BHS-based motor imagery electroencephalogram signal analysis method and system

Abstract

The invention discloses a BHS-based motor imagery electroencephalogram signal analysis method and system, and the method comprises the steps: carrying out the offline experiment of a to-be-tested target through a motor imagery brain-computer interface experiment normal form, and collecting the offline experiment electroencephalogram data; based on a BHS electroencephalogram channel selection method, performing electroencephalogram optimal channel selection for a specific subject on offline experiment electroencephalogram data to obtain a common spatial filter and a classification model; carrying out online verification, collecting optimal channel electroencephalogram data in real time and carrying out preprocessing, extracting electroencephalogram features according to a spatial filter and inputting the electroencephalogram features into a classification model for mode recognition; if the identification is wrong, returning to continue collection; and if the recognition is correct, converting to form a control instruction for driving the hand robot to act. According to the method, personalized and accurate intervention and control can be provided for the hand robot, the number of electroencephalogram signal channels is reduced by adopting a BHS channel selection method, the cost of a brain-computer interface system is controlled, and the efficiency and the user experience are improved.

Description

Technical Field

[0001] The invention belongs to the technical field of data analysis and processing, and in particular relates to a method and system for analyzing motor imagery electroencephalogram signals based on BHS. Background Art

[0002] Stroke is currently the leading cause of death and disability in adults. The number of stroke patients is large and on the rise. Although stroke patients currently receive intervention, many still have significant motor dysfunction, of which upper limb motor dysfunction accounts for about 44% and hand dysfunction accounts for about 52%.

[0003] Based on this, brain-computer interface provides a multi-modal and multi-feedback intervention method for stroke patients. Compared with previous methods, brain-computer interface can enable patients to actively participate in the training process with their intention to move. This active neural method is conducive to intervening in their motor function. Brain-computer interface technology has been widely introduced into the control technology of hand robots. Most of the existing brain-computer interface control hand robot systems use the commonly used 16-lead, 32-lead or 64-lead. Because of the large number of leads, the experimental preparation time is long, the calculation cost is high, and the comfort of the subjects is poor. Channel selection of leads can solve the above shortcomings.

[0004] As an improvement, the channel selection method can make the brain-computer interface system use as few channels as possible while ensuring ideal performance, which will reduce computing costs, reduce experimental preparation time, and enhance the comfort of subjects. The existing channel selection methods mainly include filtering selection and wrapping selection. The principle of the filtering selection method is to select EEG channels based on mathematical statistical methods, such as mutual information, correlation, Fisher criterion, etc., extract EEG data features based on the selected channels and train the classifier. The wrapping selection method is to design a fitness function related to the number of channels and classification accuracy, and then iterate through an intelligent optimization algorithm (such as genetic algorithm, particle swarm algorithm, artificial bee colony algorithm, etc.) to finally select the EEG channel corresponding to the highest classification accuracy and the least number of channels as the optimal channel combination. In this process, the selection process is related to the classifier performance, while the channel selection process in the filtering selection method is not related to the classifier performance, which will result in the final selection of the optimal channel cannot obtain the optimal classification effect. In addition, most of the intelligent optimization algorithms used in the existing wrapping selection methods are complex in principle and have high iteration time cost. In summary, in the current channel selection method of brain-computer interface intervention, the channel selection process is not related to the classifier performance, resulting in the final optimal channel not being able to obtain the optimal classification effect, and the optimization algorithm principle used is relatively complex and the iteration efficiency is low. Summary of the invention

[0005] The present invention provides a method and system for analyzing motor imagery EEG signals based on BHS, aiming to solve the problems in the current channel selection method of brain-computer interface intervention mode, that the channel selection process is irrelevant to the classifier performance, resulting in that the optimal channel finally selected cannot obtain the optimal classification effect, and the optimization algorithm principle adopted is relatively complex and the iteration efficiency is low.

[0006] In order to achieve the above object, the present invention adopts the following technical scheme: The present invention provides a method for analyzing motor imagery EEG signals based on BHS, comprising the following steps: S1. Through the motor imagery brain-computer interface experimental paradigm, the test target is subjected to offline experiments and the offline experimental EEG data is collected; S2. Based on the BHS EEG channel selection method, the optimal EEG channel selection for specific subjects is performed on the offline experimental EEG data to obtain the co-spatial filter and classification model; S3, conduct online verification experiments, collect the optimal channel EEG data in real time and perform preprocessing, extract EEG features based on spatial filters and input them into the classification model for pattern recognition; If the recognition is wrong, it returns to continue collecting offline experimental EEG data of the target to be tested; if the recognition is correct, it is converted into control instructions to drive the hand robot action.

