A gong assisting sleep sound wave coupling regulation method and system

Abstract

The application provides a singing bowl sleep-aiding sound wave coupling regulation method and system, which utilizes subject electroencephalogram (EEG) signals and electrodermal activity (EDA) signals to train a random forest regression model to generate sleep-aiding effect scores corresponding to knocking parameters; according to the predicted sleep-aiding effect scores, musical notes and musical phrases are screened and spliced into a tree-shaped hierarchical parameterized music file; according to the playing instructions of a control terminal, the corresponding parameterized music is loaded, and the guiding explanation audio is played first; the tree-shaped music is parsed to generate single-knocking control parameters to drive the hammer to hit the singing bowl to output the singing bowl sound wave; and the user state is determined according to the user heart rate, heart rate variability (HRV) and body movement physiological parameters, and the actual knocking intensity, actual execution time and silent blank ratio of the musical phrase are adjusted according to the user state. The application screens musical notes and musical phrases through sleep-aiding effect scores, so that the music file meeting the user's sleep-aiding requirement is screened; and the knocking of the bowl body is controlled through the knocking control parameters, so that the sleep-aiding effect of the bowl body sound wave is improved.

Description

Technical Field

[0001] This invention relates to the fields of intelligent sound therapy for sleep aid, as well as deep learning and artificial intelligence technologies, and in particular to a method and system for sound wave coupling and modulation for sound bowl sleep aid. Background Technology

[0002] Currently, sleep intervention devices are mainly divided into two categories: electronic sound wave sleep aid devices and mechanical sound therapy devices. Representative solutions include white noise players, ultrasonic / infrasonic sleep regulation devices, adaptive sound wave systems based on respiratory feedback, artificial singing bowl therapy, and simple automatic singing bowl tapping devices.

[0003] However, pure audio playback solutions only output fixed sounds and cannot form regular physical striking sounds and singing bowl vibrations. Traditional singing bowl healing relies on manual striking, and the timing, intensity, and rhythm are difficult to standardize and cannot be stably reproduced. Furthermore, existing automatic striking devices only support simple start and stop or fixed beats and lack editable music structures, making it impossible to achieve multiple musical phrases, multiple notes, intensity changes, and rhythmic disturbances.

[0004] For example, patent CN120324804A discloses a method and device for whole-brain sleep regulation using ultrasound-infrasound coupled sound waves. This method uses EEG acquisition equipment and physiological state monitoring equipment to obtain the user's physiological parameters, and adjusts the ultrasonic parameters generated by the sound wave emitting device through a control device. It falls under the category of physical intervention and does not involve the automatic playing of physical musical instruments, nor does it disclose a parameterized music data structure or a multi-stage automatic tapping control method. Therefore, the existing technology is unlikely to achieve a sleep-aiding effect. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and system for sound wave coupling and modulation of singing bowls for sleep aid.

[0006] In a first aspect, the present invention provides a method for sound wave coupling and modulation for sleep aid using a singing bowl, comprising:

[0007] The subject's EEG and EDA signals were acquired, and the EEG frequency band power features and EDA conductance relaxation features were extracted based on the EEG and EDA signals, respectively, and then concatenated into an input feature vector.

[0008] Based on the combination of differentiated tapping parameters as perturbation stimuli, a training sample set of tapping parameters, EEG frequency band power characteristics, and skin conductance relaxation characteristics is established to train a random forest regression model. The trained random forest regression model is used to output the sleep aid effect score corresponding to the tapping parameters.

[0009] Based on the predicted sleep-aid effect score, musical notes and phrases are selected and assembled into a tree-structured, hierarchical, parameterized music file according to the sleep-inducing logic, and then written into the database;

[0010] The control terminal establishes communication links with both the singing bowl device and the wearable device, and loads the corresponding parameterized tracks according to the playback instructions from the control terminal, first playing the guiding audio.

[0011] After the guide audio finishes playing, the tree-structured parameterized track is analyzed, and single-strike control parameters are generated in real time through the gradation base intensity of musical phrases, as well as the basic intensity and duration of notes. Based on the strike control parameters, the hammer is driven to inertially strike the singing bowl and output the singing bowl sound wave.

[0012] Real-time collection of user heart rate, heart rate variability (HRV), and body movement physiological parameters; extraction of user heart rate features, HRV features, and body movement features through a sliding window; and classification of user state into tense state, relaxed state, and sleep transition state.

[0013] Adjust the actual striking intensity and execution duration of musical phrases, as well as the proportion of silent space, based on the user's status.

[0014] Secondly, the present invention provides a sound wave coupling and modulation system for using a singing bowl to aid sleep, comprising:

[0015] The data acquisition module is used to acquire the subject's EEG signal and EDA signal, and extract the EEG frequency band power features and EDA conductance relaxation features based on the EEG signal and EDA signal, respectively, and splice them into an input feature vector;

[0016] The sleep aid effect scoring module is used to train a random forest regression model by establishing a training sample set of tapping parameters, EEG frequency band power characteristics, and skin conductance relaxation characteristics based on differentiated tapping parameter combinations as perturbation stimuli. The trained random forest regression model is then used to output the sleep aid effect score corresponding to the tapping parameters.

