Brain-computer interface based multi-sensory collaborative self-feedback psychological relaxation method and system

Abstract

The application discloses a multi-sensory collaborative self-feedback psychological relaxation method and system based on a brain-computer interface, relates to the field of brain-computer interfaces and human-computer interaction, and proposes the following scheme, which comprises the following steps: window processing and screening of effective analysis windows of user electroencephalogram (EEG) data, generation of a local correction spectrum of the effective window EEG data, extraction of an instantaneous peak center, a half-power width, and calculation of an adjacent band leakage ratio; determination of a direction consistency degree and a change discrete degree based on the instantaneous peak center of the continuous effective window, and dynamic updating of a trajectory locking center; construction of a target narrow band and a reverse tail band according to the trajectory locking center, and calculation of a residence score; determination of a feedback interaction state in combination with the residence score and the adjacent band leakage ratio, driving of at least two output modes in audio, vision and touch to execute feedback, and formation of a brain-computer interaction closed loop. The application improves the anti-interference and dynamic adaptability of relaxation state recognition, guarantees the accurate matching of feedback timing and strength, and optimizes the closed-loop intervention effect of psychological relaxation training.

Description

Technical Field

[0001] This invention relates to the field of brain-computer interface and human-computer interaction technology, and more specifically, this application relates to a multi-sensory collaborative self-feedback psychological relaxation method and system based on brain-computer interface. Background Technology

[0002] With the development of wearable EEG acquisition devices, real-time signal processing terminals, and audio-visual-tactile interactive devices, identifying the user's psychological relaxation process using EEG activity and further intervening through auditory, visual, or tactile means has become a common technical approach in brain-computer interface-assisted relaxation training. Especially in non-invasive applications, the system typically needs to determine the current relaxation process based on short-term EEG changes under constraints of limited sampling accuracy, continuous noise perturbation, and real-time feedback, thereby determining subsequent feedback output.

[0003] Most existing methods use pre-defined relaxation-related frequency bands or individual frequency bands obtained from initial calibration as the basis for subsequent judgments, and directly generate feedback signals based on energy changes within these frequency bands during continuous processing. This approach is relatively straightforward in implementation and is therefore widely adopted. However, the target rhythm in EEG is not constant throughout the training process. As the user gradually transitions from tension to relaxation, the location and distribution range of its dominant frequency continuously change with the state. At the same time, electrode contact fluctuations, body movement interference, and fluctuations in adjacent frequency components all contribute to the short-term analysis results.

[0004] Therefore, if the system continues to use static frequency bands or makes judgments based solely on local peak values ​​at the current moment, it may mistakenly identify real but slowly migrating state changes as invalid changes, and it may also misjudge rises in adjacent frequency bands or noise disturbances as relaxation enhancements. This would cause the timing and intensity of feedback to become disconnected from the actual brain state, affecting the closed-loop regulation effect between the user and the feedback terminal. Thus, how to distinguish between real state migrations and surface fluctuations caused by noise or adjacent frequency disturbances during continuous EEG processing, considering the dynamic changes in the location and range of the target rhythm, and how to stably convert EEG changes into the basis for regulating audio, visual, and tactile feedback, has become a pressing technical problem to be solved. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a multi-sensory collaborative self-feedback psychological relaxation method and system based on brain-computer interface, which is used to stably map changes in a user's brain electrical activity into a multi-sensory feedback interactive state, thereby achieving closed-loop brain-computer interaction.

[0006] To achieve the above objectives, the technical solution of the present invention is as follows:

[0007] Firstly, this application provides a multi-sensory collaborative self-feedback psychological relaxation method based on a brain-computer interface, including:

[0008] The system acquires the user's EEG data, performs windowing processing on it, calculates the quality score of each analysis window to represent the usability of the window, and determines the analysis windows with a quality score greater than a preset quality score threshold as valid analysis windows.

[0009] The EEG data within the effective analysis window are subjected to spectral processing, and a local correction spectrum is generated within the preset frequency band search interval.

[0010] Based on the local calibration spectrum, the instantaneous peak center and half-power width of the effective analysis window are extracted. The instantaneous main band corresponding to the instantaneous peak center is constructed with the half-power width as the bandwidth, and adjacent bands with the same bandwidth are constructed on both sides of the instantaneous main band. The ratio of the sum of energy in each adjacent band to the energy in the instantaneous main band is calculated as the adjacent band leakage ratio.

[0011] The degree of directional consistency and variation dispersion are determined based on the instantaneous peak centers within a preset number of continuous effective analysis windows, and the initial trajectory locking center is determined based on the median value of the instantaneous peak centers.

[0012] When the adjacent zone leakage ratio, direction consistency, and variation dispersion meet the preset migration conditions, the trajectory locking center of the current analysis window is updated based on the trajectory locking center of the previous analysis window; otherwise, it is not updated.

[0013] A target narrowband is generated based on the trajectory locking center of the current analysis window and the preset target bandwidth. A reverse trailing band with the same bandwidth as the target narrowband is generated according to the changing direction of the trajectory locking center of the current analysis window and the previous analysis window. The logarithmic difference between the energy in the target narrowband and the energy in the reverse trailing band is calculated as the dwell score.

[0014] The feedback interaction state is determined by comparing the residence score and adjacent band leakage ratio with the preset state conditions, and a feedback control quantity is generated based on the feedback interaction state to control at least two of the audio output, visual output and tactile output to provide feedback to the user, thereby forming a brain-computer interaction closed loop.

[0015] Secondly, this application provides a brain-computer interface-based multi-sensory collaborative self-feedback psychological relaxation system for implementing the aforementioned brain-computer interface-based multi-sensory collaborative self-feedback psychological relaxation method, including:

[0016] The data acquisition module is used to acquire the user's EEG data, perform windowing processing on it, calculate the quality score of each analysis window to represent the usability of the window, and determine the analysis window with a quality score greater than the preset quality score threshold as a valid analysis window.

[0017] The calibration spectrum generation module is used to perform spectral processing on EEG data within the effective analysis window and generate a local calibration spectrum within a preset frequency band search interval.

[0018] The leakage ratio calculation module is used to extract the instantaneous peak center and half-power width of the effective analysis window based on the local calibration spectrum, construct the instantaneous main band corresponding to the instantaneous peak center with the half-power width as the bandwidth, and construct adjacent bands with the same bandwidth on both sides of the instantaneous main band; calculate the ratio of the sum of energy in each adjacent band to the energy in the instantaneous main band as the adjacent band leakage ratio;

[0019] The trajectory center update module is used to determine the degree of directional consistency and the degree of variation dispersion based on the instantaneous peak centers within a preset number of consecutive effective analysis windows, and to determine the initial trajectory locking center based on the median value of its instantaneous peak centers; when the adjacent band leakage ratio, the degree of directional consistency, and the degree of variation dispersion meet the preset migration conditions, the trajectory locking center of the current analysis window is updated based on the trajectory locking center of the previous analysis window; otherwise, it is not updated.