[0007] In some embodiments, in S1, the motor imagery brain-computer interface experimental paradigm also includes, while the paradigm induces the motor imagery of the target to be tested, providing corresponding visual feedback according to the processing results of the motor imagery data of the target to be tested.

[0008] In some embodiments, in S1, offline experimental EEG data is collected using a wireless EEG cap, and the wireless EEG cap is placed according to the standard.

[0009] In some implementations, the specific process of S2 includes: S21, preprocessing the offline experimental EEG data; S22, select channels for pre-processed offline experimental EEG data based on BHS; S23, weighing the classification error rate and the number of channels, selecting the optimal solution set of the fitness function corresponding to the optimal weighted coefficient under the BHS iteration, selecting the optimal channel combination in the optimal solution set, and outputting the CSP spatial filter and SVM classification model.

[0010] Furthermore, in S21 , the offline experimental EEG data is preprocessed by using common mean reference, Min-max normalization and frequency band filtering.

[0011] Furthermore, S22 specifically includes: BHS runs on the harmony memory library and selects channels by optimizing the fitness function. The goal of channel selection is to reduce the number of channels. The fitness function is defined as the weighted sum of the classification error rate and the relative number of channels, expressed as follows: (1); in, Expressed as fitness function; is the classification error rate; is the relative number of channels; is the weight coefficient, weight w i Standardized .

[0012] In some implementations, S23 specifically includes: S231, initialization problem and algorithm parameters; S232, initializing harmony memory and calculating the fitness value of each harmony vector; S233, generating a new harmony vector according to the harmony innovation mechanism of the BHS algorithm and calculating the fitness value; S234, update harmony memory library; S235. When NI iterations are completed, the fitness values ​​under the HMS, HMCR and NI parameter values ​​are solved, and the three parameter values ​​corresponding to the lowest fitness values ​​are selected as the optimal parameters of HMS, HMCR and NI; otherwise, S233 and S234 are repeated in sequence.

[0013] Furthermore, in S231, the binary optimization problem can be described as follows: st (2); in, is the objective function; x For variables gather; n Expressed as the number of decision variables.

[0014] Furthermore, in S233, a new harmony mechanism is created as follows: (3); in, The first i -th component, HMS represents the harmonic memory size, and HMCR represents the harmonic memory consideration rate.

[0015] The present invention also provides a system based on the BHS-based motor imagery EEG signal analysis method, the system includes a paradigm presentation module, a data acquisition module, a signal processing module and a hand robot module, wherein: The paradigm presentation module is used to conduct offline experiments on subjects through the brain-computer interface experimental paradigm of motor imagery; The data acquisition module is used to acquire offline experimental EEG data; The signal processing module is used to select the optimal EEG channel for specific subjects based on the BHS EEG channel selection method for offline experimental EEG data, and obtain the co-spatial filter and classification model; The hand robot module is used to conduct online verification experiments, collect and preprocess the EEG data of the optimal channel in real time, extract EEG features based on the spatial filter and input them into the classification model for pattern recognition; if the recognition is wrong, it returns to continue collecting offline experimental EEG data of the target to be tested; if the recognition is correct, it is converted into control instructions to drive the hand robot's movements.