[0017] The track construction module is used to filter notes and phrases based on the predicted sleep aid effect score, and to concatenate them into a tree-structured hierarchical parameterized track file according to the sleep logic, and then write it to the database;

[0018] The track loading module is used to establish communication links with the singing bowl device and the wearable device through the control terminal, and load the corresponding parameterized tracks according to the playback instructions of the control terminal, first playing the guiding explanation audio;

[0019] The sound bowl striking control module is used to analyze the tree-structured parameterized track after the guide audio has finished playing. It generates single striking control parameters in real time through the gradation reference intensity of musical phrases, as well as the basic intensity and duration of notes. Based on the striking control parameters, it drives the hammer to inertially strike the singing bowl and outputs the singing bowl sound wave.

[0020] The user state segmentation module is used to collect user heart rate, heart rate variability (HRV), and body movement physiological parameters in real time. It extracts user heart rate features, HRV features, and body movement features through a sliding window and classifies user state into tense state, relaxed state, and sleep transition state.

[0021] The adjustment and optimization module is used to adjust the actual striking intensity and execution duration of musical phrases, as well as the proportion of silent space, based on the user's status.

[0022] Thirdly, the present invention provides an electronic device, comprising:

[0023] At least one processor;

[0024] and a memory communicatively connected to the at least one processor;

[0025] The memory stores a computer program that can be executed by the at least one processor, and when the computer program is executed by the at least one processor, it implements the singing bowl sleep-aiding sound wave coupling control method.

[0026] The beneficial technical effects of this invention are as follows:

[0027] 1. This invention trains a random forest regression model using EEG signals and EDA signals to predict the sleep aid effect score corresponding to the tapping parameters. The sleep aid effect score is used to filter notes and musical phrases, thereby obtaining music files that meet the user's sleep aid needs.

[0028] 2. This invention analyzes the music and generates single-strike control parameters in real time based on the gradual change of musical phrase intensity, the basic intensity of notes, and the basic duration of notes. The striking of the bowl is controlled by the striking control parameters, thereby improving the sleep-aiding effect of the bowl's sound waves.

[0029] 3. This invention determines the user's state by tracking the user's heart rate and body movement in real time, and adjusts the tapping control parameters according to the user's state, thereby adjusting the sound wave effect of the bowl in real time according to different user states.

[0030] 4. This invention can increase the power of delta deep sleep waves, shorten the user's sleep latency, and retain only low-volume guiding voice during the silent blanking stage to achieve a smooth transition and relaxation. Attached Figure Description

[0031] Figure 1 This is a flowchart illustrating the method of an embodiment of the present invention;

[0032] Figure 2 This is a structural framework diagram of the system according to an embodiment of the present invention;

[0033] Figure 3 This is a schematic diagram of the structure of the bowl body according to an embodiment of the present invention;

[0034] In the diagram, 10 is the data acquisition module; 20 is the sleep aid effect scoring module; 30 is the track construction module; 40 is the track loading module; 50 is the bowl tapping control module; 60 is the user status classification module; and 70 is the adjustment and optimization module.

[0035] 1-Bowl body; 2-Support; 3-Motor; 4-Hammer; 5-Base. Detailed Implementation

[0036] The specific embodiments of the present invention will be further described below with reference to the accompanying drawings:

[0037] like Figure 1 As shown, this embodiment provides a method for sound wave coupling and modulation for sleep aid using singing bowls, including:

[0038] S1: Acquire the subject's EEG signal and EDA signal, and extract the EEG frequency band power feature and EDA conductance relaxation feature based on the EEG signal and EDA signal respectively, and concatenate them into an input feature vector;

[0039] S2: Based on the differentiated combination of tapping parameters as perturbation stimuli, a training sample set of tapping parameters, EEG frequency band power characteristics, and skin conductance relaxation characteristics is established to train a random forest regression model. The trained random forest regression model is used to output the sleep aid effect score corresponding to the tapping parameters.

[0040] S3: Filter musical notes and phrases based on the predicted sleep-aid effect score, and concatenate them into a tree-structured, hierarchical, parameterized music track file according to the sleep-inducing logic, and write it into the database;

[0041] S4: The control terminal establishes communication links with the singing bowl device and the wearable device respectively, and loads the corresponding parameterized track according to the playback command of the control terminal, and plays the guiding explanation audio first;

[0042] S5: After the guide audio finishes playing, the tree-structured parameterized track is analyzed. Single-strike control parameters are generated in real time through the gradation base intensity of musical phrases, as well as the basic intensity and duration of notes. Based on the strike control parameters, the hammer is driven to inertially strike the singing bowl and output the singing bowl sound wave.

[0043] S6: Real-time collection of user heart rate, heart rate variability (HRV), and body movement physiological parameters; extraction of user heart rate features, HRV features, and body movement features through a sliding window; and classification of user state into tense state, relaxed state, and sleep transition state.

[0044] S7: Adjust the actual striking intensity and execution duration of musical phrases, as well as the proportion of silent space, based on the user's status.

[0045] In this embodiment, step S1 involves acquiring the subject's EEG and EDA signals, and extracting EEG frequency band power features and EDA conductance relaxation features based on the EEG and EDA signals, respectively, and concatenating them into an input feature vector, including:

[0046] The power characteristics of the EEG frequency band include Rate of change of wave power spectral density relative to resting baseline , Wave power relative to resting baseline rate of change and absolute power of wave Complex characteristics of sleep depth EEG complexity feature H;

[0047] In this embodiment, the... The wave frequency is 0.5–4 Hz; The frequency range is 8–13 Hz; The frequency range is 4–8 Hz;

[0048] The aforementioned sleep depth composite features for Wave power and The ratio of wave power; the EEG complexity feature H is the Shannon entropy corresponding to the normalized power of each frequency band.

