Sleep improvement environment sound control auxiliary method and system

Abstract

The invention relates to the technical field of sleep data processing, in particular to a sleep improvement environment sound control auxiliary method and system. The method comprises the following steps: acquiring sleep state data; performing sleep state feature extraction according to the sleep state data to obtain sleep state feature data; performing environmental sound type selection according to the sleep state feature data to obtain environmental sound type data; according to the environment sound type data and the sleep state feature data, environment audio rate adjustment and environment volume adjustment are carried out, environment audio rate adjustment data and environment volume adjustment data are obtained respectively, and sleep improvement environment sound control auxiliary work is carried out. According to the invention, the system can better adapt to the long-term sleep mode change of the user, and can help the user gradually establish a healthier sleep habit through long-term adjustment and optimization, thereby realizing long-term improvement of the sleep mode.

Description

Technical Field

[0001] The present invention relates to the technical field of sleep data processing, and in particular to a method and system for assisting in controlling ambient sound for improving sleep. Background Art

[0002] With the accelerated pace of modern life and the increase in stress, sleep quality has become one of the increasingly prominent health issues worldwide. Insufficient sleep or poor sleep quality can have a negative impact on an individual's physical and mental health, leading to fatigue, decreased concentration, reduced immunity, and even an increased risk of chronic diseases. Therefore, how to improve and enhance sleep quality has become a key area of ​​research and technology development.

[0003] There are already some sleep aids on the market, such as white noise machines, smart sleep monitoring devices, etc. These devices assist users in sleeping by playing stable background sounds or monitoring the user's physiological signals. However, traditional white noise machines or ambient sound playback devices usually play preset audio types and volumes in a static manner. This method fails to dynamically adjust the audio according to the specific needs of the user in different sleep stages, and it is difficult to adapt to state changes during sleep. The user's sleep consists of multiple stages, including light sleep, deep sleep, and rapid eye movement, and each stage has different needs for sound. Devices that lack dynamic adaptability cannot provide sufficient support when the user is in light sleep or awake due to environmental interference, affecting sleep quality. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a sleep-improving ambient sound control auxiliary method and system to solve at least one of the above technical problems.

[0005] The present application provides a method for assisting in controlling ambient sound for improving sleep, comprising the following steps:

[0006] Step S1: Obtaining sleep state data;

[0007] Step S2: extracting sleep state features according to the sleep state data to obtain sleep state feature data;

[0008] Step S3: Selecting an ambient sound type according to the sleep state characteristic data to obtain ambient sound type data;

[0009] Step S4: adjusting the ambient sound frequency and the ambient sound volume according to the ambient sound type data and the sleep state characteristic data, and obtaining ambient sound frequency adjustment data and ambient sound volume adjustment data, respectively, to perform a sleep-improving ambient sound control auxiliary operation.

[0010] In the present invention, the user's sleep state data is obtained through multimodal sensors (such as heart rate sensors, breathing sensors and acceleration sensors), so that the entire system can respond to the user's current sleep stage, which helps to achieve highly targeted sleep assistance. By extracting the features of the sleep state data, the user's sleep state characteristics can be deeply understood, and whether they are in light sleep, deep sleep or rapid eye movement (REM) can be accurately determined. The sleep state characteristic data generated in this step provides a scientific basis for the selection and adjustment of ambient sound, so that the ambient sound can be dynamically adjusted according to the actual needs of the user, and the system's response speed and accuracy to changes in the user's sleep state are improved. According to the sleep state characteristic data, selecting a suitable type of ambient sound (such as white noise, natural sound, etc.) can help the user smoothly enter the next stage or maintain the current sleep state. For example, selecting low-frequency white noise suitable for deep sleep or soft natural sound suitable for light sleep can promote the user's deep sleep and avoid sudden sound changes from interfering with sleep. Through the selection of audio types, the system can more effectively assist users to gradually enter deep sleep or maintain a stable rapid eye movement period, thereby improving overall sleep quality. The audio frequency and volume are adjusted based on the ambient sound type and sleep state characteristic data, so that the played ambient sound is more in line with the user's current sleep stage. For example, when the user enters deep sleep, the system will lower the volume or adjust the frequency to reduce interference; in the light sleep stage, the system will gradually increase the audio volume to help the user enter a deeper sleep. The dynamic adjustment function can enhance the adaptability of ambient sound and ensure that the user is always in the best sound environment throughout the sleep cycle.