[0020] The scoring module is used to generate a target narrowband based on the trajectory locking center of the current analysis window and the preset target bandwidth, and to generate a reverse trailing band with the same bandwidth as the target narrowband according to the changing direction of the trajectory locking center of the current analysis window and the previous analysis window; the logarithmic difference between the energy in the target narrowband and the energy in the reverse trailing band is calculated as the dwell score.

[0021] The multi-sensory interactive output module is used to determine the feedback interaction state based on the comparison results of the residence score and the adjacent band leakage ratio with the preset state conditions, and to generate feedback control quantity based on the feedback interaction state to control at least two of the audio output, visual output and tactile output to perform feedback to the user, thereby forming a brain-computer interaction closed loop.

[0022] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0023] This application generates a local correction spectrum within a preset frequency band search interval, and further combines the instantaneous peak center, half-power width, instantaneous main band, and adjacent bands to construct the adjacent band leakage ratio. This solves the technical problem of existing schemes that rely only on local peak values ​​and have difficulty distinguishing between the real target rhythm and adjacent frequency rise or broadband noise. It achieves a joint characterization of the target rhythm center position, bandwidth range, and frequency band leakage degree, thereby improving the pertinence and anti-interference of state recognition.

[0024] This application solves the technical problems of single-window peaks being easily affected by instantaneous disturbances and the slow migration of target rhythms being difficult to track stably by extracting the degree of consistency of direction and the degree of dispersion of change based on the instantaneous peak center within a continuous and effective analysis window, and uses this to constrain the update of the trajectory locking center. It achieves controlled updates of the target rhythm migration trend, thereby reducing the risk of frequent deviations due to local fluctuations.

[0025] This application solves the technical problem that existing solutions cannot simultaneously characterize "new center establishment" and "old center residue". It generates a target narrow band based on the trajectory locking center and constructs a reverse trailing band according to the trajectory change direction. Then, it uses the logarithmic difference between the energy in the target narrow band and the energy in the reverse trailing band to form a dwell score. Combined with the adjacent band leakage ratio, it determines the feedback interaction state. This achieves a stable mapping of EEG state to audio, visual and tactile multi-sensory feedback control quantities, thereby improving the accuracy of feedback timing and the matching of feedback intensity in the brain-computer interaction closed loop. Attached Figure Description

[0026] The disclosure of this invention is illustrated with reference to the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. Wherein:

[0027] Figure 1 This is a flowchart of the multi-sensory collaborative self-feedback psychological relaxation method based on brain-computer interface proposed in this invention;

[0028] Figure 2 This is a schematic diagram of the structure of the multi-sensory collaborative self-feedback psychological relaxation system based on brain-computer interface proposed in this invention. Detailed Implementation

[0029] It is readily understood that, based on the technical solution of this invention, those skilled in the art can propose various interchangeable structural methods and implementations without altering the essential spirit of the invention. Therefore, the following detailed embodiments and accompanying drawings are merely illustrative examples of the technical solution of this invention and should not be considered as the entirety of the invention or as limitations or restrictions on the technical solution of this invention.

[0030] In existing brain-computer interface (BCI) psychological relaxation feedback systems, state recognition typically relies on pre-set fixed relaxation frequency bands or individual frequency bands obtained during the initial calibration phase as the basis for subsequent judgment. The problem is that these systems cannot dynamically adapt to changes in the center position and bandwidth of the target rhythm during continuous relaxation. When a user gradually transitions from a tense state to a relaxed state, the target rhythm in the EEG is not static but fluctuates continuously in frequency position, energy distribution range, and adjacent frequency leakage. Simultaneously, changes in electrode contact, body motion artifacts, and fluctuations in adjacent frequency components all contribute to the short-term analysis results. At this point, a deviation arises between the static frequency band and the dynamic migration process of the actual EEG target rhythm, causing the system's feedback judgment to become disconnected from the user's true relaxation state. This results in feedback lag, false amplification, or erroneous freezing, especially when the target rhythm is slowly migrating but has not yet formed a significant energy mutation, making it difficult for the system to accurately identify the true relaxation process.

[0031] For example, during a continuous relaxation training session, the user's initial target rhythm center is located around 9.8Hz, gradually shifting towards around 10.2Hz as the training progresses, while the bandwidth increases from 2.6Hz to 3.0Hz. At this point, the EEG acquisition system detects a continuous shift in local peaks within the target frequency band, while the energy at adjacent frequencies also locally increases with background fluctuations. Traditional systems still use the fixed frequency band calibrated in the initial stage to determine the relaxation state, failing to consider the migration trend of the rhythm center and the influence of adjacent frequency energy leakage. Therefore, on the one hand, the system may misjudge genuine rhythm migration as invalid fluctuations, resulting in a delayed enhancement of the feedback output; on the other hand, it may misjudge the increase in adjacent frequency energy as an enhancement of relaxation, prematurely increasing the intensity of audio, visual, or tactile feedback, making the output feedback inconsistent with the user's actual brain state.

[0032] If the above problems are not addressed, brain-computer interface systems will be unable to stably distinguish between real state transitions and surface fluctuations caused by noise or adjacent frequency perturbations during continuous EEG processing, leading to a long-term deviation of the feedback interaction state from the actual relaxation process. If the system continues to make judgments based on mismatched static frequency bands or local peaks, it may miss the optimal feedback window during real relaxation enhancement, reducing the training induction effect; it may also mistakenly trigger enhanced feedback before the target rhythm has stabilized, causing feedback rhythm disorder, weakening the user's trust and reliance on the feedback terminal, and ultimately affecting the effectiveness and engineering applicability of the entire brain-computer interaction closed loop.

[0033] To address the aforementioned issues, this application first analyzes the root cause of the failure of traditional fixed-band determination mechanisms, finding that its core lies in the failure to consider the coordinated changes in the target rhythm center location, bandwidth range, and adjacent-band leakage during continuous relaxation. To resolve this, this application attempts to first screen the usability of the EEG analysis window. Based on this, it generates a local correction spectrum through spectral processing, extracts the instantaneous peak center and half-power width from it, and constructs the main band and adjacent bands corresponding to the instantaneous peak center. The ratio of adjacent band energy to main band energy characterizes the degree of adjacent-band leakage. Furthermore, this application introduces the directional consistency and variation dispersion of the instantaneous peak center within the continuously effective analysis window to stably constrain the migration trend of the target rhythm, thus forming the trajectory locking center update logic. Simultaneously, outside the updated target narrowband, a reverse trailing band is constructed based on the trajectory change direction. A residence score is generated by comparing the energy difference between the target narrowband and the reverse trailing band, and the feedback interaction state is determined by combining the adjacent-band leakage ratio. Through comprehensive analysis of continuous EEG fluctuations, rhythm transfer, and multisensory feedback coupling, this application ultimately develops a closed-loop interactive method capable of stably recognizing dynamic changes in target rhythms and mapping the recognition results to audio, visual, and tactile feedback control quantities.

