Science and technology competition system and method applying electroencephalogram concentration reverse training key technology

Abstract

The invention belongs to the technical field of electroencephalogram training, and particularly relates to a science and technology competition system and method applying an electroencephalogram concentration reverse training key technology, and the system comprises an electroencephalogram collection end which is used for sampling electroencephalogram signals from a specific area to carry out noise reduction, filtering and amplification, and transmitting the electroencephalogram signals to a Raycore microprocessor (RK2206) controller through a serial port; the RK2206 controller is used for loading a lightweight swan operating system and sending the processed electroencephalogram signals to a remote terminal and a computer terminal through Bluetooth by using threads; the Bluetooth transceiving module is used for amplifying the electroencephalogram signal and then sending the electroencephalogram signal to a remote terminal through Bluetooth; the computer end is provided with a user front end and a background service system, and the user front end can browse the tested user grade report or perform concentration training on a specific user; the background management system is responsible for browsing background user information and electroencephalogram data; and the remote terminal is used for installing a remote control tool to realize remote control. The method is an ICA method, artifact interference of electro-oculogram and myoelectricity is removed, meanwhile, an innovative emotion judgment mode is fused, crowds are specifically classified, sampling points are reduced, and influences of other currents are indirectly reduced.

Description

Technical Field

[0001] The present invention belongs to the field of electroencephalogram (EEG) training technology, and in particular, relates to a technology competition system and method using the key technology of EEG concentration reverse training. Background Art

[0002] The electrical signals of the brain can directly reflect the actual condition of the brain, so the analysis of this field helps humans to correctly explore and analyze the mysteries of the brain. With the deepening of EEG signal analysis, new solutions can be developed for the clinical treatment of various brain diseases, such as depression, epilepsy and other problems. It can also provide important methodological support for ADHD and concentration training.

[0003] This patented technology is designed for the service system of electronic design technology innovation competitions for teenagers to college students, with the purpose of improving the concentration of contestants during the design process. During the competition, there are usually three to four contestants, and each contestant has a different role in the production process. Therefore, a customized concentration training system is required for objects of different roles and levels. At the same time, it also provides a reference for the division of team roles for counselors to improve team collaboration efficiency.

[0004] The overall steps of EEG signal processing are divided into EEG signal extraction, signal analysis and application in specific fields. In EEG signal extraction, it is necessary to go through the steps of electrode current acquisition, artifact removal, noise reduction, signal amplification, etc. Then there is the signal analysis link, which has different analysis schemes due to different application fields. With the maturity of machine learning algorithms and the improvement of computing equipment capabilities, some studies have used artificial neural network learning methods such as decision trees, random forests, K nearest neighbor algorithms, support vector machines, naive Bayes, etc. to classify EEG features in order to analyze the maturity of the brain or provide EEG technical support for specific diseases. This direction provides professional theoretical basis support for this patent in concentration training.

[0005] According to the current status of classification technology, most classification methods are concentrated on establishing accurate EEG signal acquisition, establishing EEG models, and analyzing time-frequency characteristics. EEG acquisition is the first step in establishing an EEG signal model. Since EEG signals are microcurrents, they are often affected by ECG and EOG signals. Therefore, in order to improve the accuracy of EEG acquisition, some studies have focused on the removal of EEG artifacts. Some studies have used regression methods to remove the interference of ECG and EOG on EEG signals, but they are non-online training methods and are only suitable for processing single-channel methods. They need to increase the cost of background hardware and are not real-time. There are also studies that use adaptive filtering and Wiener filtering to remove interference from EEG signals, which can be used to remove interference EEG online, but the amount of data processing is large. There are also deep learning methods to process EEG signals, and reduce the interference of EEG, ECG, and EMG on EEG signals. The above methods mostly use multi-point and multi-channel EEG, and the actual equipment is large in size and processes a large amount of data. The huge amount of data requires the introduction of learning algorithms in the modeling process.

[0006] Existing technologies aim to improve the classification accuracy and computing efficiency of EEG models by improving the accuracy of EEG acquisition and the algorithm of EEG models. However, most studies are not targeted at specific scenarios and use multi-channel acquisition, which leads to problems such as large hardware equipment, long computing time, and high hardware overhead. For the field of science and technology competition for the general population, it is necessary to make reference adjustments based on applicable purposes. Therefore, in order to adapt to EEG classification in more scenarios, many studies have begun to incorporate richer feature signals.