[0011] Preferably, step S1 specifically includes:

[0012] Step S11: collecting the user's heart rate data through a heart rate sensor to obtain the user's heart rate data;

[0013] Step S12: collecting the user's respiratory frequency data through a respiratory sensor to obtain the user's respiratory frequency data;

[0014] Step S13: collecting user body motion data through an acceleration sensor to obtain user body motion data;

[0015] Step S14: determining the user's sleep state based on the user's heart rate data, the user's breathing rate data, and the user's body movement data to obtain sleep state data.

[0016] The heart rate changes in the present invention show significant differences in different stages of sleep (light sleep, deep sleep, rapid eye movement (REM)). By collecting the user's heart rate data, the system can detect the user's physiological state and preliminarily determine the depth of sleep or whether it is in the rapid eye movement period. Sleep quality is closely related to breathing rate. The deep sleep stage is usually accompanied by slower and more stable breathing, while the breathing rate in the light sleep stage and the rapid eye movement period is relatively unstable. By collecting body movement data through the acceleration sensor, the user's activities during sleep can be identified. Frequent movement or turning over at night usually indicates that the user's sleep quality is poor or in a light sleep period. Using movement data, the system captures small movements at night, which can not only further determine the sleep stage, but also help identify behaviors that interfere with the user's sleep. According to the heart rate, breathing rate and body movement data, the system determines the current sleep state (such as light sleep, deep sleep, rapid eye movement period, etc.) through data fusion and algorithm, providing a real-time basis for subsequent ambient sound control, so that the system can adjust the audio type, frequency and volume in time to better adapt to the user's current sleep stage.

[0017] Preferably, the user's body movement data collection specifically includes:

[0018] Acceleration amplitude data is collected by using an acceleration sensor to obtain acceleration amplitude data;

[0019] Perform frequency analysis based on acceleration amplitude data to obtain acceleration frequency data;

[0020] Perform sleep movement event detection according to the acceleration frequency data and the acceleration amplitude data to obtain the user's sleep movement event data;

[0021] The motion data is analyzed based on the user's sleep motion event data to obtain the user's body motion data.

[0022] The real-time amplitude acquisition of the acceleration sensor in the present invention can capture the user's tiny body movements during sleep. The acceleration amplitude reflects the displacement or amplitude of the user's body, and provides basic data for judging the intensity and frequency of movement during sleep. By performing frequency analysis on the acceleration amplitude data, the frequency characteristics of the user's activities during sleep are revealed. Different sleep stages correspond to body movements of different frequencies (such as turning over in light sleep and micro-movements in deep sleep). Combined with the acceleration amplitude and frequency data, the system can detect the user's motion events (such as turning over, slight shaking, etc.) in real time, and identify the type and intensity of the motion events. Using threshold judgment and pattern matching algorithms, the system can identify key motion events and determine the specific time when they occur. Sleep motion event detection can identify unstable sleep patterns at night and provide accurate event markers for further determining the user's sleep quality. Through in-depth analysis of the user's sleep motion event data, the system can comprehensively evaluate the frequency, duration and intensity of nighttime movements, and generate overall body motion data, which can reflect the user's nighttime activity characteristics and help judge the continuity and stability of the sleep cycle.

[0023] Preferably, step S2 specifically includes:

[0024] Step S21: extracting sleep state change characteristics according to the sleep state data to obtain sleep state change characteristic data;

[0025] Step S22: extracting sleep state quality features according to the sleep state data to obtain sleep state quality feature data;

[0026] Step S23: performing sleep state characteristic fusion on the sleep state change characteristic data and the sleep state quality characteristic data to obtain sleep state characteristic data.

[0027] In the present invention, by extracting the change features of sleep state data, the system can identify the user's sleep stage switching in different time periods (such as from light sleep to deep sleep, from deep sleep to rapid eye movement period). Change feature extraction can help identify the user's typical sleep cycle and pattern, and the generated data can reflect the user's sleep structure and its dynamic changes throughout the night. Quality feature extraction is based on a variety of physiological data such as heart rate, respiratory rate and movement data, and can accurately evaluate the quality of each sleep stage, such as depth, stability and continuity. This feature analysis can help identify the length of the user's deep sleep time, the stability of the sleep cycle, and the interference factors that affect sleep quality (such as waking up at night). Through the quality feature data, the system can determine whether the user's sleep quality has reached the ideal state, and adjust the ambient sound settings accordingly to increase the duration of deep sleep and optimize the overall sleep quality. The change feature data is fused with the quality feature data to generate comprehensive sleep state feature data. Feature fusion can comprehensively consider the dynamic changes and quality characteristics of the user's sleep, and the generated feature data can reflect the user's overall sleep state.