[0016] Compared with the prior art, the present invention provides a method and system for analyzing motor imagery EEG signals based on BHS, which has the following beneficial effects: The present invention is based on the BHS motor imagery EEG signal analysis method, which can provide personalized, precise intervention and control for the hand robot based on reasonable and efficient signal processing and analysis. The present invention is applicable to the robot control system based on brain-computer interface technology, improves the accuracy of the system, reduces the number of channels, shortens the preparation time of the experiment, and expands the application scenarios with special requirements for the comfort of the subjects. The present invention reduces the number of EEG signal channels and reduces the hardware cost and computing cost of the brain-computer interface system by adopting the BHS (Binary Harmony Search) channel selection method. At the same time, due to the optimization of channel selection, the experimental preparation time is shortened, thereby improving work efficiency, reducing the operating cost of medical equipment, and reducing production costs and improving the practicality and comfort of the system. It is expected to improve the overall applicability of the hand robot system and enhance the user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings in the specification are used to provide further understanding of the present invention and constitute a part of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0018] Figure 1 The schematic diagram of the architecture of a motor imagery EEG signal analysis system based on BHS of the present invention; Figure 2 A schematic diagram of an experimental paradigm in a method for analyzing motor imagery EEG signals based on BHS of the present invention; Figure 3 A motor imagery EEG channel selection method based on a BHS algorithm and an online verification schematic diagram in a motor imagery EEG signal analysis method based on a BHS of the present invention; Figure 4 A schematic flow chart of a method for selecting a motor imagery EEG channel based on a BHS algorithm in a method for analyzing motor imagery EEG signals based on BHS of the present invention; Figure 5 This is a schematic diagram of the control strategy of a brain-computer interface driving a hand robot in a method for analyzing motor imagery EEG signals based on BHS of the present invention. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0020] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0021] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0022] In the description of the embodiments of the present invention, it should be noted that if the terms "upper", "lower", "horizontal", "inner", etc. indicate an orientation or positional relationship based on the orientation or positional relationship shown in the drawings, or the orientation or positional relationship in which the product of the invention is usually placed when in use, it is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. In addition, the terms "first", "second", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.

[0023] In addition, if the term "horizontal" appears, it does not mean that the component must be absolutely horizontal, but can be slightly tilted. For example, "horizontal" only means that its direction is more horizontal than "vertical", which does not mean that the structure must be completely horizontal, but can be slightly tilted.

[0024] In the description of the embodiments of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", and "connect" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal connection of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0025] How to improve the data processing accuracy and efficiency of EEG signals in order to improve the effectiveness and practicality of hand robot control and thus enhance user experience.

[0026] Based on this, Figure 1 As shown, the present invention provides a method for analyzing motor imagery EEG signals based on BHS, comprising the following steps: S1. Through the motor imagery brain-computer interface experimental paradigm, the test target is subjected to offline experiments and the offline experimental EEG data is collected; S2. Based on the BHS EEG channel selection method, the optimal EEG channel selection for specific subjects is performed on the offline experimental EEG data to obtain the co-spatial filter and classification model; S3, conduct online verification experiments, collect the optimal channel EEG data in real time and perform preprocessing, extract EEG features based on spatial filters and input them into the classification model for pattern recognition; If the recognition is wrong, it returns to continue collecting offline experimental EEG data of the target to be tested; if the recognition is correct, it is converted into control instructions to drive the hand robot action.

[0027] The present invention is a method for analyzing motor imagery EEG signals based on BHS, which can significantly improve the performance of the hand robot system, especially maintain a high recognition rate at a low number of channels, and enhance the market competitiveness of the hand robot. In view of the trend of increasing requirements for cost, efficiency and comfort in the industry, the present invention will increase the market share of related products and promote the widespread application of more hospitals, centers and home users.

[0028] Through efficient brain-computer interface control, the present invention enables users to achieve more accurate sports training, improve efficiency, and shorten the cycle, which not only helps to improve the quality of life of users, but also reduces the continuous demand for medical services. The present invention optimizes the channel selection method of the brain-computer interface system, reduces the computational complexity and cost, lowers the application threshold of brain-computer interface technology, and improves the applicability of brain-computer interface in the fields of nerves and sports training.

[0029] In some embodiments, the present invention provides a method-based system for analyzing motor imagery EEG signals based on BHS, and the specific operations are as follows: Step 1: The subject wears a 32-lead wireless EEG cap to collect EEG signals. The computer screen in front of the subject is 60-80 cm away from the head. All electrodes are placed according to standard leads. A reference electrode is selected to ensure that all electrodes are in good contact with the scalp.

[0030] Step 2: Design a motor imagery brain-computer interface experimental paradigm. The subjects conduct offline experiments according to the requirements of the experimental paradigm and complete the number of experiments required by all experimental tasks.

[0031] Step 3: Collect offline experimental EEG data, select the optimal EEG channel for specific subjects according to the BHS algorithm, and obtain the co-spatial filter and classification model.