[0007] In order to reduce the cost of EEG sampling software and hardware caused by multiple channels, and at the same time improve the efficiency of competition production. The project divides the roles of ordinary people in the competition into creators, imagineers and logic trainers, that is, to meet the training needs of contestants of different role levels. After judging based on the EEG emotion model, the EEG sampling of the subject can be reduced to specific points, thereby reducing hardware expenses. The key to the overall project lies in the signal judgment between the emotion model and the EEG. The hot spots of emotion model judgment at home and abroad can be carried out through expressions, EEG and other methods. The above methods focus on the two directions of emotion perception judgment and the association of EEG sampling areas, but the characteristic model of emotion and population level has not yet been established.

[0008] Based on the above technologies, the existing systems that use EEG to train concentration have the following shortcomings:

[0009] 1. The single analysis of EEG signals has many sampling points and the EEG model classification accuracy is not high;

[0010] 2. EEG monitoring involves multiple sampling points, the equipment is too expensive, and it is not conducive to mobility.

[0011] 3. Currently, there is no concentration training for the classification of roles in the competition, and no customized training model has been established. Summary of the invention

[0012] In view of the problems existing in the prior art, the present invention provides a science and technology competition system and method using the key technology of reverse training of EEG concentration.

[0013] The present invention is implemented as follows: a science and technology competition system using the key technology of reverse training of EEG concentration, the system comprising:

[0014] The EEG acquisition terminal is used to sample EEG signals from a specific area for noise reduction, filtering and amplification, and transmit them to the Rockchip microprocessor (RK2206) controller via the serial port;

[0015] The RK2206 controller is connected to the EEG acquisition terminal, loads the lightweight Hongmeng operating system, and uses threads to send the processed EEG signals to the remote terminal and computer via Bluetooth;

[0016] The Bluetooth transceiver module is connected to the RK2206 controller and is used to amplify the EEG signal and send it to the remote terminal via Bluetooth;

[0017] The computer side is connected to the RK2206 controller and has a user front-end and a back-end service system. The user front-end can browse the user level report after the test or conduct concentration training for specific users; the back-end management system is responsible for browsing the back-end user information and EEG data;

[0018] The remote terminal is connected to the Bluetooth transceiver module and is used to install the remote control tool to realize remote control.

[0019] Furthermore, the EEG test and training algorithm of the EEG acquisition terminal specifically includes: an EEG data analysis part, an emotion pattern discrimination part, and a specific population direction training part.

[0020] Furthermore, the EEG data analysis part includes: anti-interference processing, noise reduction processing, and eigenvalue extraction based on power spectrum density; the anti-interference processing is to use a notch filter to process the interference value of a specific frequency, wherein the anti-interference function is:

[0021]

[0022] Where f0 is used to indicate the frequency range of the signal to be filtered out, and a and b are used to indicate the filter coefficients of the notch filter. The noise reduction process divides the EEG signal into windows through the fast Fourier transform (STFT). The signal in each window is stable, and then Fourier transform is performed on each window. The transformation formula is:

[0023]

[0024] οx(τ) is the original signal.

[0025] ο(t-τ) is the window function.

[0026] οe -j2πfτ is the kernel function of the Fourier transform.

[0027] Then, for further analysis, it is necessary to extract the EEG eigenvalues ​​and perform power spectrum density conversion. The core conversion algorithm obtains the power spectrum intensity (PSI) of the kth part through the following formula:

[0028]

[0029] Where f s is the sampling rate, and N is the sequence length.

[0030] Furthermore, the emotion pattern identification part specifically includes:

[0031] The system will establish a three-dimensional emotional pattern classification for the test population, which is divided into creativity, thinking, and imagination. The classification process first uses the EEG emotional pattern dataset for online training; the training method uses the SEED dataset, which is a collection of EEG datasets provided by the BCMI laboratory. The SEED dataset contains the EEG signals of the subjects when watching movie clips. The movie clips are carefully selected to evoke different types of emotions, including positive, negative, and neutral emotions;

[0032] A neural network algorithm is used to classify groups with positive emotional experiences under specific audio-visual device stimulation; the SEED visual library is used to train the classification model; the overall process includes the following: classify the audio-visual stimulation sources into three types: creativity-oriented, logical thinking-oriented, and imagination-oriented; then test the assessment categories of the trainees' positive emotional experiences under each type; finally, use the assessment model to judge the testees' abilities;

[0033] The algorithm model used is an improved spatiotemporal graph neural network; an effective method for learning long-distance dependencies in sequence data is used to establish a self-attention network in various time series analysis tasks.