[0049] In this embodiment, the skin conductance relaxation feature includes:

[0050] Skin conductance baseline (SCL), the average skin conductance level within the measurement window;

[0051] Skin conductance response amplitude The difference between a single SCR peak value and the current baseline;

[0052] Skin conductance response frequency The total number of SCR pulse events occurring per unit time;

[0053] Skin conductance decrease slope The slope of the SCL baseline change within the measurement window.

[0054] The power features of the EEG frequency band and the relaxation features of the skin conductance are extracted and concatenated into the input feature vector.

[0055] In some embodiments, the subject's electroencephalogram (EEG) signals can be collected using an EEG acquisition cap, and the subject's electrodermal absorption (EDA) signals can be collected using electrode patches.

[0056] In this embodiment, in step S2, based on the differentiated tapping parameter combination as a perturbation stimulus, a training sample set of tapping parameters, EEG frequency band power characteristics, and skin conductance relaxation characteristics is established to train a random forest regression model. The trained random forest regression model is then used to output a sleep-aiding effect score corresponding to the tapping parameters, including:

[0057] S21: Design multiple sets of tapping parameters, and randomly shuffle them to play the differentiated tapping parameter stimulation of each set to the subjects in turn. Collect the EEG signal and EDA signal of the subjects under the stimulation of each set of differentiated tapping parameters, and extract the power characteristics of the EEG frequency band and the relaxation characteristics of the EDA conduction.

[0058] S22: Concatenate the power features of each EEG band with the skin conductance relaxation features as a training sample, and concatenate the input feature vector with the training sample to form a training sample set;

[0059] S23: Train the random forest regression model using the tapping parameter combination and the training sample set as input, and output a sleep aid effect score, i.e.:

[0060] ;

[0061] In the formula, Rate the sleep aid effect; This represents the random forest regression mapping function; For the tapping parameters; As training samples, The aforementioned tapping parameters ; The number of times a musical phrase is repeated; These are the initial intensity and the final intensity of a musical phrase, respectively. This represents the intensity of the disturbance. This is the duration of the perturbation. The basic duration of a musical note; The basic intensity of the note; The percentage of silent white space.

[0062] In this embodiment, during the training of the random forest regression model, the actual sleep-aiding effect is used as the regression label. ; and minimize the loss function as the optimization objective; wherein, the regression label Based on input feature vectors Rate of change of wave power spectral density relative to resting baseline With the slope of the decrease in skin conductance Construction, that is:

[0063] ;

[0064] In the formula, , These are the weighting coefficients;

[0065] The loss function The mean square error (MSE) is used, i.e.:

[0066] ;

[0067] In the formula, The number of training samples; For the first The true regression labels of each training sample; For the first The predicted sleep aid effect score for each training sample.

[0068] In this embodiment, step S3 involves filtering musical notes and phrases based on the predicted sleep-aid effect score, assembling them into a tree-structured hierarchical parameterized music track file according to sleep-inducing logic, and writing it into the database, including:

[0069] S31: Select candidates whose predicted sleep aid effect score is greater than or equal to the preset sleep aid effect threshold. Combinations of musical notes and phrases;

[0070] S32: Divide the combination of notes and phrases into guided relaxation sections, deep soothing sections, and sleep transition sections. Piece together the divided phrases and notes to construct a three-layer tree-structured hierarchical data structure containing a track layer, a phrase layer, and a note layer. Embed independent sleep aid scores as biofeedback-related fields in each level of the tree-structured hierarchical data structure and encapsulate them into a tree-structured parameterized track file.

[0071] In this embodiment, the music track layer stores music track identifiers, predicted sleep aid effect scores, and a set of musical phrase sequences;

[0072] The phrase layer stores the number of phrase repetitions N and the initial intensity of the phrase. Harmony phrase ending intensity Intensity disturbance Duration of perturbation .

[0073] The aforementioned note layer stores the basic duration of the notes. Basic strength of notes .

[0074] In this embodiment, in step S5, after the guiding audio playback is complete, the tree-structured parameterized track is parsed, and single-strike control parameters are generated in real time based on the gradation reference intensity of musical phrases, the basic intensity of notes, and the basic duration of notes. Based on the strike control parameters, the hammer is driven to inertially strike the singing bowl, outputting the singing bowl sound wave. This includes:

[0075] S51: Get the current musical phrase repetition cycle The number of times the musical phrase is repeated, N, if the number of times the musical phrase is repeated... If the musical phrase is determined to have been executed only once, then the base intensity of the musical phrase will gradually change. The initial intensity of a musical phrase ;

[0076] If the musical phrase is repeated a certain number of times The musical phrase gradually changes its base intensity. for: ;

[0077] S52: Based on the base note strength of the currently pending note. Gradual change in musical intensity Get the basic striking intensity of the note ,Right now:

[0078] ;

[0079] S53: Striking intensity based on the note Based on the superimposed intensity random offset After obtaining the amplitude limit, the actual striking intensity is obtained. ; and the basic duration of the notes Add random offset of duration on top of the base The actual execution time is obtained after minimum duration protection. ;Right now:

[0080] ;

[0081] ;

[0082] In the formula, This represents a random number generation function;

[0083] S54: Actual striking intensity and actual execution time As a single-impact control parameter, it is sent to the control terminal. If the actual impact intensity after amplitude limiting... If so, the physical tapping is skipped, and a silent waiting state is entered;

[0084] S55: The motor is designed according to the actual impact intensity. and actual execution time The corresponding PWM duty cycle continuously drives the hammer, outputting the sound wave of the singing bowl.