[0028] Preferably, the sleep state change feature extraction is specifically:

[0029] Segment the sleep cycle according to the sleep state data to obtain sleep cycle segmentation data;

[0030] Perform sleep change recognition according to the sleep cycle segmentation data to obtain sleep change recognition data;

[0031] Perform feature dimensionality reduction according to the sleep change recognition data to obtain sleep change recognition dimensionality reduction data;

[0032] Perform frequency domain analysis on the dimensionality reduction data based on sleep change identification to obtain sleep change frequency feature data;

[0033] Perform sequence pattern mining based on the sleep change frequency feature data to obtain sleep change sequence pattern data;

[0034] The change index is calculated based on the sleep change sequence pattern data to obtain the sleep state change characteristic data.

[0035] In the present invention, the sleep state data is periodically segmented, which can more clearly define the user's sleep stage. The data is divided by the time series analysis method, which helps to identify the user's typical sleep cycle, including light sleep, deep sleep and rapid eye movement (REM). According to the segmented sleep cycle data, change identification can be performed to detect the state switching events in each cycle (such as switching from light sleep to deep sleep), which can effectively capture the transition moment of the sleep state, thereby revealing the user's sleep dynamic characteristics. Through the sleep change identification data, the system can better understand the user's sleep stability and state fluctuations, which is convenient for targeted adjustment of the ambient sound settings. By performing feature dimensionality reduction on the sleep change identification data, the system can reduce the data dimension while maintaining the key features of the data. Through frequency domain analysis, the system can identify the frequency characteristics of different sleep stages. For example, the deep sleep stage is accompanied by a lower movement frequency, while the light sleep and REM stages show a higher activity frequency. Through sequence pattern mining, the system can identify common patterns and laws of changes in the user's sleep state. Based on the sequence pattern data, the system can calculate the user's sleep state change index, which quantifies the user's sleep stability, stage transition frequency, and sleep pattern fluctuations, providing a specific reference indicator for the dynamic control of ambient sound. Through the change index calculation, the system can more accurately understand the user's sleep stability and adjust the ambient sound parameters when the user's sleep state changes significantly to help the user maintain a stable deep sleep.

[0036] Preferably, the sleep state quality feature extraction is specifically:

[0037] Extract sleep cycle features according to the sleep state data to obtain sleep state cycle feature data;

[0038] Performing sleep stability assessment based on sleep state cycle characteristic data to obtain sleep state stability assessment data;

[0039] Perform wakefulness detection based on sleep state stability assessment data to obtain sleep wakefulness event data;

[0040] Performing sleep stage continuity analysis based on sleep state stability assessment data to obtain sleep stage continuity feature data;

[0041] Calculate the sleep depth fluctuation frequency according to the sleep state cycle characteristic data to obtain sleep fluctuation frequency characteristic data;

[0042] The sleep quality index is calculated based on the sleep state stability assessment data, sleep wake event data, sleep stage continuity characteristic data and sleep fluctuation frequency characteristic data to obtain the sleep state quality characteristic data.

[0043] In the present invention, by extracting the periodic features of sleep state data, the system can identify the user's typical sleep cycle (such as the periodic distribution of light sleep, deep sleep and REM period). The sleep cycle features can reveal the user's overall sleep structure, help understand the duration and distribution of different sleep stages, and more accurately reflect the user's sleep pattern and rhythm, providing a data basis for subsequent quality evaluation and dynamic audio adjustment.

[0044] Based on the cycle characteristic data, the system can evaluate the stability of each sleep stage, including the stability of the deep sleep stage and the continuity of the light sleep stage. The stability characteristics of the user in each stage are evaluated through statistical methods such as standard deviation and coefficient of variation. Through the stability evaluation data, the system is able to detect brief awakening events at night. Using signal fluctuations and anomaly detection algorithms, the frequency and duration of awakening can be identified. Awake event detection can help the system analyze the user's sleep interruptions at night, and when frequent awakenings are detected, appropriate audio adjustments can be made to minimize interference with sleep.