[0034] Furthermore, the improved spatiotemporal graph neural network calculation process includes the following steps:

[0035] (1) For a given input sequence Where n is the sequence length and d is the feature dimension. The corresponding query matrix is ​​obtained by linear transformation Key Matrix Sum Matrix

[0036] Where Q = XW Q ,K=XW k ,V=XW v are three trainable parameter matrices, d q , d k , d v are the dimensions of the query vector, key vector, and value vector respectively:

[0037] (2) Then, we calculate the dot product of the transposed query matrix and the key matrix K to obtain the attention score matrix S∈R n*n :

[0038] S=QK T

[0039] (3) In order to avoid the gradient vanishing problem caused by the excessive value obtained by the dot product calculation, the obtained attention score matrix needs to be scaled, generally by d k The square root of the key vector dimension is used to implement it, then the scaled attention score matrix S'∈R n*n It can be expressed as:

[0040]

[0041] (4) Perform a softmax operation on the attention score matrix to obtain a normalized attention weight matrix;

[0042] (5) Finally, multiply the attention weight matrix by the value matrix V to obtain the final output matrix.

[0043] Furthermore, the specific population directional training part is to divide the population into three categories: lateral creativity, logic, and imagination, and further reverse training; for the above three categories of people, different stimulus source libraries are established, and for the creativity group, a video library related to building construction, transportation track design, urban greening design, etc. is established; for the logical thinking group, a chess and go game library is established; for the imagination group, abstract thinking conversion graphic training is established.

[0044] Another object of the present invention is to provide a technology competition method using the key technology of EEG concentration reverse training based on the technology competition system using the key technology of EEG concentration reverse training, the method specifically comprising:

[0045] S1: The EEG acquisition module mainly samples EEG signals from a specific area for noise reduction, filtering and amplification, and transmits them to the Rockchip microprocessor (RK2206) controller via the serial port;

[0046] S2: The RK2206 controller loads the lightweight Hongmeng operating system, uses threads to amplify the processed EEG signals and sends them to the remote terminal and computer via Bluetooth;

[0047] S3: The computer side has a user front-end and a back-end service system. The user front-end can browse the user level report after the test or conduct concentration training for specific users; the back-end management system is responsible for browsing the back-end user information and EEG data.

[0048] Another object of the present invention is to provide a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the science and technology competition method using the key technology of reverse training of EEG concentration.

[0049] Another object of the present invention is to provide a computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to execute the steps of the scientific and technological competition method using the key technology of reverse training of EEG concentration.

[0050] Another object of the present invention is to provide an information data processing terminal, which is used to implement a science and technology competition system that uses the key technology of EEG concentration reverse training as described above.

[0051] In combination with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solutions to be protected by the present invention are as follows:

[0052] First, the present invention first integrates emotional features to establish a model between emotions and brain waves. Based on the correlation between emotions and brain wave areas, this method can concentrate EEG monitoring in specific areas in subsequent concentration training, thereby reducing sampling point data and reducing costs. In order to achieve this goal, a model library between EEG signals of people with different abilities and senses is established through samples, and EEG and pleasant emotion models are mainly established based on the type of stimulation, so as to judge the type and level of the crowd.

[0053] Secondly, due to the reduction of sampling points, the size of the device for monitoring EEG can be reduced, thereby reducing hardware overhead costs. At the same time, the neural network algorithm is improved, making it possible to speed up the calculation efficiency of the classifier through graph acceleration and other methods.

[0054] Furthermore, the crowd classification method will be based on the role of the team during the competition, roughly classified into three types: creators, deducers, and innovators, focusing on the creativity, logical thinking ability, and imagination of the crowd. Through the neural network algorithm, after the model is judged, training concentration-related plans are formulated for different ranges of people. Therefore, it has more customized characteristics and can more effectively customize training plans.

[0055] Second, as auxiliary evidence of the inventiveness of the claims of the present invention, it is also reflected in the following important aspects:

[0056] (1) The expected benefits and commercial value of the technical solution of the present invention after transformation:

[0057] This solution will be designed for a youth competition reservation system, which can be applied to science and technology competition coaching institutions and concentration training institutions. After using this system, the coaching efficiency of the competition process can be improved. After adopting this system, the expected income is 300 yuan per person per time, and the number of people served in a single month can reach nearly 100. The total number of coaching times per person per month is nearly 4-6 times, which can generate economic benefits of 1.2 million yuan per year.

[0058] (2) The technical solution of the present invention fills the technical gap in the industry at home and abroad:

[0059] 1. Targeting a wider range of people

[0060] Attention training at home and abroad is mostly for children with ADHD and learning difficulties, using sensory integration training, Schulte grid attention training method and clinical drugs. This invention uses multi-attention graphs to accelerate neural networks for ordinary people in the competition process, establishes a perceptual model of EEG and emotions, and uses this model to classify creativity, logical thinking and imagination for ordinary people, thereby customizing the reverse of attention, filling the technical gap in the industry at home and abroad.