[0085] In this embodiment, step S6 involves real-time acquisition of the user's heart rate, heart rate variability (HRV), and body movement physiological parameters. The user's heart rate features, HRV features, and body movement features are extracted using a sliding window, and the user's state is categorized into a tense state, a relaxed state, and a sleep transition state, including:

[0086] S61: Real-time collection of users' heart rate (HR), heart rate variability (HRV), and body physiological parameters using wearable devices;

[0087] S62: Sample the user's heart rate (HR), heart rate variability (HRV), and body movement physiological parameters according to the preset sliding window length, and extract the heart rate features, HRV features, and body movement features of the data within the sliding window;

[0088] In this embodiment, the heart rate characteristics include average heart rate characteristics and the slope of heart rate rise or fall. The HRV characteristics include the window's average RMSSD value and the slope of HRV change. The body movement characteristics include the window's total body movement frequency and the amplitude of body movement fluctuations.

[0089] S63: If the average heart rate characteristic is greater than the resting baseline heart rate threshold, and the window average RMSSD value is less than the resting baseline RMSSD value, and the slope of heart rate rise or fall is greater than 0, and the slope of HRV change is less than 0; and the window total movement frequency is greater than the body movement threshold, then the user is judged to be in a state of tension.

[0090] If the average heart rate characteristic is less than the resting baseline heart rate threshold, and the window average RMSSD value is greater than the resting baseline RMSSD value, and the slope of heart rate rise or fall is less than 0, and the slope of HRV change is greater than 0; and the overall movement frequency of the window is less than the body movement threshold, then the user is judged to be in a relaxed state.

[0091] If the average heart rate characteristic is stable below the resting baseline heart rate threshold, and the window average RMSSD value is consistently greater than the resting baseline RMSSD value, and the window total movement frequency is consistently less than the body movement threshold, then the user is determined to be in a sleep transition state.

[0092] In some embodiments, the wearable device incorporates a PPG photoelectric sensor and a triaxial accelerometer. The PPG photoelectric sensor can collect the user's heart rate (HR) and heart rate variability (HRV). The triaxial accelerometer collects body motion physiological parameters.

[0093] In this embodiment, in step S7, the actual striking intensity of the musical phrase is adjusted according to the user's state. And the actual execution time and the proportion of silent white space, including:

[0094] If the user is in a state of tension, reduce the intensity ratio. Based on the reduced intensity ratio coefficient Adjust the base intensity of the musical phrase gradient, that is:

[0095] ;

[0096] In the formula, In this embodiment, to adjust the reference intensity of the musical phrase gradient, .

[0097] Increase duration disturbance coefficient In this embodiment, And based on the increased duration perturbation coefficient Adjust the random offset of the duration, that is:

[0098] ;

[0099] In the formula, This is the adjusted duration random offset.

[0100] By increasing the white space coefficient, the proportion of silent white space notes is increased, thus increasing the relaxation intervals without tapping.

[0101] ;

[0102] In the formula, The adjusted percentage of silent white space notes; This is a clipping function; The white space factor is the white space factor, which is adjusted to [value] in this embodiment. .

[0103] If the user is in a relaxed state, increase the intensity ratio. In this embodiment, Based on the reduced intensity ratio coefficient Adjust the base intensity of the musical phrase gradient.

[0104] The duration perturbation coefficient and baseline blank space coefficient Keep it unchanged, that is , .

[0105] If the user is in a sleep transition state, the intensity ratio coefficient is linearly reduced for each musical phrase. The intensity coefficient decreases by 0.1 with each change of musical phrase, down to a minimum of 0.2.

[0106] Reduce duration disturbance coefficient In this embodiment, the adjustment is made to ;

[0107] Step-by-step increase in white space ratio The piece concludes smoothly; in this embodiment, the blank space coefficient for each musical phrase is... Increased by 15%.

[0108] In this embodiment, if a stop command is received, the tasks of playing the track, striking the bowl, and connecting the communication link are interrupted layer by layer through the embedded asynchronous multi-task scheduling architecture.

[0109] like Figure 2 As shown, another embodiment of this application provides a sound wave coupling and modulation system for using a singing bowl to aid sleep, comprising:

[0110] The data acquisition module 10 is used to acquire the subject's EEG signal and EDA signal, and extract the EEG frequency band power features and EDA conductance relaxation features based on the EEG signal and EDA signal, respectively, and splice them into an input feature vector;

[0111] The sleep aid effect scoring module 20 is used to train a random forest regression model by establishing a training sample set of tapping parameters, EEG frequency band power characteristics, and skin conductance relaxation characteristics based on differentiated tapping parameter combinations as perturbation stimuli. The trained random forest regression model is used to output the sleep aid effect score corresponding to the tapping parameters.

[0112] The track construction module 30 is used to filter notes and phrases based on the predicted sleep aid effect score, and to concatenate them into a tree-structured hierarchical parameterized track file according to the sleep logic, and then write it into the database;

[0113] The track loading module 40 is used to establish communication links with the singing bowl device and the wearable device through the control terminal, and load the corresponding parameterized track according to the playback instructions of the control terminal, and first play the guiding explanation audio.