[0045] Based on the stability assessment data, the system can analyze the continuous time of each sleep stage to quantify the continuity characteristics of the user in different stages. Through sliding window and continuous segment analysis, the system can understand the continuous time distribution of each sleep stage. Continuity analysis can help determine whether the user's deep sleep and REM stages are long enough, and provide support for optimizing audio adjustments during the deep sleep period. Through periodic feature data, the system can identify the user's deep fluctuation frequency in each sleep stage. For example, the deep sleep stage shows a lower fluctuation frequency, while the REM stage has a higher fluctuation frequency. Based on data such as sleep stability, wakefulness events, stage continuity and fluctuation frequency, the system can calculate an overall sleep quality index to quantify the user's overall sleep quality and help the system determine whether the current ambient sound meets the user's sleep needs.

[0046] Preferably, the sleep state feature fusion is specifically:

[0047] Generate a feature fusion matrix based on the sleep state change feature data and the sleep state quality feature data to obtain sleep feature fusion matrix data;

[0048] Performing feature weighting on the sleep state change feature data and the sleep state quality feature data according to the sleep feature fusion matrix data to obtain sleep state change feature weighted data and sleep state quality feature weighted data;

[0049] Performing deep feature weighting according to the weighted data of sleep state change characteristics and the weighted data of sleep state quality characteristics to obtain preliminary sleep state characteristic data;

[0050] Feature rebalancing is performed based on the preliminary sleep state feature data to obtain sleep state feature data.

[0051] In the present invention, by constructing a fusion matrix, the system can integrate the sleep state change feature data and the quality feature data to generate a preliminary feature matrix. The fusion matrix helps to associate multiple feature dimensions together to reflect the composite characteristics of the user's overall sleep state. The sleep state change feature data and the quality feature data are weighted according to the fusion matrix data, and the system can quantitatively evaluate the importance of different features. Feature weighting can highlight key features (such as deep sleep stability, frequency of wake events, etc.) and weaken unimportant features, thereby improving the assessment accuracy of the sleep state. Deep weighting of the weighted change feature data and quality feature data can further extract more representative sleep state features, and deep processing of the feature data through deep neural networks or other machine learning algorithms to generate preliminary sleep state feature data. Feature rebalancing is performed based on the preliminary sleep state feature data to ensure the stability and rationality of the features. The feature data is rebalanced by regularization methods or other statistical analysis methods to eliminate data imbalance problems caused by excessive weighting or deviation. The rebalanced data is more in line with the user's actual sleep condition, so that the system's judgment and response are more in line with the user's needs.

[0052] Preferably, step S3 is specifically:

[0053] Step S31: mapping according to a preset ambient sound type library and sleep state characteristic data to obtain primary ambient sound type data;

[0054] Step S32: Acquire environmental noise data;

[0055] Step S33: performing simulated playback according to the primary ambient sound type data and the ambient noise data to obtain ambient sound playback effect simulation data;

[0056] Step S34: extracting playback effect features according to the ambient sound playback effect simulation data to obtain playback effect feature data;

[0057] Step S35: adjusting the sleep state characteristic data according to the playback effect characteristic data to obtain sleep state characteristic adjustment data;

[0058] Step S36: mapping according to the preset ambient sound type library and the sleep state characteristic adjustment data to obtain secondary ambient sound type data;

[0059] Step S37: Perform audio fusion according to the primary ambient sound type data and the secondary ambient sound type data to obtain the ambient sound type data.

[0060] In the present invention, mapping is performed based on the preset ambient sound type library and the sleep state characteristic data, and the system can quickly select the preliminary audio type (such as white noise, natural sound, etc.) suitable for the current sleep stage. The primary ambient sound type data can provide audio settings that meet the current sleep state requirements and lay the foundation for subsequent ambient sound adjustment and optimization. By collecting the current ambient noise data, the system can understand the acoustic characteristics of the surrounding environment, such as the intensity and frequency distribution of external noise.

[0061] The acquired ambient noise data can help the system determine whether it is necessary to add noise cancellation or masking features to the played audio, thereby avoiding external noise from interfering with the user's sleep and optimizing the playback effect of ambient sound. By simulating the playback of primary ambient sound types and ambient noise data, the system can estimate the actual effect of ambient sound and analyze whether it can meet the user's sleep needs. Simulated playback effect The simulated data can reflect the playback effect of primary ambient sound types in the current noise environment, providing a data basis for audio optimization and fine-tuning, making the ambient sound settings more in line with actual needs.