[0032] Step 4: Conduct an online verification experiment to collect and preprocess the EEG data of the optimal channel in real time, extract EEG features based on the spatial filter and input them into the classification model for pattern recognition. If the recognition is correct, it is converted into control instructions, and the control instructions are transmitted via Bluetooth to drive the hand robot.

[0033] As an option, the hand robot may be, but is not limited to, a "guardian" hand robot.

[0034] Step 3: Preprocessing of EEG signals, channel selection based on BHS, feature extraction and classification model construction. Specific steps: Step 3-1: The preprocessing methods mainly include Common Average Reference (CAR), Min-max normalization and frequency band filtering. Min-max normalization is performed independently on each channel in each trial. The main purpose is to convert each value of the EEG signal into a value between 0 and 1 to reduce the influence of the amplitude difference between channels.

[0035] Step 3-2: Channel selection based on the BHS algorithm. Offline training data is mainly used to select the optimal channel subset. First, design the fitness function and set the fitness parameters. w 1 Parameter range w 1=0.1~0.9, the process is 0.1. Traverse each w The fitness function under the value of 1 is used to iterate the BHS algorithm.

[0036] Further, in step 3-2-1, the problem and algorithm parameters are initialized, and the range of BHS algorithm parameters (harmonic memory size HMS, harmonic memory consideration rate HMCR and number of iterations NI) is set.

[0037] Step 3-2-2, initialize the harmony memory HM and calculate the fitness value of each harmony vector.

[0038] Step 3-2-3, generate a new harmony vector according to the harmony innovation mechanism of the BHS algorithm and calculate the fitness value.

[0039] Step 3-2-4, update the harmony memory library.

[0040] Step 3-2-5, continue iterating until the stopping condition is met and output the optimal fitness value.

[0041] The fitness values ​​under all parameters are calculated, and the parameters HMS, HMCR and NI corresponding to the optimal fitness values ​​are determined as the best parameters.

[0042] Step 3-3, weigh the classification error rate and the number of channels and select the optimal weighting coefficient w The fitness value and harmony vector corresponding to 1 are the candidate optimal channel subsets, and the co-spatial pattern CSP spatial filter and SVM classifier corresponding to the optimal channel subset are output at the same time.

[0043] The following is a detailed description of a method and system for analyzing motor imagery EEG signals based on BHS of the present invention through specific embodiments: like Figure 1 As shown, the BHS-based motor imagery EEG signal analysis system of the present invention includes a paradigm presentation module, a data acquisition module, a signal processing module and a hand robot module, wherein: The paradigm presentation module is used to conduct offline experiments on subjects through the brain-computer interface experimental paradigm of motor imagery; The data acquisition module is used to acquire offline experimental EEG data; The signal processing module is used to select the optimal EEG channel for specific subjects based on the BHS EEG channel selection method for offline experimental EEG data, and obtain the co-spatial filter and classification model; The hand robot module is used to conduct online verification experiments, collect and preprocess the EEG data of the optimal channel in real time, extract EEG features based on the spatial filter and input them into the classification model for pattern recognition; if the recognition is wrong, it returns to continue collecting offline experimental EEG data of the target to be tested; if the recognition is correct, it is converted into control instructions to drive the hand robot's movements.

[0044] In a specific embodiment; Paradigm presentation module: Based on the mirror neuron theory, the present invention designs an experimental paradigm. In the motor imagery stage, the subject is visually induced by displaying a video of grasping a cup. The subject follows the screen prompts and imagines grasping the cup with the right hand. While the paradigm induces the subject's motor imagery, the paradigm also needs to provide corresponding visual feedback based on the processing results of the subject's motor imagery data.

[0045] Signal acquisition module: The main functions implemented are signal acquisition parameter setting, real-time signal acquisition and real-time signal presentation, and data storage. The g. USBamp EEG signal acquisition amplifier is used in the present invention to mainly realize the acquisition, amplification and hardware filtering of EEG. The signal acquisition amplifier is connected to the computer via USB. The collected scalp EEG signal is stored in the computer. The EEG acquisition module is a 32-lead dry electrode EEG cap, and it can also be 64, 128 leads or any number of leads (at least greater than 16). It should be noted here that the more leads there are, the more necessary the channel selection is, and the BHS algorithm channel selection is more effective.