[0114] The sound bowl striking control module 50 is used to analyze the tree-structured parameterized track after the guide audio has finished playing. It generates single striking control parameters in real time through the gradation reference intensity of musical phrases, as well as the basic intensity and duration of notes. Based on the striking control parameters, it drives the hammer to inertially strike the singing bowl and outputs the singing bowl sound wave.

[0115] The user state segmentation module 60 is used to collect user heart rate, heart rate variability (HRV), and body movement physiological parameters in real time. It extracts user heart rate features, HRV features, and body movement features through a sliding window and classifies user state into tense state, relaxed state, and sleep transition state.

[0116] The adjustment and optimization module 70 is used to adjust the actual striking intensity and actual execution duration of musical phrases, as well as the proportion of silent space, according to the user's status.

[0117] In this embodiment, the data acquisition module 10 extracts brainwave frequency band power features based on the EEG signal, and the brainwave frequency band power features include Rate of change of wave power spectral density relative to resting baseline , Wave power relative to resting baseline rate of change and absolute power of wave Complex characteristics of sleep depth EEG complexity feature H.

[0118] Among them, the aforementioned The wave frequency is 0.5–4 Hz; The frequency range is 8–13 Hz; The frequency range is 4–8 Hz;

[0119] The aforementioned sleep depth composite features for Wave power and The ratio of wave power; the EEG complexity feature H is the Shannon entropy corresponding to the normalized power of each frequency band.

[0120] In this embodiment, the data acquisition module 10 extracts skin conductance relaxation features based on the skin conductance EDA signal, and the skin conductance relaxation features include:

[0121] Skin conductance baseline (SCL), the average skin conductance level within the measurement window;

[0122] Skin conductance response amplitude The difference between a single SCR peak value and the current baseline;

[0123] Skin conductance response frequency The total number of SCR pulse events occurring per unit time;

[0124] Skin conductance decrease slope The slope of the SCL baseline change within the measurement window.

[0125] The power features of the EEG frequency band and the relaxation features of the skin conductance are extracted and concatenated into the input feature vector.

[0126] In some embodiments, the subject's electroencephalogram (EEG) signals can be collected using an EEG acquisition cap, and the subject's electrodermal absorption (EDA) signals can be collected using electrode patches.

[0127] In this embodiment, the sleep aid effect scoring module 20 uses differentiated tapping parameter combinations as perturbation stimuli to establish a training sample set of tapping parameters, EEG frequency band power characteristics, and skin conductance relaxation characteristics to train a random forest regression model. The trained random forest regression model then outputs a sleep aid effect score corresponding to the tapping parameters, including:

[0128] Multiple sets of tapping parameters were designed, and the different tapping parameters of each set were played to the subjects in a randomized order. The subjects' EEG and EDA signals were collected under the stimulation of each set of different tapping parameters. At the same time, the power characteristics of the EEG frequency band and the relaxation characteristics of the EDA conduction were extracted.

[0129] Each group of EEG frequency band power features and skin conductance relaxation features are concatenated as a training sample, and the input feature vector is concatenated with the training sample to form a training sample set;

[0130] The random forest regression model is trained using the tapping parameter combination and the training sample set as input, and the output is a sleep aid effect score, i.e.:

[0131] ;

[0132] In the formula, Rate the sleep aid effect; This represents the random forest regression mapping function; For the tapping parameters; As training samples, The aforementioned tapping parameters ; The number of times a musical phrase is repeated; These are the initial intensity and the final intensity of a musical phrase, respectively. This represents the intensity of the disturbance. This is the duration of the perturbation. The basic duration of a musical note; The basic intensity of the note; The percentage of silent white space.

[0133] In this embodiment, during the training of the random forest regression model, the actual sleep-aiding effect is used as the regression label. ; and minimize the loss function as the optimization objective; wherein, the regression label Based on input feature vectors Rate of change of wave power spectral density relative to resting baseline With the slope of the decrease in skin conductance Construction, that is:

[0134] ;

[0135] In the formula, , These are the weighting coefficients;

[0136] The loss function The mean square error (MSE) is used, i.e.:

[0137] ;

[0138] In the formula, The number of training samples; For the first The true regression labels of each training sample; For the first The predicted sleep aid effect score for each training sample.

[0139] In this embodiment, the music track construction module 30 filters notes and phrases based on the predicted sleep-aid effect score, and concatenates them into a tree-structured hierarchical parameterized music track file according to sleep logic, and writes it into the database, including:

[0140] Screening for predicted sleep aid scores greater than or equal to the preset sleep aid threshold Combinations of musical notes and phrases;

[0141] The notes and phrases are divided into guided relaxation sections, deep soothing sections, and sleep transition sections. The divided phrases and notes are spliced ​​together to construct a three-layer tree-structured hierarchical data structure containing a track layer, a phrase layer, and a note layer. Independent sleep aid scores are embedded in each level of the tree-structured hierarchical data structure as biofeedback-related fields and encapsulated into a tree-structured parameterized track file.

[0142] In this embodiment, the music track layer stores music track identifiers, predicted sleep aid effect scores, and a set of musical phrase sequences;

[0143] The phrase layer stores the number of phrase repetitions N and the initial intensity of the phrase. Harmony phrase ending intensity Intensity disturbance Duration of perturbation .

[0144] The aforementioned note layer stores the basic duration of the notes. Basic strength of notes .

