels corresponding to each target training physiological electrical signal; adjust the model parameters of the initial physiological electrical signal classification model based on the prediction labels and training labels corresponding to the same target training physiological electrical signal until the convergence conditions are met to obtain the target physiological electrical signal classification model.

[0370] The above-mentioned physiological electrical signal classification and processing device first performs data alignment on the corresponding initial physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier, which can reduce the distribution differences between the physiological electrical signals of different training users. Then, a universal target spatial filter matrix can be generated based on the target physiological electrical signal obtained by data alignment and the corresponding training label. The target spatial filter matrix can be used to extract spatial features in the physiological electrical signal that can be used to distinguish the categories of physiological electrical signals. Then, when classifying the physiological electrical signals of unknown users, the initial physiological electrical signals to be classified corresponding to the target user identifier are first data aligned based on the target signal spatial information corresponding to the target user identifier to reduce the distribution differences between the physiological electrical signals of the target user and the training user. Then, the spatial features of the target physiological electrical signals to be classified obtained by data alignment are extracted based on the universal target spatial filter matrix, so that the classification results corresponding to the initial physiological electrical signals to be classified can be obtained based on the extracted spatial features to be classified. In this way, the physiological electrical signals of the target user can be classified without obtaining the physiological electrical signals of the target user in advance, which is more convenient and efficient.

[0371] In one embodiment, Figure 13As shown, a physiological electrical signal classification and processing device is provided. The device can adopt a software module or a hardware module, or a combination of the two to become part of a computer device. The device specifically includes: a signal acquisition module 1302, a data alignment module 1304, a spatial filter matrix generation module 1306, a feature extraction module 1308 and a model training module 1310, wherein:

[0372] The signal acquisition module 1302 is configured to acquire initial training physiological electrical signals corresponding to a plurality of training user identifiers; the initial training physiological electrical signals carry training labels;

[0373] The data alignment module 1304 is configured to perform data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier, so as to obtain target training physiological electrical signals corresponding to each training user identifier.

[0374] A spatial filter matrix generating module 1306 is configured to generate a target spatial filter matrix based on signal differences between target training physiological electrical signals corresponding to different training labels;

[0375] A feature extraction module 1308 is configured to extract spatial features of each target training physiological electrical signal based on the target spatial filter matrix to obtain training spatial features corresponding to each target training physiological electrical signal;

[0376] The model training module 1310 is used to perform model training on the initial physiological electrical signal classification model based on the training space features and training labels corresponding to each target training physiological electrical signal until the convergence condition is met, thereby obtaining the target physiological electrical signal classification model.

[0377] The above-mentioned physiological electrical signal classification and processing device performs data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier, which can reduce the distribution differences between the physiological electrical signals of different training users. Then, based on the target physiological electrical signals obtained through data alignment and the corresponding training labels, a universal target spatial filter matrix can be generated. The target spatial filter matrix can be used to extract spatial features in the physiological electrical signals that can be used to distinguish the categories of physiological electrical signals. In this way, without pre-acquiring the physiological electrical signals of the target user, a target spatial filter matrix and a target physiological electrical signal classification model that can be used to classify the physiological electrical signals of the target user can be trained. The target spatial filter matrix and the target physiological electrical signal classification model can be used to classify the physiological electrical signals of the target user, which is more convenient and efficient.

[0378] For the specific definition of the physiological electrical signal classification and processing device, please refer to the definition of the physiological electrical signal classification and processing method above, which will not be repeated here. The various modules in the above-mentioned physiological electrical signal classification and processing device can be implemented in whole or in part by software, hardware, and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0379] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 14 As shown. The computer device includes a processor, a memory and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store target spatial filter matrix, target signal spatial information, and target physiological electrical signal classification model data. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a physiological electrical signal classification processing method is implemented.

[0380] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 15 As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a method for classifying and processing physiological electrical signals is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad provided on the computer device housing, or an external keyboard, touchpad or mouse.