[0034] Example 1:

[0035] Reference Figure 1 As shown, this application introduces a multi-sensory collaborative self-feedback psychological relaxation method based on brain-computer interface, including:

[0036] S1. Acquire the user's EEG data, perform windowing processing on it, calculate the quality score of each analysis window to represent the usability of the analysis window, and determine the analysis window with a quality score greater than the preset quality score threshold as a valid analysis window.

[0037] Specifically, the EEG data output from the user's EEG acquisition device is acquired, and DC-DC removal, power frequency notch filtering, and bandpass filtering are performed on the EEG data to obtain preprocessed EEG data. For example, EEG data can be acquired using three channels: O1, O2, and Pz, with a sampling frequency of 250Hz and a bandpass filtering range of 1Hz to 30Hz. Then, the preprocessed EEG data is windowed based on a preset window length and a preset step size. The windowing of the EEG data can be performed using a known sliding window method, that is, continuous signals are truncated according to a fixed window length, and then updated by rolling according to a fixed step size, thereby forming a sequence of consecutive analysis windows on a continuous time axis. For example, if the preset window length is 2s and the preset step size is 0.5s, 17 analysis windows can be obtained within a 10s acquisition time, with each analysis window corresponding to 500 sampling points.

[0038] After obtaining each analysis window, the amplitude stability data, high-frequency contamination data, low-frequency anomaly data, and channel consistency data corresponding to each analysis window are acquired and normalized using the min-max method to obtain multiple quality sub-indicators. Then, based on a preset weighting factor (e.g., all values ​​are 0.25), a weighted fusion process is performed on the quality sub-indicators to obtain the quality score for each analysis window. For example, the quality score threshold can be set to 0.75. When the quality score corresponding to a certain analysis window is 0.835, the analysis window is determined to be a valid analysis window; when the quality score corresponding to a certain analysis window is 0.62, the analysis window is determined to be an invalid analysis window. Finally, analysis windows with quality scores greater than the preset quality score threshold are output as valid analysis windows.

[0039] Among them, amplitude stable data is obtained by peak-to-peak value, extreme value distribution or abnormal jumps within the statistical analysis window; high-frequency pollution data is obtained by the power ratio or energy ratio in the high-frequency band within the statistical analysis window; low-frequency abnormal data is obtained by the energy change, drift amplitude or slow change trend in the low-frequency range within the statistical analysis window; and channel consistency data is obtained by comparing the peak position, peak shape or energy distribution relationship of multiple channels in the target frequency band.

[0040] Through the above technical solution, this application first performs preprocessing on EEG data to reduce the influence of power frequency interference, baseline drift and noise outside the target frequency band. Then, a continuous analysis window is obtained with a fixed window length and a fixed step size. Furthermore, a quality score is constructed from four dimensions: amplitude stability, high frequency contamination, low frequency anomaly and channel consistency. This allows analysis windows with higher usability to be selected as effective analysis windows, avoiding low-quality analysis windows from entering the subsequent spectrum analysis process.

[0041] S2. Perform spectral processing on the EEG data within the effective analysis window and generate a local correction spectrum within the preset frequency band search interval;

[0042] The process of generating the local correction spectrum includes:

[0043] EEG data within the effective analysis window is acquired, and spectral estimation processing is performed on it to obtain the power spectrum characterizing the energy intensity at different frequency positions;

[0044] Obtain the low-frequency reference band located on the low-frequency side of the frequency band search interval and the high-frequency reference band located on the high-frequency side;

[0045] Calculate the logarithmic power of the power spectrum in the low-frequency and high-frequency reference bands and average them to obtain the low-frequency average logarithmic power and the high-frequency average logarithmic power.

[0046] The local background line is obtained by linear interpolation of the frequency band search interval based on the low-frequency average logarithmic power and the high-frequency average logarithmic power, and a local correction spectrum is generated based on the difference between the logarithmic spectrum of the power spectrum and the local background line.

[0047] The process of obtaining the local background line includes:

[0048] The center frequency of the low-frequency reference band is used as the low-frequency reference position, the center frequency of the high-frequency reference band is used as the high-frequency reference position, and the positions of each frequency within the preset frequency band search interval are obtained.

[0049] Calculate the normalized position ratio of each frequency position relative to the low-frequency reference position and the high-frequency reference position;

[0050] Based on the normalized location ratio, linear interpolation is performed on the low-frequency average logarithmic power and the high-frequency average logarithmic power to obtain the background power value at that frequency location.

[0051] Linear interpolation is repeatedly performed on each frequency position within the frequency band search interval to obtain the background power value sequence at each frequency position, and the background power value sequence is determined as the local background line;

[0052] Wherein, the background power value corresponding to frequency position f satisfies:

[0053] ;

[0054] in, Indicates the background power value. This represents the low-frequency average logarithmic power. Represents the high-frequency average logarithmic power. Indicates the low-frequency reference position. Indicates the high-frequency reference position.

[0055] Specifically, the EEG data within the current valid analysis window is acquired, and spectral estimation processing is performed on it to obtain the power spectrum. For example, the current valid analysis window corresponds to EEG data sampled at 2s and 250Hz, with each channel containing 500 sampling points. Welch spectral estimation can be performed on the O1, O2, and Pz channels respectively, and the average can be calculated to obtain the power spectrum corresponding to the current valid analysis window.

[0056] Then, a preset frequency band search range and reference frequency bands located on its low-frequency and high-frequency sides are obtained. For example, the preset frequency band search range can be set to 7.5Hz to 12.5Hz, the low-frequency reference frequency band can be set to 6.5Hz to 7.5Hz, and the high-frequency reference frequency band can be set to 12.5Hz to 13.5Hz.

[0057] The logarithmic power mean of the power spectrum in the low-frequency and high-frequency reference bands is further calculated to obtain the low-frequency average logarithmic power and the high-frequency average logarithmic power. For example, the low-frequency average logarithmic power can be 1.72, and the high-frequency average logarithmic power can be 1.34. Then, using the center frequency of the low-frequency reference band and the center frequency of the high-frequency reference band as the two reference positions, linear interpolation is performed on each frequency position within the preset frequency band search interval to obtain the local background line;

[0058] The logarithmic spectrum of the power spectrum is then subtracted point-by-point from the local background line to obtain the local correction spectrum. For example, at the 10.0 Hz position, if the logarithmic spectrum value corresponding to the power spectrum is 1.96 and the background power value corresponding to the local background line is 1.53, then the value of the local correction spectrum at that position is 0.43. The final output local correction spectrum is used for subsequent extraction of the instantaneous peak center and half-power width.