[0061] 2. Use multi-attention graph accelerated neural network to establish a new EEG-emotion-crowd type

[0062] Xing Model

[0063] The graph-accelerated neural network algorithm using multiple attentions can perform parallel operations, greatly improving the efficiency of training and thus improving the real-time nature of training feedback.

[0064] (3) The technical solution of the present invention solves the technical problems that people have been eager to solve but have never been able to solve successfully:

[0065] The technical solution of the present invention establishes a new model of EEG-emotional perception-concentration training, and utilizes new methods to expand the program suitable for people with specific clinical diseases to the general population in a specific e-sports competition link. It not only effectively improves the efficiency of competition cooperation, but also provides better guidance programs for counselors of the participating groups.

[0066] Concentration training at home and abroad is mostly aimed at children with ADHD, but the technical solution of the present invention expands the applicable population to the general population. By using the relationship between EEG and emotion perception, a role classification of the competition service system is established to fully enhance the concentration of the general population.

[0067] Third, the technical problems solved by the present invention are:

[0068] 1. Insufficient accuracy of EEG signal processing

[0069] Existing EEG signal acquisition and processing systems are often affected by environmental interference, equipment noise and other factors, resulting in low signal quality and inaccurate feature extraction. The present invention uses a noise reduction algorithm combining a notch filter and short-time Fourier transform (STFT) to effectively suppress interference and improve signal quality, providing a reliable data basis for subsequent analysis.

[0070] 2. Concentration training lacks personalization and scientificity

[0071] Existing concentration training methods often lack accurate recognition of individual EEG characteristics and emotional patterns, and the training effect is limited. The present invention designs personalized concentration training programs for specific groups of people through feature extraction based on power spectral density and emotional pattern discrimination algorithm, which significantly improves the pertinence and effect of training.

[0072] 3. Low efficiency of data transmission and system management between devices

[0073] Traditional EEG devices are inefficient in data transmission and system management, and are inconvenient for users to operate. This invention uses the RK2206 controller and Bluetooth module, combined with the lightweight Hongmeng operating system, to achieve rapid transmission of EEG signals and synchronous management of multiple terminals, significantly improving the system's operating efficiency and user experience.

[0074] 4. Insufficient integration of EEG training system with application scenarios of science and technology competitions

[0075] Existing EEG training systems are rarely used in science and technology competitions, lacking real-time interactivity and contextual support. This invention integrates EEG signal acquisition, training, remote control and background management systems to achieve real-time evaluation and training optimization of EEG concentration in a competition environment, thus filling this gap.

[0076] Fourth, the significant technological advancement brought about by the present invention:

[0077] 1. Innovation and improvement of signal processing technology

[0078] The present invention innovatively combines notch filter anti-interference, short-time Fourier transform noise reduction and a feature extraction algorithm based on power spectral density, which not only significantly improves the analysis accuracy of EEG signals, but also expands the practical application scope of EEG signals.

[0079] 2. Breakthrough in personalized concentration training technology

[0080] By introducing an emotion pattern recognition algorithm and a specific directional training design, the present invention realizes personalized EEG concentration training to meet the needs of different users, and especially provides precise training plans for specific groups of people (such as teenagers or professional players).

[0081] 3. Implementation of multi-terminal collaborative work architecture

[0082] The present invention loads the lightweight Hongmeng operating system on the RK2206 controller, and realizes seamless connection between multiple terminal devices through the Bluetooth module, providing users with an efficient and convenient operation experience, and ensuring the real-time and consistency of EEG data.

[0083] 4. Optimization of application scenarios of the science and technology competition system

[0084] The user front-end and back-end management system designed by the present invention organically combines EEG signal analysis with user experience, providing powerful data support and intelligent management capabilities for the science and technology competition system. At the same time, it supports real-time evaluation and report generation of user concentration performance, promoting the widespread application of EEG technology in education, training and competition scenarios.

[0085] In summary, the present invention not only solves several key problems in the prior art, but also achieves significant technological progress in EEG signal processing, personalized concentration training, and application scenario optimization, and has broad industrial application prospects and market value. BRIEF DESCRIPTION OF THE DRAWINGS

[0086] Figure 1 This is a structural diagram of a science and technology competition system using a key technology of reverse training of EEG concentration provided by an embodiment of the present invention;

[0087] Figure 2 is a schematic diagram of attention in the Transformer provided by an embodiment of the present invention;

[0088] Figure 3 is a general architecture diagram based on the attention mechanism provided by an embodiment of the present invention;

[0089] Figure 4 It is a user terminal interface 1 provided by an embodiment of the present invention;

[0090] Figure 5 It is a user terminal interface 2 provided by an embodiment of the present invention;