[0381] Those skilled in the art will understand that Figure 14 、 15The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0382] In one embodiment, a computer device is further provided, including a memory and a processor. The memory stores a computer program, and the processor implements the steps in the above method embodiments when executing the computer program.

[0383] In one embodiment, a computer-readable storage medium is provided, storing a computer program, which implements the steps in the above-mentioned method embodiments when executed by a processor.

[0384] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the steps of each of the above-described method embodiments.

[0385] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0386] The technical features of the above embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0387] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A method for classifying and processing physiological electrical signals, characterized in that: The method comprises: Obtaining an initial to-be-classified physiological electrical signal corresponding to the target user identifier; Obtaining a starting reference matrix corresponding to the initial physiological electrical signal to be classified; Obtaining a statistical result of the number of classified physiological electrical signals corresponding to the target user identifier, calculating a covariance matrix to be classified corresponding to the initial physiological electrical signals to be classified, and correcting the starting reference matrix based on the statistical result and the covariance matrix to be classified to obtain a corrected reference matrix corresponding to the initial physiological electrical signals to be classified; Using the corrected reference matrix as target signal space information corresponding to the target user identifier; Performing data alignment on the initial physiological electrical signal to be classified based on the target signal spatial information to obtain a target physiological electrical signal to be classified; Performing spatial feature extraction on the target physiological electrical signal to be classified based on a target spatial filter matrix to obtain a spatial feature to be classified, wherein the target spatial filter matrix is ​​generated based on target training physiological electrical signals corresponding to multiple training user identifiers and training labels corresponding to each target training physiological electrical signal, and the target training physiological electrical signal is obtained by performing data alignment on the initial training physiological electrical signal based on the training signal spatial information corresponding to the training user identifier; A classification result corresponding to the initial physiological electrical signal to be classified is obtained based on the spatial feature to be classified.

2. The method according to claim 1, characterized in that The step of obtaining the initial physiological electrical signal to be classified corresponding to the target user identifier includes: Obtain candidate physiological electrical signals to be classified corresponding to the target user identifier; Performing signal extraction of at least one target frequency band on the candidate physiological electrical signal to be classified to obtain initial sub-signals to be classified corresponding to the candidate physiological electrical signal to be classified in each target frequency band; An initial physiological electrical signal to be classified corresponding to the target user identifier is obtained based on each initial sub-signal to be classified.

3. The method according to claim 1, characterized in that The initial reference matrix is ​​a modified reference matrix corresponding to the last physiological electrical signal to be classified corresponding to the target user identifier.

4. The method according to claim 1, wherein The initial reference matrix includes at least one initial reference sub-matrix corresponding to each target frequency band, and the initial physiological electrical signal to be classified includes at least one initial sub-signal to be classified corresponding to each target frequency band; The step of correcting the initial reference matrix based on the number statistics result and the covariance matrix to be classified to obtain a corrected reference matrix corresponding to the initial physiological electrical signal to be classified includes: Based on the initial to-be-classified sub-signals corresponding to the target frequency band and the number statistics result, the starting reference sub-matrix corresponding to the target frequency band is modified to obtain modified reference sub-matrices corresponding to each target frequency band; A revised reference matrix corresponding to the initial physiological electrical signal to be classified is obtained based on each revised reference sub-matrix.

5. The method according to claim 1, wherein The modified reference matrix corresponding to the initial physiological electrical signal to be classified includes at least one modified reference sub-matrix corresponding to each target frequency band, and the initial physiological electrical signal to be classified includes at least one initial sub-signal to be classified corresponding to each target frequency band; The step of performing data alignment on the initial physiological electrical signal to be classified based on the target signal spatial information to obtain the target physiological electrical signal to be classified includes: The modified reference sub-matrix corresponding to the same target frequency band and the initial sub-signal to be classified are fused to obtain the target sub-signals to be classified corresponding to each target frequency band; The target physiological electrical signal to be classified is obtained based on each target sub-signal to be classified.