[0059] Through the above technical solution, this application first transforms the EEG data in the effective analysis window to the frequency domain to obtain the power spectrum; then, it constructs a local background line using the reference frequency bands on both sides of the preset frequency band search interval; finally, it obtains a local correction spectrum by subtracting the logarithmic spectrum from the local background line, thereby reducing the influence of local background tilt on subsequent peak recognition, and making the subsequent instantaneous peak center extraction more focused on the real local convex structure of the target rhythm.

[0060] S3. Based on the local correction spectrum, extract the instantaneous peak center and half-power width of the effective analysis window, construct the instantaneous main band corresponding to the instantaneous peak center with the half-power width as the bandwidth, and construct adjacent bands with the same bandwidth on both sides of the instantaneous main band; calculate the ratio of the sum of energy in each adjacent band to the energy in the instantaneous main band as the adjacent band leakage ratio.

[0061] Among them, the instantaneous peak center and half-power width of the effective analysis window extracted based on the local correction spectrum include:

[0062] The local peak positions of the local calibration spectrum within the preset frequency band search interval are obtained, and the frequencies corresponding to the local peak positions are determined as candidate peak centers.

[0063] When there is only one candidate peak center, the candidate peak center is determined as the instantaneous peak center;

[0064] When the number of candidate peak centers is greater than one, obtain the peak size and peak width corresponding to each candidate peak center. When the instantaneous peak center exists in the previous effective analysis window, obtain the frequency distance between each candidate peak center and the instantaneous peak center of the previous effective analysis window. Determine the instantaneous peak center based on the peak size, peak width, and frequency distance.

[0065] The instantaneous peak center spectral value in the local correction spectrum is obtained as the peak center spectral value, and the half power threshold is obtained by subtracting the preset attenuation amount from the peak center spectral value.

[0066] Search from the instantaneous peak center towards the low frequency direction in the local correction spectrum. When the corresponding spectral value is less than or equal to the half power threshold, the corresponding frequency position is determined as the low frequency half power point.

[0067] Search from the instantaneous peak center towards the high frequency direction in the local correction spectrum. When the corresponding spectral value is less than or equal to the half-power threshold, the corresponding frequency position is determined as the high-frequency half-power point.

[0068] The half-power width is determined based on the frequency interval between the low-frequency half-power point and the high-frequency half-power point.

[0069] The process of determining the instantaneous peak center based on peak size, peak width, and frequency distance includes:

[0070] Compare the peak values ​​of each candidate peak center, and when there is only one candidate peak center with the largest peak value, determine the candidate peak center with the largest peak value as the instantaneous peak center;

[0071] When there is not a unique candidate peak center with the largest peak value, compare the peak widths of the candidate peak centers with the largest peak values ​​and determine the candidate peak center with the smallest peak width as the instantaneous peak center;

[0072] When the candidate peak center with the smallest peak width is not unique and the instantaneous peak center of the previous effective analysis window exists, the frequency distances are compared, and the candidate peak center with the smallest frequency distance is determined as the instantaneous peak center.

[0073] Specifically, the local calibration spectrum corresponding to the current effective analysis window is obtained, the local peak position is searched within the preset frequency band search interval, and the frequency corresponding to the local peak position is determined as the candidate peak center; when the candidate peak center is unique, the candidate peak center is determined as the instantaneous peak center; when the candidate peak center is not unique, the instantaneous peak center is determined by combining the peak size, peak width, and frequency distance between the candidate peak center and the instantaneous peak center corresponding to the previous effective analysis window.

[0074] Then, the spectral value of the instantaneous peak center in the local correction spectrum is obtained as the peak center spectral value, and a preset attenuation is subtracted from the peak center spectral value to obtain the half-power threshold. The system searches from the instantaneous peak center towards both low and high frequencies, determining the low-frequency and high-frequency half-power points when the corresponding spectral values ​​are less than or equal to the half-power threshold, and determining the half-power width based on the frequency interval between them. For example, if the local correction spectral value of the 8th effective analysis window at 10.0 Hz is 0.43 and the preset attenuation is 0.30, then the half-power threshold is 0.13; if the corresponding low-frequency half-power point is 8.5 Hz and the high-frequency half-power point is 11.5 Hz, then the half-power width is 3.0 Hz.

[0075] Subsequently, an instantaneous main band corresponding to the instantaneous peak center is constructed using the half-power width as the bandwidth, and adjacent bands with the same bandwidth are constructed on both sides. The spectral values ​​of the local correction spectrum that are greater than zero in the instantaneous main band and adjacent bands are integrated or summed to obtain the energy in the instantaneous main band and the sum of the energies in each adjacent band. The ratio of the sum of the energies in each adjacent band to the energy in the instantaneous main band is then determined as the adjacent band leakage ratio. For example, if the energy in the instantaneous main band is 1.52 and the sum of the energies in each adjacent band is 0.62, then the adjacent band leakage ratio is approximately 0.41. Finally, the instantaneous peak center, half-power width, and adjacent band leakage ratio are output.

[0076] Through the above technical solution, this application first extracts the instantaneous peak center in the local correction spectrum, then determines the half-power width based on the peak center spectral value and the half-power threshold, and then constructs the instantaneous main band and adjacent bands around the instantaneous peak center. The adjacent band leakage ratio is obtained by calculating the ratio of the adjacent band energy to the instantaneous main band energy, so that the subsequent trajectory update no longer depends on the single peak position, but simultaneously considers the peak concentration and the adjacent band leakage.

[0077] S4. Determine the degree of directional consistency and the degree of variation dispersion based on the instantaneous peak centers within a preset number of continuous effective analysis windows, and determine the initial trajectory locking center based on the median value of its instantaneous peak centers; when the adjacent zone leakage ratio, the degree of directional consistency, and the degree of variation dispersion meet the preset migration conditions, update the trajectory locking center of the current analysis window based on the trajectory locking center of the previous analysis window, otherwise do not update;

[0078] Among them, determining the degree of directional consistency and variation dispersion based on the instantaneous peak centers within a preset number of continuous effective analysis windows, and determining the initial trajectory locking center based on the median value of its instantaneous peak centers, includes:

[0079] Obtain the instantaneous peak centers corresponding to a preset number of consecutive valid analysis windows, and arrange them in the order of the analysis windows to form a peak center sequence;

[0080] A center change sequence is generated based on the difference between adjacent instantaneous peak centers in the peak center sequence.

[0081] The degree of directional consistency is determined based on the proportion of the number of differences with the same sign in the central change sequence to the total number of differences;

[0082] The degree of dispersion of change is determined based on the median absolute deviation of the central change sequence;

[0083] The initial trajectory locking center is determined based on the median value of the instantaneous peak center in the peak center sequence.