[0091] Figure 6 It is the user terminal interface 3 provided by the embodiment of the present invention;

[0092] Figure 7 It is the system data end interface 1 provided by the embodiment of the present invention;

[0093] Figure 8 It is the system data end interface 2 provided by the embodiment of the present invention;

[0094] Fig. 9 is the data monitoring result provided by the embodiment of the present invention;

[0095] Fig.10 is a circuit diagram of an EEG acquisition terminal provided by an embodiment of the present invention;

[0096] Fig.11 is a data transmission circuit diagram provided by an embodiment of the present invention;

[0097] Fig.12 It is a data receiving circuit diagram provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0098] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0099] like Figure 1 As shown, an embodiment of the present invention provides a science and technology competition system using a key technology of reverse training of EEG concentration, the system comprising:

[0100] The EEG acquisition terminal is used to sample EEG signals from a specific area for noise reduction, filtering and amplification, and transmit them to the Rockchip microprocessor (RK2206) controller via the serial port;

[0101] The RK2206 controller is connected to the EEG acquisition terminal, loads the lightweight Hongmeng operating system, and uses threads to send the processed EEG signals to the remote terminal and computer via Bluetooth;

[0102] The Bluetooth transceiver module is connected to the RK2206 controller and is used to amplify the EEG signal and send it to the remote terminal via Bluetooth;

[0103] The computer is connected to the RK2206 controller and has a user front-end and a back-end service system. The user front-end can browse the user level report after the test or conduct concentration training for specific users; the back-end management system is responsible for browsing the back-end user information and EEG data;

[0104] The remote terminal is connected to the Bluetooth transceiver module and is used to install the remote control tool to realize remote control.

[0105] The EEG test and training algorithm of the EEG acquisition terminal specifically includes: an EEG data analysis part, an emotion pattern discrimination part, and a specific population direction training part.

[0106] The EEG data analysis part includes: anti-interference processing, noise reduction processing, and eigenvalue extraction based on power spectrum density; the anti-interference processing is to use a notch filter to process the interference value of a specific frequency, wherein the anti-interference function is:

[0107]

[0108] Where f0 is used to indicate the frequency range of the signal to be filtered out, and a and b are used to indicate the filter coefficients of the notch filter. The noise reduction process divides the EEG signal into windows through the fast Fourier transform (STFT). The signal in each window is stable, and then Fourier transform is performed on each window. The transformation formula is:

[0109]

[0110] οx(τ) is the original signal.

[0111] οw(t-τ) is the window function.

[0112] οe -j2πfτ is the kernel function of the Fourier transform.

[0113] Then, for further analysis, it is necessary to extract the EEG eigenvalues ​​and perform power spectrum density conversion. The core conversion algorithm obtains the power spectrum intensity (PSI) of the kth part through the following formula:

[0114]

[0115] Where f s is the sampling rate, and N is the sequence length.

[0116] 1. Signal acquisition and preprocessing at the EEG acquisition terminal

[0117] The EEG acquisition terminal is responsible for collecting EEG signals from specific areas of the user's head. The collected raw EEG signals contain a variety of interference and noise, and the signal quality is enhanced after noise reduction and filtering. First, a notch filter is used to suppress interference signals within a specific frequency range, and a stable filtering effect is ensured by adjusting the filter coefficients \(a\) and \(b\). Secondly, the short-time Fourier transform (STFT) is used to segment the EEG signal into multiple stable windows, and the spectrum information of each window is obtained by Fourier transform, preparing for subsequent feature extraction.

[0118] 2. Signal processing and transmission of RK2206 controller

[0119] The RK2206 controller runs the lightweight Hongmeng operating system, further processes the signals transmitted from the EEG acquisition terminal, and sends them to the remote terminal or computer through the Bluetooth module. The controller uses a multi-threaded mode to ensure that the processing and transmission of EEG signals are carried out in parallel. The loaded algorithms include feature extraction based on power spectral density, which accurately extracts the key features of EEG signals by calculating the power spectral intensity (PSI) of a specific frequency, providing support for emotional pattern discrimination and concentration training.

[0120] 3. Linkage between Bluetooth module and terminal equipment

[0121] The Bluetooth transceiver module serves as a bridge for data transmission, sending the processed EEG signals to the remote terminal or computer in real time. The control tool installed on the remote terminal enables dynamic monitoring of the EEG test process and adjustment of the concentration training program. Through this wireless communication mode, users can view EEG signals and training results on mobile devices at any time, while ensuring the efficiency and convenience of the system.