[0046] Signal processing module: It is the core of the BCI system. From the perspective of processing steps, it can be divided into EEG signal preprocessing, feature extraction and pattern recognition. According to the real-time requirements, it can be divided into offline processing and online processing. In the EEG lead selection method based on the binary harmony search algorithm, offline processing needs to obtain the optimal channel combination for a specific subject. In addition, all three methods require 30 minutes of offline training time. The online experiment requires preprocessing of the signal, and then feature extraction based on the spatial filter obtained by the corresponding offline training method, and then the feature samples are imported into the classification model for intent recognition.

[0047] Hand robot module: Write and package the corresponding command action program of the hand robot. Convert the recognition results obtained by the data processing module into corresponding action control instructions, transmit the control instructions to the hand robot through wireless communication, and control it to drive the user's hand to complete the corresponding action according to the instructions.

[0048] In a specific embodiment, Figure 2 As shown: 1. Design a brain interface experimental paradigm and recruit healthy right-handed subjects to participate in the experiment. Before the experiment, the subjects wore an EEG cap, remained relaxed, and sat in front of the display screen; they remained relaxed for the first 5 seconds. When the subjects heard the sound prompt "Beep", they tried to reduce blinking, body shaking and other actions that affect the quality of signal acquisition to reduce artifacts in the EEG acquisition process. After 10 seconds, the "+" on the screen disappeared and the hand grasping cup action video started to run. The motor imagery task was performed in 10-14 seconds. The corresponding actions needed to be performed according to the offline and online experimental tasks in the 14-18 second period. In the offline experiment, the subjects were in a relaxed state; in the online experiment, the system recognized the subject's motor imagery intention and converted it into control instructions to drive the robot to drive hand training. Each subject performed 5 rounds, each round contained 10 trials (Trial), that is, each subject performed 50 trials. There was no limit on the rest time between each round to ensure that the subjects were not tired when the next round of trials was carried out to prevent fatigue from affecting the quality of EEG signals.

[0049] 2. Based on BHS EEG channel selection method, such as Figure 3 and Figure 4 As shown, the offline training phase mainly performs channel selection, and the specific steps are as follows: 1) The preprocessing methods mainly include Common Average Reference (CAR), Min-max normalization and band filtering. Min-max normalization is performed independently on each channel in each trial. The main purpose is to convert each value of the EEG signal into a value between 0 and 1 to reduce the influence of the amplitude difference between channels.

[0050] 2) Channel selection based on BHS algorithm.

[0051] BHS runs on the harmony memory library and selects channels by optimizing the fitness function. The goal of channel selection is to reduce the number of channels while minimizing or improving the classification accuracy of the system. The fitness function is defined as the weighted sum of the classification error rate and the relative number of channels, expressed as follows: (1); in, Expressed as fitness function; is the classification error rate; is the relative number of channels; is the weight coefficient, weight w i Standardized .

[0052] In an embodiment of the present invention, is the error rate of 10-fold cross validation of the training data, where the classification error rate is obtained by extracting the EEG training features of the EEG data of the selected channel through the common spatial pattern CSP algorithm, and then calculating the classification error rate as the 10-fold cross validation error rate. The value is usually the harmony vector x The number of channels selected in is divided by the total number of channels in the dataset (i.e. the length of the binary string). and ranges from 0 to 1, so The change is also 0 to 1.

[0053] Set the weight coefficient in the fitness function to a range of w1 = 0.1 to 0.9, where the process is 0.1; finally, traverse each w The fitness function under the value of 1 is used to iterate the BHS algorithm.

[0054] In the channel selection based on the BHS algorithm, specifically: 21) Initialization problem and algorithm parameters; The binary optimization problem can be described as follows: st (2); in, is the objective function; x For variables gather; n It is expressed as the number of decision variables. The BHS algorithm includes the harmonic memory size HMS and the harmonic memory consideration rate HMCR.