[0084] Specifically, when the adjacent zone leakage ratio, directional consistency, and variation dispersion meet preset migration conditions, the trajectory locking center of the current analysis window is updated based on the trajectory locking center of the previous analysis window; otherwise, it is not updated. This includes:

[0085] Obtain the trajectory locking center of the previous analysis window, as well as the adjacent band leakage ratio, directional consistency, and variation dispersion of the current analysis window;

[0086] When the adjacent band leakage ratio, directional consistency and change dispersion meet the preset migration conditions, the median change of the central change sequence is obtained as the update step size.

[0087] Based on the trajectory locking center and update step size of the previous analysis window, determine the trajectory locking center of the current analysis window;

[0088] The current analysis window's trajectory locking center satisfies:

[0089] ;

[0090] in, This indicates the center of the trajectory lock corresponding to the current analysis window. This indicates the center of the trajectory lock corresponding to the previous analysis window. This indicates the update step size corresponding to the current analysis window, expressed as the median change. It's absolutely worth it. This indicates the migration condition determination result of the current analysis window. A value of 1 indicates that the migration condition is met, and a value of 0 indicates that the migration condition is not met.

[0091] The migration conditions include: the adjacent band leakage ratio corresponding to the current analysis window is not greater than the preset adjacent band leakage ratio threshold, the directional consistency degree corresponding to the current analysis window is not less than the preset directional consistency degree threshold, and the change dispersion degree corresponding to the current analysis window is not greater than the preset change dispersion degree threshold.

[0092] When the adjacent zone leakage ratio, direction consistency, and variation dispersion do not meet the preset migration conditions, the trajectory locking center of the previous analysis window is used as the trajectory locking center of the current analysis window.

[0093] Specifically, the instantaneous peak centers corresponding to a preset number of consecutive effective analysis windows are obtained and arranged in the order of the analysis windows to form a peak center sequence. Then, based on the differences between adjacent instantaneous peak centers in the peak center sequence, a center change sequence is obtained. Next, the degree of directional consistency is determined based on the proportion of the number of differences with the same sign in the center change sequence to the total number of differences; the degree of change dispersion is determined based on the median absolute deviation of the center change sequence; and the initial trajectory locking center is determined based on the median value of the instantaneous peak centers in the peak center sequence. For example, if the instantaneous peak centers corresponding to five consecutive effective analysis windows are 9.8Hz, 10.0Hz, 10.1Hz, 10.2Hz, and 10.2Hz, respectively, then the center change sequence corresponding to the peak center sequence is 0.2Hz, 0.1Hz, 0.1Hz, 0Hz; 0.2Hz, 0.1Hz, 0Hz, 0.1Hz, 0Hz; 0.2Hz, 0.1Hz, 0.1Hz, 0Hz; 0.2Hz, 0.1Hz, 0.1Hz, 0Hz. This yields a directional consistency of 0.75, a change dispersion of 0.05Hz, and an initial trajectory locking center of 10.1Hz.

[0094] Furthermore, when the adjacent band leakage ratio, directional consistency, and variation dispersion meet preset migration conditions, the median change of the center variation sequence is obtained as the update step size, and the trajectory locking center of the current analysis window is updated based on the trajectory locking center of the previous analysis window; otherwise, the trajectory locking center of the previous analysis window remains unchanged. For example, if the adjacent band leakage ratio corresponding to the current analysis window is 0.41, the preset adjacent band leakage ratio threshold is 0.60, the preset directional consistency threshold is 0.75, and the preset variation dispersion threshold is 0.10Hz, then the current analysis window meets the preset migration conditions; if the trajectory locking center of the previous analysis window is 10.0Hz and the median change of the center variation sequence is 0.1Hz, then the trajectory locking center of the current analysis window can be updated to 10.1Hz; if the adjacent band leakage ratio corresponding to the current analysis window is 0.82, then the trajectory locking center of the previous analysis window remains unchanged. The final output is the directional consistency, variation dispersion, and the trajectory locking center of the current analysis window.

[0095] Through the above technical solution, this application, at the level of continuous effective analysis windows, no longer determines the subsequent target frequency band position solely based on the instantaneous peak center of a single effective analysis window. Instead, it first extracts the degree of directional consistency and the degree of variation dispersion through the peak center sequence, then combines the adjacent band leakage ratio to determine whether the current analysis window meets the migration conditions. Finally, when the migration conditions are met, the trajectory locking center of the current analysis window is updated based on the trajectory locking center and update step size of the previous analysis window, thereby reducing the impact of local jitter of the instantaneous peak center on the narrowband position of the subsequent target.

[0096] S5. Generate a target narrow band based on the trajectory locking center of the current analysis window and the preset target bandwidth, and generate a reverse trailing band with the same bandwidth as the target bandwidth according to the changing direction of the trajectory locking center of the current analysis window and the previous analysis window; calculate the logarithmic difference between the energy in the target narrow band and the energy in the reverse trailing band as the dwell score.

[0097] Specifically, a target narrowband is generated based on the trajectory locking center of the current analysis window and the preset target bandwidth, and a reverse trailing band is generated according to the changing direction of the trajectory locking center between the current analysis window and the previous analysis window, including:

[0098] Obtain the trajectory lock center and preset target bandwidth of the current analysis window;

[0099] Based on the trajectory locking center and preset target bandwidth of the current analysis window, a target narrowband is generated;

[0100] The preset target bandwidth is obtained by performing a weighted fusion process on the half-power width corresponding to the current effective analysis window and the median value of the historical half-power width based on a preset weight factor.

[0101] Compare the size relationship between the trajectory lock center of the current analysis window and the trajectory lock center of the previous analysis window to determine the direction of change;

[0102] When the trajectory lock center of the current analysis window is greater than the trajectory lock center of the previous analysis window, the direction of change is determined as the high-frequency direction;

[0103] When the trajectory locking center of the current analysis window is smaller than the trajectory locking center of the previous analysis window, the direction of change is determined to be the low-frequency direction;

[0104] When the trajectory lock center of the current analysis window is equal to the trajectory lock center of the previous analysis window, the direction of change is determined as the direction of no change.

[0105] When the direction of change is high frequency, a reverse trailing band is generated on the low frequency side of the target narrowband;

[0106] When the direction of change is low frequency, a reverse trailing band is generated on the high frequency side of the target narrow band;

[0107] When the trajectory lock center of the current analysis window is the same as the trajectory lock center of the previous analysis window, the reverse trailing band corresponding to the previous analysis window remains unchanged.

[0108] The process of calculating the logarithmic difference between the energy within the target narrowband and the energy within the reverse trailing band as the dwell score includes:

[0109] The energy within the target narrowband is obtained by integrating and summing the spectral values ​​of the local correction spectrum that are greater than zero within the target narrowband.

[0110] The energy within the reverse tail band is obtained by integrating and summing the spectral values ​​of the local correction spectrum that are greater than zero within the reverse tail band.

[0111] The dwell score is obtained based on the logarithmic difference between the energy in the target narrow band and the energy in the reverse trailing band.