[0122] 4. Functions of user front-end and back-end management systems

[0123] The computer terminal integrates the user front-end and back-end management systems, providing an intuitive user interaction interface. The user front-end can browse the test results and concentration level reports in real time, and also supports customized EEG training tasks. The back-end management system is responsible for storing and analyzing user EEG data, and providing a comprehensive evaluation report on user behavior patterns and emotional states. In addition, the back-end management system can also optimize directional training for specific groups of people, and use big data analysis and deep learning technology to further improve the system training effect, ensuring the scientificity and effectiveness of the competition tasks.

[0124] The emotion pattern identification part specifically includes:

[0125] The system will establish a three-dimensional emotional pattern classification for the test population, which is divided into creativity, thinking, and imagination. The classification process first uses the EEG emotional pattern dataset for online training; the training method uses the SEED dataset, which is a collection of EEG datasets provided by the BCMI laboratory and the Shanghai Jiaotong University (SEED) dataset, which contains EEG signals when the subject watches movie clips. The movie clips are carefully selected to induce different types of emotions, including positive, negative, and neutral emotions;

[0126] A neural network algorithm is used to classify the groups with positive emotional experience under specific audio-visual device stimulation; the classification model is trained using the Shanghai Jiao Tong University (SEED) visual library; the overall process includes the following: classifying the audio-visual stimulation sources into three types: creativity-oriented, logical thinking-oriented, and imagination-oriented; then testing the assessment categories of the trainees' positive emotional experience under each type; finally, using the assessment model to judge the testees' abilities;

[0127] The algorithm model used is an improved spatiotemporal graph neural network; using an effective method for learning long-distance dependencies in sequence data, a self-attention network is established in a variety of time series analysis tasks, such as Figure 2 shown.

[0128] The improved spatiotemporal graph neural network calculation process has the following steps:

[0129] (1) For a given input sequence Where n is the sequence length and d is the feature dimension. The corresponding query matrix is ​​obtained by linear transformation Key Matrix Sum Matrix

[0130] Where Q = XW Q ,K=XW k ,V=XW v are three trainable parameter matrices, d q , d k , d v are the dimensions of the query vector, key vector, and value vector respectively:

[0131] (2) Then, we calculate the dot product of the transposed query matrix and the key matrix K to obtain the attention score matrix S∈R n*n :

[0132] S=QK T

[0133] (3) In order to avoid the gradient vanishing problem caused by the excessive value obtained by the dot product calculation, the obtained attention score matrix needs to be scaled, generally by d k The square root of the key vector dimension is used to implement it, then the scaled attention score matrix S'∈R n*n It can be expressed as:

[0134]

[0135] (4) Perform a softmax operation on the attention score matrix to obtain a normalized attention weight matrix.

[0136] (5) Finally, multiply the attention weight matrix by the value matrix V to obtain the final output matrix.

[0137] In order to more effectively utilize graph acceleration methods to process EEG data in parallel, a multi-attention mechanism can be used so that the subsequent computer can use hardware graph acceleration methods for parallel processing during training. Combining the above core models, the overall structure diagram can be expressed as follows Figure 3 shown.

[0138] The above framework will test the type of pleasant emotions and the type of test sources of the test population, and after comprehensive classification of the types, a preliminary judgment model will be obtained, dividing the population into three characteristic patterns: creativity, logic, and imagination.

[0139] The specific population directional training part is to divide the population into three categories: lateral creativity, logic, and imagination, and further reverse training; for the above three categories of people, different stimulus source libraries are established, and for the creativity group, a video library related to building construction, transportation track design, urban greening design, etc. is established; for the logical thinking group, a chess and go game library is established; for the imagination group, abstract thinking conversion graphic training is established.

[0140] The embodiment of the present invention provides a technology competition method using the key technology of EEG concentration reverse training based on the technology competition system using the key technology of EEG concentration reverse training, and the method specifically includes:

[0141] S1: The EEG acquisition module mainly samples EEG signals from a specific area for noise reduction, filtering and amplification, and transmits them to the Rockchip microprocessor (RK2206) controller via the serial port;

[0142] S2: The RK2206 controller loads the lightweight Hongmeng operating system, uses threads to amplify the processed EEG signals and sends them to the remote terminal and computer via Bluetooth;

[0143] S3: The computer side has a user front-end and a back-end service system. The user front-end can browse the user level report after the test or conduct concentration training for specific users; the back-end management system is responsible for browsing the back-end user information and EEG data.

[0144] The present invention provides a system and method based on EEG concentration reverse training technology, which is used to support concentration training and monitoring in science and technology competitions. The system consists of multiple modules, which complete EEG signal acquisition, processing, transmission, and data analysis operations in sequence to help contestants improve their concentration and monitor the training effect in real time. By collecting EEG signals in specific areas, the system can identify changes in concentration and improve contestants' ability to continue to concentrate through reverse training mode, providing scientific support for science and technology competitions.