[0145] In this embodiment, after the guiding audio playback is complete, the singing bowl striking control module 50 analyzes the tree-structured parameterized music track, and generates single-strike control parameters in real time based on the gradation reference intensity of musical phrases, as well as the basic intensity and duration of notes. Based on these striking control parameters, it drives the hammer to inertially strike the singing bowl and outputs the singing bowl sound wave, including:

[0146] Get the current musical phrase repetition cycle The number of times the musical phrase is repeated, N, if the number of times the musical phrase is repeated... If the musical phrase is determined to have been executed only once, then the base intensity of the musical phrase will gradually change. The initial intensity of a musical phrase ;

[0147] If the musical phrase is repeated a certain number of times The musical phrase gradually changes its base intensity. for: ;

[0148] Based on the base note strength of the currently pending note. Gradual change in musical intensity Get the basic striking intensity of the note ,Right now:

[0149] ;

[0150] Striking intensity based on musical notes Based on the superimposed intensity random offset After obtaining the amplitude limit, the actual striking intensity is obtained. ; and the basic duration of the notes Add random offset of duration on top of the base The actual execution time is obtained after minimum duration protection. ;Right now:

[0151] ;

[0152] ;

[0153] In the formula, This represents a random number generation function;

[0154] Actual striking intensity and actual execution time As a single-impact control parameter, it is sent to the control terminal. If the actual impact intensity after amplitude limiting... If so, the physical tapping is skipped, and a silent waiting state is entered;

[0155] The motor is adjusted according to the actual impact intensity. and actual execution time The corresponding PWM duty cycle continuously drives the hammer, outputting the sound wave of the singing bowl.

[0156] In some embodiments, a schematic diagram of the bowl body may be referenced. Figure 3 The device includes a bowl body 1 and a hammer 4 disposed on one side of the bowl body 1. The hammer 4 is driven by a motor 3. In this embodiment, the bowl body 1 is disposed on a base 5, the motor 3 is disposed on a bracket 2, and the hammer 4 is connected to the motor shaft of the motor 3 through a hammer handle. The motor 3 is controlled by a control terminal.

[0157] In this embodiment, the user state classification module 60 collects the user's heart rate, heart rate variability (HRV), and body movement physiological parameters in real time. It extracts the user's heart rate features, HRV features, and body movement features through a sliding window and classifies the user's state into tension, relaxation, and sleep transition states, including:

[0158] Wearable devices are used to collect users’ heart rate (HR), heart rate variability (HRV), and body physiological parameters in real time.

[0159] The user's heart rate (HR), heart rate variability (HRV), and body movement physiological parameters are sampled according to a preset sliding window length, and the heart rate features, HRV features, and body movement features of the data within the sliding window are extracted.

[0160] In this embodiment, the heart rate characteristics include average heart rate characteristics and the slope of heart rate rise or fall. The HRV characteristics include the window's average RMSSD value and the slope of HRV change. The body movement characteristics include the window's total body movement frequency and the amplitude of body movement fluctuations.

[0161] If the average heart rate characteristic is greater than the resting baseline heart rate threshold, and the window average RMSSD value is less than the resting baseline RMSSD value, and the slope of heart rate rise or fall is greater than 0, and the slope of HRV change is less than 0; and the window total movement frequency is greater than the body movement threshold, then the user is judged to be in a state of tension.

[0162] If the average heart rate characteristic is less than the resting baseline heart rate threshold, and the window average RMSSD value is greater than the resting baseline RMSSD value, and the slope of heart rate rise or fall is less than 0, and the slope of HRV change is greater than 0; and the overall movement frequency of the window is less than the body movement threshold, then the user is judged to be in a relaxed state.

[0163] If the average heart rate characteristic is stable below the resting baseline heart rate threshold, and the window average RMSSD value is consistently greater than the resting baseline RMSSD value, and the window total movement frequency is consistently less than the body movement threshold, then the user is determined to be in a sleep transition state.

[0164] In some embodiments, the wearable device incorporates a PPG photoelectric sensor and a triaxial accelerometer. The PPG photoelectric sensor can collect the user's heart rate (HR) and heart rate variability (HRV). The triaxial accelerometer collects body motion physiological parameters. The wearable device can be a wristband or a smartwatch. In this embodiment, the wristband or smartwatch can connect to the control terminal via the standard Bluetooth GATT protocol.

[0165] In this embodiment, the adjustment and optimization module 70 adjusts the actual striking intensity of the musical phrase according to the user's state. And the actual execution time and the proportion of silent white space, including:

[0166] If the user is in a state of tension, reduce the intensity ratio. Based on the reduced intensity ratio coefficient Adjust the base intensity of the musical phrase gradient, that is:

[0167] ;

[0168] In the formula, In this embodiment, to adjust the reference intensity of the musical phrase gradient, .

[0169] Increase duration disturbance coefficient In this embodiment, And based on the increased duration perturbation coefficient Adjust the random offset of the duration, that is:

[0170] ;

[0171] In the formula, This is the adjusted duration random offset.

[0172] By increasing the white space coefficient, the proportion of silent white space notes is increased, thus increasing the relaxation intervals without tapping.

[0173] ;

[0174] In the formula, The adjusted percentage of silent white space notes; This is a clipping function; The white space factor is the white space factor, which is adjusted to [value] in this embodiment. .

[0175] If the user is in a relaxed state, increase the intensity ratio. In this embodiment, Based on the reduced intensity ratio coefficient Adjust the base intensity of the musical phrase gradient.