[0112] Among them, the residency score satisfies:

[0113] ;

[0114] in, Indicates the score for stay. This represents the energy within the target's narrow band. This indicates the energy within the reverse trailing band. Indicates a preset positive number;

[0115] Specifically, the trajectory lock center of the current analysis window, the half-power width corresponding to the current effective analysis window, and the median value of the half-power width corresponding to the historical effective analysis windows are obtained. Then, based on a preset weighting factor, a weighted fusion is performed on the half-power width corresponding to the current effective analysis window and the median value of the half-power width corresponding to the historical effective analysis windows to obtain the preset target bandwidth. For example, if the trajectory lock center of the current analysis window is 10.1Hz, the half-power width corresponding to the current effective analysis window is 3.0Hz, and the median value of the half-power width corresponding to the historical effective analysis windows is 2.8Hz, and the weights of the two are 0.6 and 0.4 respectively, then the preset target bandwidth can be obtained as 2.92Hz.

[0116] Subsequently, a target narrowband is generated with the trajectory lock center of the current analysis window as the center and the preset target bandwidth as the bandwidth. If the trajectory lock center of the previous analysis window is 10.0Hz, since the trajectory lock center of the current analysis window (10.1Hz) is greater than that of the previous analysis window (10.0Hz), the direction of change can be determined as the high-frequency direction, and a reverse trailing band with the same bandwidth as the target narrowband is generated on the low-frequency side of the target narrowband. For example, a target narrowband of 8.64Hz to 11.56Hz and a reverse trailing band of 5.72Hz to 8.64Hz can be obtained. If the trajectory lock center of the current analysis window is smaller than that of the previous analysis window, a reverse trailing band is generated on the high-frequency side of the target narrowband; if the two are equal, the reverse trailing band of the previous effective analysis window remains unchanged. Finally, the target narrowband and the reverse trailing band are output.

[0117] Through the above technical solution, this application determines the target narrowband position based on the trajectory locking center of the current analysis window, and determines the target narrowband bandwidth by combining the half-power width corresponding to the current effective analysis window and the median value of the half-power width corresponding to the historical effective analysis windows. Then, it generates a reverse trailing band based on the changing direction between the trajectory locking centers of the current analysis window and the previous analysis window, thereby providing a basis for the subsequent simultaneous evaluation of the energy in the target narrowband and the energy in the reverse trailing band.

[0118] S6. Based on the comparison results of the residence score and adjacent band leakage ratio with the preset state conditions, determine the feedback interaction state, and generate a feedback control quantity based on the feedback interaction state to control at least two of the audio output, visual output and tactile output to provide feedback to the user, thereby forming a brain-computer interaction closed loop.

[0119] The preset state conditions include enhancement conditions, maintenance conditions, and freeze conditions;

[0120] The enhancement condition is that the residence score is not lower than the enhancement score threshold and the adjacent zone leakage ratio is not higher than the enhancement leakage threshold. The maintenance condition is that the residence score is not lower than the maintenance score threshold and the adjacent zone leakage ratio is not higher than the maintenance leakage threshold. The freezing condition is that the enhancement condition or the maintenance condition is not met.

[0121] Among them, the enhancement score threshold is higher than the maintenance score threshold, and the enhancement leakage threshold is lower than the maintenance leakage threshold.

[0122] Specifically, the target narrowband and reverse trailing band corresponding to the current analysis window are obtained. The spectral values ​​of the local correction spectrum that are greater than zero within the target narrowband and reverse trailing band are integrated or summed to obtain the energy within the target narrowband and the energy within the reverse trailing band, respectively. The dwell score is then determined based on the logarithmic difference between the two. For example, if the energy within the target narrowband is 1.48, the energy within the reverse trailing band is 0.52, and the preset positive value is 0.01, then the dwell score is approximately 1.03.

[0123] Then, the adjacent band leakage ratio corresponding to the current analysis window is obtained, and the dwell score and adjacent band leakage ratio are compared with preset state conditions to determine the feedback interaction state. For example, if the enhancement score threshold is 0.90 and the enhancement leakage threshold is 0.50, then when the dwell score is 1.03 and the adjacent band leakage ratio is 0.41, the feedback interaction state is determined to be the enhancement state; if the dwell score is 0.56 and the adjacent band leakage ratio is 0.62, the feedback interaction state can be determined to be the maintenance state; if the dwell score is 0.18 or the adjacent band leakage ratio is 0.83, the feedback interaction state can be determined to be the frozen state.

[0124] Furthermore, a feedback control quantity is generated based on the feedback interaction state, and the feedback control quantity is used to control at least two of the following output modes: audio output, visual output, and tactile output, to perform feedback. For example, in the enhanced state, the volume coefficient of the audio output and the brightness coefficient of the visual output can be increased; in the maintenance state, the corresponding feedback parameters can be kept unchanged; and in the frozen state, the feedback parameters can be decreased or stopped from increasing further. Finally, the feedback interaction state and the feedback control quantity are output, and the corresponding feedback is executed, thereby forming a closed loop of brain-computer interaction.

[0125] Through the above technical solution, this application uses the energy in the target narrow band and the energy in the reverse trailing band to obtain the dwell score, and then substitutes the dwell score and the adjacent band leakage ratio into the preset state conditions to determine the feedback interaction state. Finally, based on the feedback interaction state, a feedback control quantity is generated and at least two output modes are driven to perform feedback, so that the feedback signal reflects both the dwell status of the target rhythm in the target narrow band and is constrained by the adjacent band leakage situation.

[0126] Example 2:

[0127] like Figure 2 As shown, this application provides a brain-computer interface-based multi-sensory collaborative self-feedback psychological relaxation system for implementing the aforementioned brain-computer interface-based multi-sensory collaborative self-feedback psychological relaxation method, including:

[0128] The data acquisition module 100 is used to acquire the user's EEG data, perform windowing processing on it, calculate the quality score of each analysis window representing the usability of the window, and determine the analysis window with a quality score greater than the preset quality score threshold as a valid analysis window.

[0129] The calibration spectrum generation module 200 is used to perform spectral processing on the EEG data within the effective analysis window and generate a local calibration spectrum within a preset frequency band search interval.

[0130] The leakage ratio calculation module 300 is used to extract the instantaneous peak center and half-power width of the effective analysis window based on the local correction spectrum, construct the instantaneous main band corresponding to the instantaneous peak center with the half-power width as the bandwidth, and construct adjacent bands with the same bandwidth on both sides of the instantaneous main band; calculate the ratio of the sum of energy in each adjacent band to the energy in the instantaneous main band as the adjacent band leakage ratio;

[0131] The trajectory center update module 400 is used to determine the degree of directional consistency and the degree of variation dispersion based on the instantaneous peak centers within a preset number of consecutive effective analysis windows, and to determine the initial trajectory locking center based on the median value of its instantaneous peak centers; when the adjacent band leakage ratio, the degree of directional consistency, and the degree of variation dispersion meet the preset migration conditions, the trajectory locking center of the current analysis window is updated based on the trajectory locking center of the previous analysis window; otherwise, it is not updated.