6. The method according to claim 1, characterized in that The generation of the target spatial filter matrix comprises the following steps: Acquire initial training physiological electrical signals corresponding to a plurality of training user identifiers, wherein the initial training physiological electrical signals carry training labels; Performing data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier to obtain target training physiological electrical signals corresponding to each training user identifier; The target spatial filter matrix is ​​generated based on signal differences between target training physiological electrical signals corresponding to different training labels.

7. The method according to claim 6, characterized in that Before performing data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier to obtain the target training physiological electrical signals corresponding to each training user identifier, the method further includes: Generate corresponding initial reference matrices based on the initial training physiological electrical signals corresponding to the same training user identifier, and obtain initial reference matrices corresponding to the respective training user identifiers; The initial reference matrix corresponding to the same training user identifier is used as the corresponding training signal space information.

8. The method according to claim 7, characterized in that The initial training physiological electrical signal includes at least one initial training sub-signal corresponding to each target frequency band. The initial reference matrix corresponding to each initial training physiological electrical signal corresponding to the same training user identifier is generated based on the initial training physiological electrical signal, and the initial reference matrix corresponding to each training user identifier is obtained, including: Calculate the initial covariance matrix corresponding to each initial training sub-signal; Calculate the corresponding initial reference submatrix based on each initial covariance matrix corresponding to the same training user identifier and the same target frequency band, and obtain the initial reference submatrix corresponding to each training user identifier in each target frequency band; Based on each initial reference sub-matrix, an initial reference matrix corresponding to each training user identifier is obtained.

9. The method according to claim 8, characterized in that The initial training physiological electrical signal includes channel signals corresponding to multiple acquisition channels on the physiological electrical signal acquisition device, and the initial training sub-signal includes channel sub-signals corresponding to each acquisition channel; The calculating of the initial covariance matrices corresponding to the initial training sub-signals includes: In the current initial training sub-signal, the covariance between the sub-signals of each channel is calculated; An initial covariance matrix corresponding to the current initial training sub-signal is generated based on the covariance between the respective channel sub-signals.

10. The method according to claim 7, characterized in that The initial reference matrix includes at least one initial reference sub-matrix corresponding to each target frequency band, and the initial training physiological electrical signal includes at least one initial training sub-signal corresponding to each target frequency band; The step of performing data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier to obtain target training physiological electrical signals corresponding to each training user identifier includes: The initial reference sub-matrix and the initial training sub-signal corresponding to the same training user identifier and the same target frequency band are fused to obtain the target training sub-signals corresponding to each training user identifier in each target frequency band; Based on the target training sub-signals corresponding to the respective training user identifiers in the respective target frequency bands, the respective target training physiological electrical signals corresponding to the respective training user identifiers are obtained.

11. The method according to claim 6, characterized in that The target training physiological electrical signal includes at least one target training sub-signal corresponding to each target frequency band. The generating of the target spatial filter matrix based on the signal difference between the target training physiological electrical signals corresponding to different training labels includes: In the same target frequency band, the corresponding target spatial filter submatrix is ​​generated based on the signal difference between the target training sub-signals corresponding to different training labels, thereby obtaining the target spatial filter submatrix corresponding to each target frequency band; The target spatial filter matrix is ​​generated based on each target spatial filter sub-matrix.