[0062] By extracting the features of the simulated playback effects, the system can quantify the playback effects of the current ambient sound, such as the clarity of the sound, coverage, and masking effect with ambient noise. The playback effect feature data can reflect the advantages and disadvantages of the primary ambient sound type, so that the system can make targeted adjustments in subsequent audio selections to ensure the effectiveness and adaptability of the played audio. Based on the playback effect feature data, the sleep state feature data is adjusted to ensure that the choice of ambient sound is more in line with the user's current sleep state. Through feature adjustment, the system can identify whether the user's needs under the current ambient sound have been met, and optimize the sleep state feature data accordingly, making the ambient sound type more targeted and adaptable.

[0063] By re-mapping the ambient sound type library based on the feature-adjusted data, the system can generate secondary ambient sound type data. This remapping ensures that the audio type is highly matched with the user's actual sleep needs, and can more effectively assist the user in obtaining a suitable audio environment in different sleep stages. Through secondary mapping, the system can optimize the selection of primary ambient sound types to further enhance the audio playback effect. By fusing the primary and secondary ambient sound type data, the system can generate ambient sound type data, combining the preliminary selection and optimized audio types to ensure that the ambient sound settings are consistent with the user's sleep state characteristics and can adapt to the current noise environment. The fused audio type data ensures the best playback effect and improves the system's ability to adapt to the environment.

[0064] Preferably, step S4 is specifically:

[0065] Step S41: adjusting the ambient sound frequency according to the ambient sound type data and the sleep state characteristic data to obtain ambient sound frequency adjustment data;

[0066] Step S42: Acquire real-time sleep status data;

[0067] Step S43: updating and iterating the sleep state characteristic data according to the real-time sleep state data to obtain the real-time sleep state characteristic data;

[0068] Step S44: adjusting the ambient sound volume according to the ambient sound type data and the real-time sleep state characteristic data to obtain ambient sound volume adjustment data;

[0069] Step S45: performing a sleep-improving ambient sound control auxiliary operation according to the ambient sound type data, the ambient sound frequency adjustment data, and the ambient sound volume adjustment data.

[0070] In the present invention, according to the ambient sound type data and the sleep state characteristic data, the system can adjust the frequency of the audio for different sleep stages. For example, in the deep sleep stage, the system can reduce the frequency to make the sound softer, and in the light sleep stage, the appropriate frequency can be increased to provide mild stimulation to promote the user to enter a deeper sleep. Frequency adjustment ensures that the ambient sound can better adapt to the user's current sleep stage characteristics, thereby optimizing the sleep assistance effect. Through the update and iteration of real-time data, the system can optimize and improve the original sleep state characteristic data. According to the ambient sound type data and the updated real-time sleep state characteristic data, the system can dynamically adjust the volume of the ambient sound. The volume adjustment can enable the system to gradually reduce the volume when the user enters the deep sleep stage to reduce interference, and gradually increase the volume in the light sleep or awake stage to help the user enter a deeper sleep. The present invention can dynamically adapt to the user's sleep state, help the user enter a deep sleep or maintain the current ideal sleep state, thereby significantly improving the overall sleep quality.

[0071] Preferably, a sleep-improving ambient sound control assistance system is used to execute the sleep-improving ambient sound control assistance method as described above, and the sleep-improving ambient sound control assistance system includes:

[0072] A sleep state acquisition module, used to acquire sleep state data;

[0073] A sleep state feature extraction module, used to extract sleep state features according to the sleep state data to obtain sleep state feature data;

[0074] An ambient sound type selection module is used to select an ambient sound type according to the sleep state characteristic data to obtain ambient sound type data;

[0075] The ambient sound frequency adjustment and ambient sound volume adjustment modules are used to adjust the ambient sound frequency and ambient sound volume according to the ambient sound type data and the sleep state characteristic data, and obtain the ambient sound frequency adjustment data and the ambient sound volume adjustment data respectively, so as to perform auxiliary operations of sleep improvement ambient sound control.