[0132] The scoring calculation module 500 is used to generate a target narrowband based on the trajectory locking center of the current analysis window and the preset target bandwidth, and to generate a reverse trailing band with the same bandwidth as the target narrowband according to the changing direction of the trajectory locking center of the current analysis window and the previous analysis window; the logarithmic difference between the energy in the target narrowband and the energy in the reverse trailing band is calculated as the dwell score.

[0133] The multi-sensory interactive output module 600 is used to determine the feedback interaction state based on the comparison results of the residence score and the adjacent band leakage ratio with the preset state conditions, and to generate a feedback control quantity based on the feedback interaction state to control at least two of the audio output, visual output and tactile output to give feedback to the user, thereby forming a brain-computer interaction closed loop.

[0134] The multi-sensory collaborative self-feedback psychological relaxation system based on brain-computer interface proposed in this embodiment belongs to the same inventive concept as the multi-sensory collaborative self-feedback psychological relaxation method based on brain-computer interface proposed in the above embodiments. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0135] The technical scope of this invention is not limited to the content described above. Those skilled in the art can make various modifications and variations to the above embodiments without departing from the technical concept of this invention, and all such modifications and variations should fall within the protection scope of this invention.

Claims

1. A multi-sensory collaborative self-feedback psychological relaxation method based on brain-computer interface, characterized in that, include: The system acquires the user's EEG data, performs windowing processing on it, calculates the quality score of each analysis window to represent the usability of the window, and determines the analysis windows with a quality score greater than a preset quality score threshold as valid analysis windows. The EEG data within the effective analysis window are subjected to spectral processing, and a local correction spectrum is generated within the preset frequency band search interval. Based on the local calibration spectrum, the instantaneous peak center and half-power width of the effective analysis window are extracted. The instantaneous main band corresponding to the instantaneous peak center is constructed with the half-power width as the bandwidth, and adjacent bands with the same bandwidth are constructed on both sides of the instantaneous main band. The ratio of the sum of energy in each adjacent band to the energy in the instantaneous main band is calculated as the adjacent band leakage ratio. The degree of directional consistency and variation dispersion are determined based on the instantaneous peak centers within a preset number of continuous effective analysis windows, and the initial trajectory locking center is determined based on the median value of the instantaneous peak centers. When the adjacent zone leakage ratio, direction consistency, and variation dispersion meet the preset migration conditions, the trajectory locking center of the current analysis window is updated based on the trajectory locking center of the previous analysis window; otherwise, it is not updated. A target narrowband is generated based on the trajectory locking center of the current analysis window and the preset target bandwidth, and a reverse trailing band with the same bandwidth as the target narrowband is generated according to the changing direction of the trajectory locking center of the current analysis window and the previous analysis window. The logarithmic difference between the energy in the target narrowband and the energy in the reverse trailing band is calculated as the dwell score; The feedback interaction state is determined by comparing the residence score and adjacent band leakage ratio with the preset state conditions, and a feedback control quantity is generated based on the feedback interaction state to control at least two of the audio output, visual output and tactile output to provide feedback to the user, thereby forming a brain-computer interaction closed loop.

2. The multi-sensory collaborative self-feedback psychological relaxation method based on brain-computer interface according to claim 1, characterized in that, The process of generating a local correction spectrum includes: Spectral estimation processing is performed on the EEG data within the effective analysis window to obtain the power spectrum characterizing the energy intensity at different frequency locations; Obtain the low-frequency reference band located on the low-frequency side of the frequency band search interval and the high-frequency reference band located on the high-frequency side; Calculate the logarithmic power of the power spectrum in the low-frequency and high-frequency reference bands and average them to obtain the low-frequency average logarithmic power and the high-frequency average logarithmic power. The local background line is obtained by linear interpolation of the frequency band search interval based on the low-frequency average logarithmic power and the high-frequency average logarithmic power, and a local correction spectrum is generated based on the difference between the logarithmic spectrum of the power spectrum and the local background line.

3. The multi-sensory collaborative self-feedback psychological relaxation method based on brain-computer interface according to claim 2, characterized in that, The process of obtaining a local background line includes: The center frequency of the low-frequency reference band is used as the low-frequency reference position, the center frequency of the high-frequency reference band is used as the high-frequency reference position, and the positions of each frequency within the preset frequency band search interval are obtained. Calculate the position ratio of each frequency position relative to the low-frequency reference position and the high-frequency reference position; Based on the location ratio, linear interpolation is performed on the low-frequency average logarithmic power and the high-frequency average logarithmic power to obtain the background power value at that frequency location. Linear interpolation is repeatedly performed on each frequency position within the frequency band search interval to obtain the background power value sequence at each frequency position, and the background power value sequence is determined as the local background line; Wherein, the background power value corresponding to frequency position f satisfies: ； in, Indicates the background power value. This represents the low-frequency average logarithmic power. Represents the high-frequency average logarithmic power. Indicates the low-frequency reference position. Indicates the high-frequency reference position.

4. The multi-sensory collaborative self-feedback psychological relaxation method based on brain-computer interface according to claim 1, characterized in that, The process of extracting the instantaneous peak center and half-power width includes: The local peak positions of the local calibration spectrum within the preset frequency band search interval are obtained, and the frequencies corresponding to the local peak positions are determined as candidate peak centers. When there is only one candidate peak center, the candidate peak center is determined as the instantaneous peak center; When the number of candidate peak centers is greater than one, obtain the peak size and peak width corresponding to each candidate peak center, as well as the frequency distance between each candidate peak center and the instantaneous peak center of the previous effective analysis window, and determine the instantaneous peak center based on the peak size, peak width, and frequency distance. The instantaneous peak center spectral value in the local correction spectrum is obtained as the peak center spectral value, and the half power threshold is obtained by subtracting the preset attenuation amount from the peak center spectral value. Search from the instantaneous peak center to the low frequency direction in the local correction spectrum. When the corresponding spectral value is less than or equal to the half power threshold, the corresponding frequency position is determined as the low frequency half power point. In the local correction spectrum, search from the instantaneous peak center towards the high frequency direction. When the corresponding spectral value is less than or equal to the half-power threshold, the corresponding frequency position is determined as the high-frequency half-power point. The half-power width is determined based on the frequency interval between the low-frequency half-power point and the high-frequency half-power point.