[0176] The duration perturbation coefficient and baseline blank space coefficient Keep it unchanged, that is , .

[0177] If the user is in a sleep transition state, the intensity ratio coefficient is linearly reduced for each musical phrase. The intensity coefficient decreases by 0.1 with each change of musical phrase, down to a minimum of 0.2.

[0178] Reduce duration disturbance coefficient In this embodiment, the adjustment is made to ;

[0179] Step-by-step increase in white space ratio The piece concludes smoothly; in this embodiment, the blank space coefficient for each musical phrase is... Increased by 15%.

[0180] In this embodiment, if a stop command is received, the tasks of playing the track, striking the bowl, and connecting the communication link are interrupted layer by layer through the embedded asynchronous multi-task scheduling architecture.

[0181] A third embodiment of the present invention provides an electronic device, comprising:

[0182] At least one processor;

[0183] and a memory communicatively connected to the at least one processor;

[0184] The memory stores a computer program that can be executed by the at least one processor, and when the computer program is executed by the at least one processor, it implements the singing bowl sleep-aiding sound wave coupling control method.

[0185] In this embodiment, the memory can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. A processor, coupled to the memory, is used to execute computer programs stored in the memory.

[0186] This application also provides a computer storage medium storing a computer program, which, when executed by a processor, implements the sound wave coupling and modulation method for sleep aid using singing bowls as described in the embodiments.

[0187] Computer programs include computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. Computer-readable media can include at least: any entity or device capable of carrying computer program code to a photographic / electronic device, recording media, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media. Examples include USB flash drives, portable hard drives, magnetic disks, or optical discs.

[0188] The embodiments and descriptions above are merely illustrative of the principles and preferred embodiments of the present invention. Various changes and modifications may be made to the present invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed.

Claims

1. A method for sound wave coupling and modulation for sleep aid using a singing bowl, characterized in that, include: The subject's EEG and EDA signals were acquired, and the EEG frequency band power features and EDA conductance relaxation features were extracted based on the EEG and EDA signals, respectively, and then concatenated into an input feature vector. Based on the combination of differentiated tapping parameters as perturbation stimuli, a training sample set of tapping parameters, EEG frequency band power characteristics, and skin conductance relaxation characteristics is established to train a random forest regression model. The trained random forest regression model is used to output the sleep aid effect score corresponding to the tapping parameters. Based on the predicted sleep-aid effect score, musical notes and phrases are selected and assembled into a tree-structured, hierarchical, parameterized music file according to the sleep-inducing logic, and then written into the database; The control terminal establishes communication links with both the singing bowl device and the wearable device, and loads the corresponding parameterized tracks according to the playback instructions from the control terminal, first playing the guiding audio. After the guide audio finishes playing, the tree-structured parameterized track is analyzed, and single-strike control parameters are generated in real time through the gradation base intensity of musical phrases, as well as the basic intensity and duration of notes. Based on the strike control parameters, the hammer is driven to inertially strike the singing bowl and output the singing bowl sound wave. Real-time collection of user heart rate, heart rate variability (HRV), and body movement physiological parameters; extraction of user heart rate features, HRV features, and body movement features through a sliding window; and classification of user state into tense state, relaxed state, and sleep transition state. Adjust the actual striking intensity and execution duration of musical phrases, as well as the proportion of silent space, based on the user's status.

2. The method for sound wave coupling and modulation for sleep aid using singing bowls according to claim 1, characterized in that: The power characteristics of the EEG frequency band include The rate of change of wave power spectral density relative to the resting baseline The rate of change of wave power relative to the resting baseline and Absolute power of waves, complex characteristics of sleep depth, and complex characteristics of electroencephalogram (EEG); The skin conductance relaxation characteristics include skin conductance baseline (SCL), skin conductance response amplitude, skin conductance response frequency, and skin conductance descent slope. The power features of the EEG frequency band and the relaxation features of the skin conductance are extracted and concatenated into the input feature vector.

3. The method for sound wave coupling and modulation for sleep aid using singing bowls according to claim 2, characterized in that: Using differentiated tapping parameter combinations as perturbation stimuli, a random forest regression model is trained using a training sample set of tapping parameters, EEG frequency power characteristics, and skin conductance relaxation characteristics. The trained random forest regression model then outputs a sleep-aiding effect score corresponding to the tapping parameters, including: Multiple sets of tapping parameters were designed, and the different tapping parameters of each set were played to the subjects in a randomized order. The subjects' EEG and EDA signals were collected under the stimulation of each set of different tapping parameters. At the same time, the power characteristics of the EEG frequency band and the relaxation characteristics of the EDA conduction were extracted. Each group of EEG frequency band power features and skin conductance relaxation features are concatenated as a training sample, and the input feature vector is concatenated with the training sample to form a training sample set; The random forest regression model is trained by using the combination of tapping parameters and the training sample set as input, and the output is a sleep aid effect score.

4. The method for sound wave coupling and modulation for sleep aid using singing bowls according to claim 3, characterized in that: Based on the predicted sleep-aid effect score, musical notes and phrases are selected and assembled into a hierarchical, parameterized music track file according to sleep onset logic, and then written into a database, including: Screening for predicted sleep aid scores greater than or equal to the preset sleep aid threshold Combinations of musical notes and phrases; The notes and phrases are divided into guided relaxation sections, deep soothing sections, and sleep transition sections. The divided phrases and notes are spliced ​​together to construct a three-layer tree-structured hierarchical data structure containing a track layer, a phrase layer, and a note layer. Independent sleep aid scores are embedded in each level of the tree-structured hierarchical data structure as biofeedback-related fields and encapsulated into a tree-structured parameterized track file.