[0076] The beneficial effects of the present invention are: the system can obtain highly accurate sleep state data through multimodal sensor collection of heart rate, respiratory rate and body movement data, can reflect the changes in the user's physiological state in real time, and enable the system to make dynamic analysis based on actual physiological characteristics. By extracting the features of the sleep state data, the system can identify the characteristics of the user's sleep state change and quality characteristics. This feature extraction method based on data analysis enables the system to accurately distinguish the user's light sleep, deep sleep and rapid eye movement (REM) period, and can extract representative features (such as depth, stability, wakefulness events, etc.) from them. The system uses a preset ambient sound type library and sleep state feature data for mapping, so that the system can quickly select a preliminary suitable ambient sound type. At the same time, by combining environmental noise data and simulated playback effects, the system can determine whether the primary ambient sound can effectively help the user enter or maintain an ideal sleep state in the actual environment. Through the update of real-time sleep data and the dynamic adjustment of ambient sound, the system can continuously adjust the audio frequency and volume during sleep. The adjustment method based on real-time data enables the system to quickly respond to changes in the user's sleep state and ensure that the ambient sound settings at each stage can closely fit the current sleep needs. Frequency adjustment can reduce high-frequency interference in the deep sleep stage and improve the stability of deep sleep; while in the light sleep or awake stage, volume adjustment can enhance audio support and help users return to deep sleep as soon as possible. The system's real-time data iteration update further ensures the accuracy of the ambient sound settings, making audio control more responsive and flexible, and avoiding sleep interruptions or quality degradation caused by static audio settings. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting implementations made with reference to the following drawings:

[0078] Figure 1 A flowchart showing a method for assisting in controlling ambient sound to improve sleep according to an embodiment of the present invention is provided;

[0079] Figure 2 A flowchart showing a method for obtaining a sleep state according to an embodiment of the present invention is shown;

[0080] Figure 3 A flowchart showing a method for extracting sleep state features according to an embodiment is shown;

[0081] Figure 4A flowchart showing a method for selecting an ambient sound type according to an embodiment is shown;

[0082] Figure 5 A flowchart of a method for adjusting ambient sound frequency and ambient sound volume according to an embodiment is shown. DETAILED DESCRIPTION

[0083] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0084] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0085] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0086] User A exhibits the following sleep cycle pattern after falling asleep every night: Initial stage of falling asleep (light sleep): lasts about 20 minutes, accompanied by slight body movement, heart rate is about 70-75BPM (times / minute), and breathing rate is 18 times / minute. Deep sleep stage: lasts about 60 minutes, accompanied by very little body movement, heart rate drops to 50-55BPM, and breathing rate is about 12 times / minute. Rapid eye movement (REM): lasts about 10 minutes, heart rate is about 65BPM, breathing rate increases slightly to 15 times / minute, and is accompanied by slight body movement. Light sleep transition period: lasts about 10 minutes, body movement increases, heart rate returns to 70BPM, and breathing rate is 17 times / minute. Occasional awakening at night: awakening an average of 2 times per night, each time awakening for about 3 minutes.

[0087] The system collects user A's physiological data in real time through heart rate sensors, respiratory sensors, and acceleration sensors. The data collection frequency is once per second, and the collected data includes heart rate, respiratory rate, and three-axis acceleration. Heart rate data: 70, 69, 68, 66, ..., 55 BPM, respiratory rate data: 18, 17, 16, 15, ..., 12 times / minute, body movement data: x: 0.01, 0.02, ..., 0.005; y: 0.02, 0.03, ..., 0.01; z: 0.00, 0.01, ..., 0.002. Based on the downward trend of heart rate and respiratory rate and the reduction of movement data, the system recognizes that the user has entered a deep sleep state. It is detected that the user's sleep state gradually enters the deep sleep stage from the light sleep stage.

[0088] Based on the collected data, the system segments the time periods of different sleep stages. The user's sleep cycle is segmented as follows: light sleep (20 minutes), deep sleep (60 minutes), REM period (10 minutes), and light sleep transition period (10 minutes). The heart rate, breathing rate, and movement amplitude data of each stage are extracted, and the characteristic values, such as the mean and standard deviation, are calculated. For example, the statistical results of the heart rate data in the deep sleep stage: the average heart rate is 52BPM, and the standard deviation is 2BPM. Features such as the duration and stability of the deep sleep period are extracted: the duration is 60 minutes, and the movement amplitude is low. Awakening events: awakened twice at night, at 02:00AM and 04:00AM respectively. The system fuses the sleep change features and quality features to generate sleep state feature data. Feature data: {stage: deep sleep, heart rate: 52BPM, breathing rate: 12 times / minute, awakening events: 2 times}.

[0089] Combined with the preset audio type library, the system selects white noise suitable for the deep sleep stage, with a preliminary audio frequency of 60Hz. Audio type: low-frequency white noise, preset volume: 30dB. The external environmental noise detected was about 45dB, including slight wind noise. Noise data: frequency 35Hz, amplitude 45dB. The system simulates the playback effect of white noise and analyzes the noise masking effect. The results show that the primary white noise does not completely mask the external noise in the current noise environment. The simulation data analysis recommends increasing the frequency to 70Hz.