5. The multi-sensory collaborative self-feedback psychological relaxation method based on brain-computer interface according to claim 4, characterized in that, The process of determining the instantaneous peak center based on peak size, peak width, and frequency distance includes: Compare the peak values ​​of each candidate peak center, and when there is only one candidate peak center with the largest peak value, determine the candidate peak center with the largest peak value as the instantaneous peak center; When there is not a unique candidate peak center with the largest peak value, compare the peak widths of the candidate peak centers with the largest peak values ​​and determine the candidate peak center with the smallest peak width as the instantaneous peak center; When the candidate peak center with the smallest peak width is not unique and the instantaneous peak center of the previous effective analysis window exists, the frequency distances are compared, and the candidate peak center with the smallest frequency distance is determined as the instantaneous peak center.

6. The multi-sensory collaborative self-feedback psychological relaxation method based on brain-computer interface according to claim 1, characterized in that, The process of determining the degree of directional consistency and the degree of dispersion of change, as well as the initial trajectory center locking, includes: Obtain the instantaneous peak centers corresponding to a preset number of consecutive valid analysis windows, and arrange them in the order of the analysis windows to form a peak center sequence; A center change sequence is generated based on the difference between adjacent instantaneous peak centers in the peak center sequence. The degree of directional consistency is determined based on the proportion of the number of differences with the same sign in the central change sequence to the total number of differences; The degree of dispersion of change is determined based on the median absolute deviation of the central change sequence; The initial trajectory locking center is determined based on the median value of the instantaneous peak center in the peak center sequence.

7. The multi-sensory collaborative self-feedback psychological relaxation method based on brain-computer interface according to claim 1, characterized in that, The process of updating the trajectory locking center includes: When the adjacent band leakage ratio, directional consistency and change dispersion meet the preset migration conditions, the median change of the central change sequence is obtained as the update step size. The migration conditions include that the adjacent band leakage ratio is not greater than a preset adjacent band leakage ratio threshold, the directional consistency is not less than a preset directional consistency threshold, and the degree of variation dispersion is not greater than a preset degree of variation dispersion threshold. Based on the trajectory locking center and update step size of the previous analysis window, determine the trajectory locking center of the current analysis window; Wherein, the trajectory locking center satisfies: ； in, This indicates the updated trajectory center lock. This indicates that the trajectory of the previous analysis window is locked at the center. This indicates the update step size, derived from the median change. It's absolutely worth it. This indicates the migration condition determination result of the current analysis window.

8. The multi-sensory collaborative self-feedback psychological relaxation method based on brain-computer interface according to claim 1, characterized in that, The process of generating the target narrowband and the reverse trailing band includes: Obtain the trajectory lock center, half-power width, and historical half-power width median value of the current analysis window; The preset target bandwidth is obtained by performing a weighted fusion of the half-power width of the current effective analysis window and the median value of the historical half-power width based on the preset weighting factors. Based on the trajectory locking center and preset target bandwidth of the current analysis window, a target narrowband is generated; Compare the trajectory lock center of the current analysis window with the trajectory lock center of the previous analysis window. If the current window's center is larger, the direction of change is determined to be a high-frequency direction; if the current window's center is smaller, the direction of change is determined to be a low-frequency direction; if the current window's center is equal to the previous window's center, the direction of change is determined to be a direction with no change. When the direction of change is high frequency, a reverse trailing band with the same bandwidth as the target narrowband is generated on the low frequency side of the target narrowband; When the direction of change is low frequency, a reverse trailing band is generated on the high frequency side of the target narrowband; When the direction of change is unchanged, the reverse trailing band of the previous effective analysis window remains unchanged.

9. The multi-sensory collaborative self-feedback psychological relaxation method based on brain-computer interface according to claim 8, characterized in that, The process of calculating the residence score and determining the feedback state based on the comparison between the residence score and the adjacent zone leakage ratio and the preset state conditions includes: Integrating and summing the spectral values ​​of the local correction spectrum that are greater than zero in the target narrowband and the reverse tail band, we obtain the energy in the target narrowband and the energy in the reverse tail band. The dwell score is obtained based on the logarithmic difference between the energy in the target narrow band and the energy in the reverse trailing band. The residency score is satisfied: ; in, Indicates the score for stay. This represents the energy within the target's narrow band. This indicates the energy within the reverse trailing band. Indicates a preset positive number; The feedback state is determined based on the comparison results of the residence score and adjacent zone leakage ratio with the preset state conditions; The preset state conditions include enhancement conditions, maintenance conditions, and freeze conditions; The enhancement condition is that the residence score is not lower than the enhancement score threshold and the adjacent zone leakage ratio is not higher than the enhancement leakage threshold. The maintenance condition is that the residence score is not lower than the maintenance score threshold and the adjacent zone leakage ratio is not higher than the maintenance leakage threshold. The freezing condition is that the enhancement condition or the maintenance condition is not met. Among them, the enhancement score threshold is higher than the maintenance score threshold, and the enhancement leakage threshold is lower than the maintenance leakage threshold.

10. A multi-sensory collaborative self-feedback psychological relaxation system based on brain-computer interface, characterized in that, A brain-computer interface-based multi-sensory collaborative self-feedback psychological relaxation method for implementing any one of claims 1-9 includes: The data acquisition module is used to acquire the user's EEG data, perform windowing processing on it, calculate the quality score of each analysis window to represent the usability of the window, and determine the analysis window with a quality score greater than the preset quality score threshold as a valid analysis window. The calibration spectrum generation module is used to perform spectral processing on EEG data within the effective analysis window and generate a local calibration spectrum within a preset frequency band search interval. The leakage ratio calculation module is used to extract the instantaneous peak center and half-power width of the effective analysis window based on the local calibration spectrum, construct the instantaneous main band corresponding to the instantaneous peak center with the half-power width as the bandwidth, and construct adjacent bands with the same bandwidth on both sides of the instantaneous main band; calculate the ratio of the sum of energy in each adjacent band to the energy in the instantaneous main band as the adjacent band leakage ratio; The trajectory center update module is used to determine the degree of directional consistency and the degree of variation dispersion based on the instantaneous peak centers within a preset number of consecutive effective analysis windows, and to determine the initial trajectory locking center based on the median value of its instantaneous peak centers; when the adjacent band leakage ratio, the degree of directional consistency, and the degree of variation dispersion meet the preset migration conditions, the trajectory locking center of the current analysis window is updated based on the trajectory locking center of the previous analysis window; otherwise, it is not updated. The scoring module is used to generate a target narrowband based on the trajectory locking center of the current analysis window and the preset target bandwidth, and to generate a reverse trailing band with the same bandwidth as the target narrowband according to the changing direction of the trajectory locking center of the current analysis window and the previous analysis window; the logarithmic difference between the energy in the target narrowband and the energy in the reverse trailing band is calculated as the dwell score. The multi-sensory interactive output module is used to determine the feedback interaction state based on the comparison results of the residence score and the adjacent band leakage ratio with the preset state conditions, and to generate feedback control quantity based on the feedback interaction state to control at least two of the audio output, visual output and tactile output to perform feedback to the user, thereby forming a brain-computer interaction closed loop.

Images