5. The method for sound wave coupling and modulation for sleep aid using singing bowls according to claim 4, characterized in that: The aforementioned track layer stores track identifiers, predicted sleep-aid effect scores, and sets of musical phrase sequences; The phrase layer stores the number of phrase repetitions and the initial intensity of the phrase. Harmony phrase ending intensity, intensity perturbation amount, and duration perturbation amount; The note layer stores the basic duration and basic intensity of the notes.

6. The method for sound wave coupling and modulation for sleep aid using singing bowls according to claim 5, characterized in that: After the introductory audio finishes playing, the tree-structured, parameterized track is analyzed. Single-strike control parameters are generated in real-time based on the phrase gradient intensity, note base intensity, and note base duration. These parameters drive the hammer to inertially strike the singing bowl, outputting the singing bowl's sound waves, including: Get the current phrase repetition round and phrase repetition count. If the phrase repetition count is 1, it is determined that the phrase is executed only once, and the phrase gradient base intensity is the phrase starting intensity. If the number of repetitions of a musical phrase is greater than 1, then the basic intensity of the phrase will gradually change. for: ； In the formula, , The starting intensity of the musical phrase Harmony with the intensity of the musical phrase ending; The number of repetitions; For the current round; Based on the base intensity of the note to be executed and the baseline intensity of the phrase gradient. Get the basic striking intensity of the note ,Right now: ； Striking intensity based on musical notes Based on the superimposed intensity random offset After obtaining the amplitude limit, the actual striking intensity is obtained. ; and the basic duration of the notes Add random offset of duration on top of the base The actual execution time is obtained after minimum duration protection. ;Right now: ； ； In the formula, This represents a random number generation function; Actual striking intensity and actual execution time As a single-impact control parameter, it is sent to the control terminal. If the actual impact intensity after amplitude limiting... If so, the physical tapping is skipped, and a silent waiting state is entered; The motor is adjusted according to the actual impact intensity. and actual execution time The corresponding PWM duty cycle continuously drives the hammer, outputting the sound wave of the singing bowl.

7. The method for sound wave coupling and modulation for sleep aid using singing bowls according to claim 6, characterized in that: If the user is in a tense state, the intensity ratio coefficient is reduced, and the base intensity of the musical phrase is adjusted based on the reduced intensity ratio coefficient. Increase the duration perturbation coefficient, and based on the increased duration perturbation coefficient Adjust the random offset of the duration; Increase the proportion of silent notes by raising the white space coefficient, and increase the relaxation intervals without tapping.

8. The method for sound wave coupling and modulation for sleep aid using singing bowls according to claim 7, characterized in that: If the user is in a relaxed state, increase the intensity ratio coefficient; adjust the base intensity of the musical phrase gradient based on the increased intensity ratio coefficient. The duration perturbation coefficient and the baseline blanking coefficient remain unchanged; If the user is in a sleep transition state, the intensity ratio coefficient is linearly reduced for each musical phrase, decreasing by 0.1 for each musical phrase switched, down to a minimum of 0.2; The duration disturbance coefficient was lowered and the white space coefficient was increased in a stepped manner to smoothly end the track.

9. A sound wave coupling and control system for using a singing bowl to aid sleep, characterized in that, include: The data acquisition module is used to acquire the subject's EEG signal and EDA signal, and extract the EEG frequency band power features and EDA conductance relaxation features based on the EEG signal and EDA signal, respectively, and splice them into an input feature vector; The sleep aid effect scoring module is used to train a random forest regression model by establishing a training sample set of tapping parameters, EEG frequency band power characteristics, and skin conductance relaxation characteristics based on differentiated tapping parameter combinations as perturbation stimuli. The trained random forest regression model is then used to output the sleep aid effect score corresponding to the tapping parameters. The track construction module is used to filter notes and phrases based on the predicted sleep aid effect score, and to concatenate them into a tree-structured hierarchical parameterized track file according to the sleep logic, and then write it to the database; The track loading module is used to establish communication links with the singing bowl device and the wearable device through the control terminal, and load the corresponding parameterized tracks according to the playback instructions of the control terminal, first playing the guiding explanation audio; The sound bowl striking control module is used to analyze the tree-structured parameterized track after the guide audio has finished playing. It generates single striking control parameters in real time through the gradation reference intensity of musical phrases, as well as the basic intensity and duration of notes. Based on the striking control parameters, it drives the hammer to inertially strike the singing bowl and outputs the singing bowl sound wave. The user state segmentation module is used to collect user heart rate, heart rate variability (HRV), and body movement physiological parameters in real time. It extracts user heart rate features, HRV features, and body movement features through a sliding window and classifies user state into tense state, relaxed state, and sleep transition state. The adjustment and optimization module is used to adjust the actual striking intensity and execution duration of musical phrases, as well as the proportion of silent space, based on the user's status.

10. An electronic device, comprising: At least one processor; and a memory communicatively connected to the at least one processor; The memory stores a computer program that can be executed by the at least one processor, characterized in that, when the computer program is executed by the at least one processor, it implements the sound wave coupling and modulation method for sleep aiding with singing bowls as described in any one of claims 1-8.

Images