[0090] The clarity and masking characteristics of the playback effect are extracted, showing that the primary white noise does not mask the external noise sufficiently. Based on the adjusted frequency and ambient sound type library, the secondary white noise type is mapped again, and the audio frequency is amplified to 70Hz. The system combines the characteristics of primary and secondary white noise, and the audio type is determined to be white noise, with a frequency of 70Hz and a volume of 35dB.

[0091] The ambient sound frequency increases from 60Hz to 70Hz, and the system continues to adjust the frequency based on real-time feature data to ensure the best effect. Actual playback frequency setting: 70Hz. Real-time monitoring of user heart rate, breathing rate and movement data: The real-time data at this stage is {heart rate: 52BPM, breathing rate: 12 times / minute, movement amplitude: low}. Sleep state feature data is updated to: deep sleep stage. The system detects that the user's deep sleep state is stable, so the volume is reduced from 35dB to 30dB to reduce potential interference. Actual playback volume setting: 30dB. Based on the determined audio frequency (70Hz) and volume (30dB), the system plays adjusted white noise to optimize the deep sleep environment.

[0092] Comprehensive controls ensure audio settings are in sync with the user's sleep state in real time.

[0093] See also Figures 1 to 5 , the present application provides a sleep improvement ambient sound control auxiliary method, comprising the following steps:

[0094] Step S1: Obtaining sleep state data;

[0095] In one embodiment, sensors or wearable devices (such as smart bracelets, sleep monitors, etc.) are used to monitor the user's physiological signals in real time, such as heart rate, breathing rate, body activity, etc. These physiological signals are converted into digital signals through the data acquisition module and transmitted to the data processing unit.

[0096] Step S2: extracting sleep state features according to the sleep state data to obtain sleep state feature data;

[0097] In one embodiment, based on the acquired sleep state data, the user's current sleep stage (such as falling asleep, light sleep, deep sleep, and rapid eye movement) is analyzed to identify key features associated with different sleep stages, such as changes in heart rate, breathing stability, frequency of body movements, etc.

[0098] Step S3: Selecting an ambient sound type according to the sleep state characteristic data to obtain ambient sound type data;

[0099] In one embodiment, the most suitable ambient sound type for the current sleep stage is determined based on the user's sleep stage characteristic data. For example, soft natural sounds, such as rain or waves, can be selected in the falling asleep stage to help relax. In the light sleep stage, relatively stable sounds, such as wind, can be used to maintain a sleeping state. In the deep sleep stage, lower-frequency sounds, such as bass resonance, can be selected to further promote deep sleep.

[0100] Step S4: adjusting the ambient sound frequency and the ambient sound volume according to the ambient sound type data and the sleep state characteristic data, and obtaining ambient sound frequency adjustment data and ambient sound volume adjustment data, respectively, to perform a sleep-improving ambient sound control auxiliary operation.

[0101] In one embodiment, the frequency of the ambient sound is fine-tuned based on the characteristic data of the current sleep stage. In the falling asleep stage, a higher frequency adjustment can be used to make the ambient sound more active but not harsh, so as to relax and fall asleep quickly. In the light sleep stage, the frequency is adjusted to a medium level to ensure that the ambient sound is soft and balanced, so as to maintain light sleep. In the deep sleep stage, the frequency is adjusted to a low frequency band to enhance the deep sleep effect. According to the characteristics of the sleep stage and the type of ambient sound, the volume is adjusted to ensure the best effect. In the falling asleep stage, the volume is slightly lowered to avoid stimulating the nervous system. In the light sleep stage, the volume is maintained at a steady level to avoid fluctuations to prevent waking the user. In the deep sleep stage, the volume is slightly reduced to an extremely low level to ensure that the sound exists only in the background. The adjusted frequency and volume data are converted into control signals to control the playback device to adjust the frequency and volume of the ambient sound in real time to achieve the best sleep improvement effect.

[0102] In one embodiment, based on the characteristics of the light sleep stage, the frequency of wind sound needs to be maintained at a medium level. For example: the frequency range is set to 400Hz to 800Hz, and the frequency adjustment coefficient used in light sleep is 0.75. Therefore, the adjusted frequency = upper limit of the frequency range × adjustment coefficient = 800 × 0.75 = 600Hz. The ambient sound frequency adjustment data is 600Hz. The volume is set according to the user's light sleep state to avoid interference caused by excessive volume. The volume range is set to 20 to 50 decibels, and the volume adjustment coefficient in light sleep is 0.5. Therefore, the adjusted volume = upper limit of the volume range × adjustment coefficient = 50 × 0.5 = 25 decibels. The ambient sound volume adjustment data is 25 decibels.