[0145] First, in the EEG signal acquisition stage, the system obtains EEG signals from specific areas of the user's head through the EEG acquisition module. This module has signal noise reduction, filtering and signal amplification functions to ensure the accuracy and stability of the acquired signals. The processed signal will be transmitted to the RK2206 controller through the serial port for further data processing. The RK2206 controller is loaded with a lightweight Hongmeng operating system to ensure the stable operation of the system, and uses threads to efficiently call the Bluetooth function to transmit the processed EEG data to the remote terminal and computer, realizing real-time data sharing and visualization.

[0146] In the data receiving and display part, the computer side has built a user front-end and back-end service system. The user front-end provides an intuitive interface for contestants to view their concentration test results and generate a concentration level report based on the data. In addition, users can also conduct concentration training on the front-end and adjust their status in real time to improve their concentration ability. The back-end service system is used to manage and analyze user information and EEG data for administrators or coaches to monitor the training status and concentration status of contestants in real time. The back-end also has a data storage function to facilitate subsequent data analysis and competition evaluation.

[0147] The computer device provided by the present invention includes a memory and a processor, which can execute the above-mentioned reverse training method. A computer program is stored in the memory. When the program is loaded and run on the processor, the processor can perform functions such as EEG acquisition, signal transmission, data analysis and feedback. The system is implemented through a computer-readable storage medium, and the computer program runs in the processor, which can ensure the efficiency and portability of the system. The information data processing terminal, as the hardware implementation unit of the system, can perform real-time data acquisition, transmission and feedback.

[0148] The system provides a variety of data monitoring and user feedback interfaces, including the EEG acquisition circuit diagram ( Fig.10 )、Data transmission circuit diagram( Fig.11 ), and data receiving circuit diagram ( Fig.12 ). These circuit designs optimize the transmission stability of the system and ensure low-latency transmission of signals. Figures 4 to 6 Different user interfaces are displayed, on which users can view their concentration status in real time, receive training feedback, and conduct interactive concentration training. Figures 7 and 8 It is the system data end interface, displaying the user information management, data storage and analysis functions in the background service. Fig. 9 It displays the data monitoring results, visually showing the changes in the contestants' concentration during the training process.

[0149] Through this system, users can conduct systematic and scientific concentration training before the competition, significantly improving their concentration during the competition. At the same time, the system can effectively solve the problem of unquantifiable training effects in traditional concentration training methods, and use EEG data to provide real-time feedback on training effects, thereby improving the effectiveness and scientific nature of the training. The application of this system can help contestants in science and technology competitions improve their concentration in a short period of time, helping them achieve better results in the competition, while also providing strong data support for concentration research in the fields of education and competition.

[0150] It should be noted that the embodiments of the present invention can be implemented by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. It can be understood by a person of ordinary skill in the art that the above-mentioned devices and methods can be implemented using computer executable instructions and / or contained in a processor control code, such as a carrier medium such as a disk, CD or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. Such code is provided on the carrier medium. The device and its modules of the present invention can be implemented by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, and can also be implemented by a combination of the above-mentioned hardware circuits and software, such as firmware.

[0151] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any modifications, equivalent substitutions and improvements made by any technician familiar with the technical field within the technical scope disclosed by the present invention and within the spirit and principle of the present invention should be covered by the protection scope of the present invention.

Claims

1. A science and technology competition system using the key technology of EEG concentration reverse training, characterized in that: The system includes: The EEG acquisition terminal is used to sample EEG signals from a specific area for noise reduction, filtering and amplification, and transmit them to the RK2206 controller via the serial port; The RK2206 controller is connected to the EEG acquisition terminal, loads the lightweight Hongmeng operating system, and uses threads to send the processed EEG signals to the remote terminal and computer via Bluetooth; The Bluetooth transceiver module is connected to the RK2206 controller and is used to amplify the EEG signal and send it to the remote terminal via Bluetooth; The computer side is connected to the RK2206 controller and has a user front-end and a back-end service system. The user front-end can browse the user level report after the test or conduct concentration training for specific users; the back-end management system is responsible for browsing the back-end user information and EEG data; The remote terminal is connected to the Bluetooth transceiver module and is used to install the remote control tool to realize remote control.

2. According to claim 1, the technology competition system using the key technology of EEG concentration reverse training is characterized by: The EEG test and training algorithm of the EEG acquisition terminal specifically includes: an EEG data analysis part, an emotion pattern discrimination part, and a specific population direction training part.