[0055] 22) Initialize harmony memory; This HM can be expressed as , where each variable is randomly generated in the feasible solution space. Binary harmony memory HM adopts binary encoding, where the length of each harmony vector is equal to the number of channels of each data set. Each decision variable 𝑥 𝑖 Represents a specific channel. If the value of the element is "1", the corresponding channel is selected for subsequent signal processing. If the value of the element is "0", it indicates that the channel at this position is not selected. The initial harmonic memory consists of HMS harmonic vectors. After initializing the harmonic memory library, the fitness value of each harmonic vector is calculated. The specific method is to extract the channel EEG data corresponding to the element 1 in each vector, extract the EEG features through the common spatial pattern CSP, and build a classification model according to the SVM algorithm to calculate the 10-fold cross-validation error rate. , is the number of selected channels, and the fitness value of each harmony vector is obtained through the above calculation principle.

[0056] 23) According to the HMCR harmony creation mechanism, a new harmony vector is improvised, which is expressed as

[0057] The mechanism for creating new harmony is as follows: (3); Where: The first i -th component. Next, solve The fitness value of .

[0058] 24) When the fitness value of the newly generated harmony is better than the worst harmony in HM, the worst harmony is replaced by the newly generated one. replace.

[0059] 25) The iterative search terminates, and when the stopping criterion is met, the optimal solution is output. Otherwise, 23) and 24) are repeated. Here, the stopping criterion is NI iterations completed.

[0060] In the whole BHS algorithm, the parameters of the BHS algorithm are the harmonic memory size HMS, the harmonic memory consideration rate HMCR and the number of iterations NI. The values ​​of HMS are 1, 10, 20, 30, 40 and 50. The values ​​of HMCR are 0.7, 0.75, 0.8, 0.85, 0.9, 0.95 and 0.99. The range of NI is 0 to 500.

[0061] Solve the fitness values ​​under all HMS, HMCR and NI parameter values, and finally select the three parameter values ​​corresponding to the lowest fitness value as the optimal parameters of HMS, HMCR and NI.

[0062] 3) Weighing the classification error rate and the number of channels, the optimal solution set corresponding to the fitness function of the optimal weighted coefficient is selected under the BHS algorithm, and the channel corresponding to "1" in the optimal solution set is selected as the optimal channel combination for the specific subject. In this process, the CSP (common spatial pattern) spatial filter and SVM (support vector machine) classification model are output.

[0063] 4) According to the optimal channel combination obtained offline, in the online testing phase, it is only necessary to collect the EEG data corresponding to the optimal channel combination and pre-process the EEG signals. It should be noted here that it is only necessary to process the EEG signals of the motor imagery time stage, and then perform feature extraction through the spatial filter obtained offline, and finally input them into the SVM classifier for classification and recognition.

[0064] 5) If the recognition is wrong, the hand robot control will not be executed. If the recognition is correct, the recognition result will be converted into a corresponding action control instruction, and the control instruction will be transmitted to the hand robot through wireless communication. According to the instruction, it will be controlled to drive the user's hand to complete the corresponding action.

[0065] like Figure 5 As shown in the figure, the control scheme of the hand robot is shown. The motor imagery EEG signal analysis system based on BHS of the present invention adopts the classic Bluetooth module to form a wireless communication system. The hand robot integrates a controller, a motor and a Bluetooth module. The system software transmits the training mode information to the controller via Bluetooth, and controls the motor to drive the manipulator to realize hand grasping training.

[0066] In summary, the present invention provides a method and system for analyzing motor imagery EEG signals based on BHS, which significantly improves the accuracy and efficiency of hand robot control. The present invention maintains a high recognition rate at a low number of channels, optimizes the channel selection of the brain-computer interface system, reduces computational complexity and cost, and enhances the market competitiveness of the hand robot. In addition, the present invention can enable users to achieve more accurate sports training, improve training efficiency, shorten the cycle, help improve the quality of life of users, and reduce the demand for medical services. The present invention lowers the application threshold of brain-computer interface technology and improves its applicability.

[0067] Finally, it should be noted that the above is only a preferred embodiment of the present invention and does not limit the present invention in any form. Any ordinary technician in the industry can smoothly implement the present invention as shown in the specification and described above, and any equivalent changes, modifications and evolutions made by using the technical content disclosed above are all equivalent embodiments of the present invention. At the same time, any equivalent changes, modifications and evolutions made to the above embodiments based on the essential technology of the present invention are still within the protection scope of the technical solution of the present invention.