12. The method according to claim 11, characterized in that The method of generating a corresponding target spatial filter submatrix based on the signal difference between target training sub-signals corresponding to different training labels in the same target frequency band, thereby obtaining target spatial filter submatrices corresponding to each target frequency band, includes: In the current target frequency band, calculate the target covariance matrix corresponding to each target training sub-signal; Calculate the corresponding target reference matrix based on the target covariance matrix corresponding to the same training label, and obtain the target reference matrix corresponding to each training label; fusing the target reference matrices to obtain a fused reference matrix, performing eigenvalue decomposition on the fused reference matrix to obtain an initial eigenvalue matrix and an initial eigenvector matrix corresponding to the fused reference matrix; Obtaining a whitening matrix based on the initial eigenvalue matrix and the initial eigenvector matrix; Performing whitening transformation on each target reference matrix based on the whitening matrix to obtain a transformation reference matrix corresponding to each target reference matrix; Performing eigenvalue decomposition on any transformation reference matrix to obtain an eigenvalue decomposition result, and obtaining a target eigenvector matrix based on the eigenvalue decomposition result; A target spatial filter submatrix corresponding to the current target frequency band is generated based on the whitening matrix and the target eigenvector matrix.

13. The method according to claim 12, characterized in that The generating the target spatial filter submatrix corresponding to the current target frequency band based on the whitening matrix and the target eigenvector matrix includes: Fusing the whitening matrix and the target eigenvector matrix to obtain an initial spatial filter matrix; Extracting at least one initial spatial filter submatrix from the initial spatial filter matrix to obtain at least one initial spatial filter submatrix; A target spatial filter submatrix corresponding to the current target frequency band is obtained based on each initial spatial filter submatrix.

14. The method according to claim 1, wherein The target spatial filter matrix includes at least one target spatial filter sub-matrix corresponding to each target frequency band, and the target physiological electrical signal to be classified includes at least one target sub-signal to be classified corresponding to each target frequency band; The step of extracting spatial features of the target physiological electrical signal to be classified based on the target spatial filter matrix to obtain spatial features to be classified includes: Based on the target spatial filter submatrix corresponding to the same target frequency band, the spatial features of the corresponding target sub-signal to be classified are extracted to obtain the spatial sub-features to be classified corresponding to each target frequency band; The spatial features to be classified are generated based on the respective spatial sub-features to be classified.

15. The method according to claim 14, characterized in that The target spatial filter submatrix includes at least one initial spatial filter submatrix, and the target spatial filter submatrix corresponding to the same target frequency band is used to extract the spatial features of the corresponding target sub-signals to be classified, so as to obtain the spatial sub-features to be classified corresponding to each target frequency band, including: In the current target frequency band, based on each initial spatial filter sub-matrix, the corresponding target sub-signals to be classified are projected to obtain target projection sub-signals corresponding to each target sub-signal to be classified; Calculate the initial variance data corresponding to each target projection sub-signal; Normalize each initial variance data separately to obtain the corresponding target variance data; The spatial sub-features to be classified corresponding to the current target frequency band are obtained based on the respective target variance data.

16. The method according to claim 1, characterized in that The obtaining of a classification result corresponding to the initial physiological electrical signal to be classified based on the spatial feature to be classified includes: The spatial features to be classified are input into a target physiological electrical signal classification model to obtain a classification result corresponding to the initial physiological electrical signal to be classified.

17. The method according to claim 16, characterized in that The training process of the target physiological electrical signal classification model includes the following steps: Performing spatial feature extraction on each target training physiological electrical signal based on the target spatial filter matrix to obtain training spatial features corresponding to each target training physiological electrical signal; Inputting each training spatial feature into the initial physiological electrical signal classification model to obtain the prediction label corresponding to each target training physiological electrical signal; The model parameters of the initial physiological electrical signal classification model are adjusted based on the prediction label and the training label corresponding to the same target training physiological electrical signal until a convergence condition is met, thereby obtaining the target physiological electrical signal classification model.