3. The technology competition system using the key technology of EEG concentration reverse training as claimed in claim 2 is characterized by: The EEG data analysis part includes: anti-interference processing, noise reduction processing, and eigenvalue extraction based on power spectrum density; the anti-interference processing is to use a notch filter to process the interference value of a specific frequency, wherein the anti-interference function is: Where f0 is used to indicate the frequency range of the signal to be filtered out, and a and b are used to indicate the filter coefficients of the notch filter. The noise reduction process divides the EEG signal into windows through the fast Fourier transform (STFT). The signal in each window is stable, and then Fourier transform is performed on each window. The transformation formula is: οx(τ) is the original signal. οw(tT) is a window function. e -j2πfr is the kernel function of the Fourier transform. Then, for further analysis, it is necessary to extract the EEG eigenvalues ​​and perform power spectrum density conversion. The core conversion algorithm obtains the power spectrum intensity (PSI) of the kth part through the following formula: Where f is the sampling rate and N is the sequence length.

4. The technology competition system using the key technology of EEG concentration reverse training as claimed in claim 2 is characterized by: The emotion pattern identification part specifically includes: The system will establish a three-dimensional emotional pattern classification for the test population, which is divided into creativity, thinking, and imagination. The classification process first uses the EEG emotional pattern dataset for online training; the training method uses the SEED dataset, which is a collection of EEG datasets provided by the BCMI laboratory. The SEED dataset contains the EEG signals of the subjects when watching movie clips. The movie clips are carefully selected to evoke different types of emotions, including positive, negative, and neutral emotions; A neural network algorithm is used to classify groups with positive emotional experiences under specific audio-visual device stimulation; the SEED visual library is used to train the classification model; the overall process includes the following: classify the audio-visual stimulation sources into three types: creativity-oriented, logical thinking-oriented, and imagination-oriented; then test the assessment categories of the trainees' positive emotional experiences under each type; finally, use the assessment model to judge the testees' abilities; The algorithm model used is an improved spatiotemporal graph neural network; an effective method for learning long-distance dependencies in sequence data is used to establish a self-attention network in various time series analysis tasks.

5. The technology competition system using the key technology of EEG concentration reverse training as claimed in claim 4 is characterized in that: The improved spatiotemporal graph neural network calculation process has the following steps: (1) For a given input sequence Where n is the sequence length and d is the feature dimension. The corresponding query matrix is ​​obtained by linear transformation Key Matrix Sum Matrix Where Q = XW Q ,K=XW k ,V=XW v are three trainable parameter matrices, d q , d k , d v are the dimensions of the query vector, key vector, and value vector respectively: (2) Then, we calculate the dot product of the transposed query matrix and the key matrix K to obtain the attention score matrix S∈R n*n : S=QK T (3) In order to avoid the gradient vanishing problem caused by the excessive value obtained by the dot product calculation, the obtained attention score matrix needs to be scaled, generally by d k The square root of the key vector dimension is used to implement it, then the scaled attention score matrix S'∈R n*n It can be expressed as: (4) Perform a softmax operation on the attention score matrix to obtain a normalized attention weight matrix; (5) Finally, multiply the attention weight matrix by the value matrix V to obtain the final output matrix.

6. The technology competition system using the key technology of EEG concentration reverse training as claimed in claim 2 is characterized by: The specific population directional training part is to divide the population into three categories: lateral creativity, logic, and imagination, and further reverse training; for the above three categories of people, different stimulus source libraries are established, and for the creativity group, a video library related to building construction, transportation track design, urban greening design, etc. is established; for the logical thinking group, a chess and go game library is established; for the imagination group, abstract thinking conversion graphic training is established.

7. A scientific and technological competition method using the key technology of EEG concentration reverse training based on the scientific and technological competition system using the key technology of EEG concentration reverse training as described in claims 1-6, characterized in that: The method specifically includes: S1: The EEG acquisition module mainly samples EEG signals from a specific area for noise reduction, filtering and amplification, and transmits them to the RK2206 controller via the serial port; S2: The RK2206 controller loads the lightweight Hongmeng operating system and uses threads to send the processed EEG signals to the remote terminal and PC via Bluetooth. S3: The PC side has a user webpage and a background service system. The user webpage can browse the user level report after the test or conduct concentration training for specific users; the background management system is responsible for browsing the background user information and EEG data.

8. A computer device, characterized in that: The computer device includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the scientific and technological competition method using the key technology of reverse training of EEG concentration as described in claim 7.

9. A computer-readable storage medium storing a computer program, which, when executed by a processor, enables the processor to execute the steps of the scientific and technological competition method using the key technology of reverse training of EEG concentration as described in claim 7.

10. An information data processing terminal, characterized in that: The information data processing terminal is used to implement a science and technology competition system that uses the key technology of reverse training of EEG concentration as described in claims 1-6.

Images