Claims

1. A method for analyzing motor imagery EEG signals based on BHS, characterized in that: The steps include: S1. Through the motor imagery brain-computer interface experimental paradigm, the test target is subjected to offline experiments and the offline experimental EEG data is collected; S2. Based on the BHS EEG channel selection method, the optimal EEG channel selection for specific subjects is performed on the offline experimental EEG data to obtain the co-spatial filter and classification model; S3, conduct online verification experiments, collect the optimal channel EEG data in real time and perform preprocessing, extract EEG features based on spatial filters and input them into the classification model for pattern recognition; If the recognition is wrong, it returns to continue collecting offline experimental EEG data of the target to be tested; if the recognition is correct, it is converted into control instructions to drive the hand robot action.

2. The method for analyzing motor imagery EEG signals based on BHS according to claim 1, characterized in that: In said S1, the motor imagery brain-computer interface experimental paradigm also includes, while the paradigm induces the motor imagery of the subject to be tested, providing corresponding visual feedback according to the processing results of the motor imagery data of the subject to be tested.

3. The method for analyzing motor imagery EEG signals based on BHS according to claim 1, characterized in that: In S1, the offline experimental EEG data is collected using a wireless EEG cap, which is placed according to the standard.

4. The method for analyzing motor imagery EEG signals based on BHS according to claim 1, characterized in that: The specific process of S2 includes: S21, preprocessing the offline experimental EEG data; S22, select channels for pre-processed offline experimental EEG data based on BHS; S23, weighing the classification error rate and the number of channels, selecting the optimal solution set of the fitness function corresponding to the optimal weighted coefficient under the BHS iteration, selecting the optimal channel combination in the optimal solution set, and outputting the CSP spatial filter and SVM classification model.

5. The method for analyzing motor imagery EEG signals based on BHS according to claim 4, characterized in that: In S21, the offline experimental EEG data is preprocessed by using common average reference, Min-max normalization and frequency band filtering.

6. The method for analyzing motor imagery EEG signals based on BHS according to claim 4, characterized in that: The S22 specifically includes: BHS runs on the harmony memory library and selects channels by optimizing the fitness function. The goal of channel selection is to reduce the number of channels. The fitness function is defined as the weighted sum of the classification error rate and the relative number of channels, expressed as follows: （1）； in, Expressed as fitness function; is the classification error rate; is the relative number of channels; is the weight coefficient, weight w i Standardized .

7. The method for analyzing motor imagery EEG signals based on BHS according to claim 1, characterized in that: The S23 specifically includes: S231, initialization problem and algorithm parameters; S232, initializing harmony memory and calculating the fitness value of each harmony vector; S233, generating a new harmony vector according to the harmony innovation mechanism of the BHS algorithm and calculating the fitness value; S234, update harmony memory library; S235. When NI iterations are completed, the fitness values ​​under the HMS, HMCR and NI parameter values ​​are solved, and the three parameter values ​​corresponding to the lowest fitness values ​​are selected as the optimal parameters of HMS, HMCR and NI; otherwise, S233 and S234 are repeated in sequence.

8. The method for analyzing motor imagery EEG signals based on BHS according to claim 7, characterized in that: In S231, the binary optimization problem can be described as follows: s.t. (2); in, is the objective function; x For variables gather; n Expressed as the number of decision variables.

9. The method for analyzing motor imagery EEG signals based on BHS according to claim 7, characterized in that: In the S233, a new harmony mechanism is created as follows: （3）； in, The first i -th component, HMS represents the harmonic memory size, and HMCR represents the harmonic memory consideration rate.

10. A system based on the BHS-based motor imagery EEG signal analysis method according to any one of claims 1 to 9, characterized in that: The system includes a paradigm presentation module, a data acquisition module, a signal processing module and a hand robot module, wherein: The paradigm presentation module is used to conduct offline experiments on subjects through the brain-computer interface experimental paradigm of motor imagery; The data acquisition module is used to acquire offline experimental EEG data; The signal processing module is used to select the optimal EEG channel for specific subjects based on the BHS EEG channel selection method for offline experimental EEG data, and obtain the co-spatial filter and classification model; The hand robot module is used to conduct online verification experiments, collect and preprocess the EEG data of the optimal channel in real time, extract EEG features based on the spatial filter and input them into the classification model for pattern recognition; if the recognition is wrong, it returns to continue collecting offline experimental EEG data of the target to be tested; if the recognition is correct, it is converted into control instructions to drive the hand robot's movements.

Images