18. A method for classifying and processing physiological electrical signals, characterized in that: The method comprises: Acquire initial training physiological electrical signals corresponding to a plurality of training user identifiers, wherein the initial training physiological electrical signals carry training labels; Obtaining a starting reference matrix corresponding to the initial training physiological electrical signals corresponding to the training user identifier, and obtaining a statistical result of the number of classified physiological electrical signals corresponding to the training user identifier; Calculating the covariance matrix to be classified corresponding to the initial training physiological electrical signal; Modify the initial reference matrix based on the number statistics result and the covariance matrix to be classified to obtain a modified reference matrix corresponding to the initial training physiological electrical signal; Using the corrected reference matrix as training signal space information corresponding to the training user identifier; Performing data alignment on the corresponding initial training physiological electrical signals based on the training signal spatial information corresponding to the same training user identifier to obtain target training physiological electrical signals corresponding to each training user identifier; Generate a target spatial filter matrix based on the signal difference between target training physiological electrical signals corresponding to different training labels; Extracting spatial features of each target training physiological electrical signal based on the target spatial filter matrix to obtain training spatial features corresponding to each target training physiological electrical signal; The initial physiological electrical signal classification model is trained based on the training space features and training labels corresponding to each target training physiological electrical signal until the convergence condition is met, thereby obtaining the target physiological electrical signal classification model.

19. A physiological electrical signal classification and processing device, characterized in that: The device comprises: A signal acquisition module is used to acquire an initial physiological electrical signal to be classified corresponding to the target user identification; a data alignment module, configured to obtain a starting reference matrix corresponding to the initial physiological electrical signal to be classified, obtain a number statistical result of the classified physiological electrical signals corresponding to the target user identifier, calculate a covariance matrix to be classified corresponding to the initial physiological electrical signal to be classified, correct the starting reference matrix based on the number statistical result and the covariance matrix to be classified, obtain a corrected reference matrix corresponding to the initial physiological electrical signal to be classified, use the corrected reference matrix as the target signal space information corresponding to the target user identifier, perform data alignment on the initial physiological electrical signal to be classified based on the target signal space information, and obtain a target physiological electrical signal to be classified; a feature extraction module for extracting spatial features of the target physiological electrical signal to be classified based on a target spatial filtering matrix to obtain spatial features to be classified, wherein the target spatial filtering matrix is ​​generated based on target training physiological electrical signals corresponding to multiple training user identifiers and training labels corresponding to each target training physiological electrical signal, and the target training physiological electrical signal is obtained by aligning the initial training physiological electrical signal with the training signal spatial information corresponding to the training user identifier; The signal classification module is used to obtain a classification result corresponding to the initial physiological electrical signal to be classified based on the spatial feature to be classified.

20. A physiological electrical signal classification and processing device, characterized in that: The device comprises: A signal acquisition module, configured to acquire initial training physiological electrical signals corresponding to a plurality of training user identifiers; the initial training physiological electrical signals carry training labels; A data alignment module is used to obtain a starting reference matrix corresponding to the initial training physiological electrical signal corresponding to the training user identifier, and obtain a number statistical result of the classified physiological electrical signal corresponding to the training user identifier; calculate a covariance matrix to be classified corresponding to the initial training physiological electrical signal; correct the starting reference matrix based on the number statistical result and the covariance matrix to be classified to obtain a corrected reference matrix corresponding to the initial training physiological electrical signal; use the corrected reference matrix as the training signal space information corresponding to the training user identifier; perform data alignment on the corresponding initial training physiological electrical signals based on the training signal space information corresponding to the same training user identifier to obtain target training physiological electrical signals corresponding to each training user identifier; A spatial filter matrix generation module is used to generate a target spatial filter matrix based on the signal difference between target training physiological electrical signals corresponding to different training labels; A feature extraction module is used to extract spatial features of each target training physiological electrical signal based on the target spatial filter matrix to obtain training spatial features corresponding to each target training physiological electrical signal; The model training module is used to train the initial physiological electrical signal classification model based on the training space features and training labels corresponding to each target training physiological electrical signal until the convergence condition is met to obtain the target physiological electrical signal classification model.

21. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 18 are implemented.

22. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 18 are implemented.

23. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 18 are implemented.

Images