[0103] Preferably, step S1 specifically includes:

[0104] Step S11: collecting the user's heart rate data through a heart rate sensor to obtain the user's heart rate data;

[0105] In one embodiment, an optical heart rate sensor (located on a smart bracelet or wearable device) is used to monitor pulse waves in the blood. The sensor detects tiny changes in each heartbeat through the skin. The sensor emits a light beam and detects fluctuations in the intensity of the reflected light, which reflect changes in heart rate. Each time a peak is detected, it represents a heartbeat. The number of peaks per minute is calculated to obtain the heart rate per minute (unit: bpm). After obtaining the preliminary heart rate data, an average calculation (or sliding window average) is performed to smooth the data to obtain more stable heart rate data.

[0106] Step S12: collecting the user's respiratory frequency data through a respiratory sensor to obtain the user's respiratory frequency data;

[0107] In one embodiment, the respiratory rate sensor can be a chest strap or a sensor using infrared detection (located in a professional monitoring device). The accelerometer in a smart bracelet can also be used in combination with heart rate changes to indirectly monitor the respiratory rate. The sensor monitors the expansion and contraction of the chest, or captures the respiratory cycle by detecting slight temperature changes in the skin. Each expansion and contraction (or increase and decrease in temperature) represents a complete breath. The complete respiratory cycles per minute (the rise and fall of breathing are each a complete cycle) are counted to obtain the respiratory rate (unit: breaths / minute). The data is further smoothed, and sudden fluctuations in the respiratory signal are removed by techniques such as moving average to obtain more stable respiratory rate data.

[0108] Step S13: collecting user body motion data through an acceleration sensor to obtain user body motion data;

[0109] In one embodiment, an acceleration sensor (usually built into a smart bracelet or wearable device) is used to monitor tiny movements of the body. The sensor can measure changes in acceleration in three axes (x, y, and z axes). The sensor detects changes in acceleration values ​​in real time and records tiny movements of the user in all directions, such as turning over, moving the arms, and so on. The collected three-axis acceleration signals are processed to calculate the intensity of the movement. The modulus of the acceleration (i.e., the total acceleration) is used to reflect the intensity changes of the movement. A smoothing filter is used to eliminate subtle signal fluctuations and remove small random movements. A threshold is determined for the smoothed motion signal. If the total acceleration exceeds a certain value, it is marked as a large movement, otherwise it is stationary or a small movement, thereby obtaining the user's motion state data.

[0110] Specifically, the sensor records the acceleration value in each direction in real time to reflect the user's motion intensity in that direction. For example, the following acceleration data is detected within 5 seconds: x-axis: 0.01g, 0.02g, 0.03g, 0.01g, 0.02g, y-axis: 0.00g, 0.01g, 0.01g, 0.00g, 0.01g, z-axis: 0.02g, 0.02g, 0.03g, 0.01g, 0.02g. Calculate the sum of the three-axis acceleration (i.e., motion intensity), and the average acceleration within 5 seconds is 0.015g, reflecting that the user's overall motion intensity is small. Apply a smoothing filter to the data to remove high-frequency noise, and set a motion threshold to distinguish between small movements and no movement. User body motion data = 0.015g (average motion intensity, indicating a small amount of movement).

[0111] Step S14: determining the user's sleep state based on the user's heart rate data, the user's breathing rate data, and the user's body movement data to obtain sleep state data.

[0112] In one embodiment, the user's sleep state is determined based on heart rate, breathing rate and body movement data. Generally, the sleep state can be divided into awake, light sleep, deep sleep and rapid eye movement (REM). According to the heart rate, breathing rate and body movement data per minute, the threshold range is set respectively. For example, in the light sleep state, the heart rate is low, the breathing is stable, and the body movement is less; while in the deep sleep state, the heart rate and breathing rate further decrease, and there is almost no body movement. In the determination process, the change rate of these parameters can be calculated. For example: heart rate change rate: if the heart rate decreases greatly and remains stable, it is in a deep sleep state; if the heart rate fluctuates greatly, it is in a light sleep or REM stage. Breathing change rate: if the breathing frequency is low and stable, it indicates that the user is in a deep sleep stage. Exercise intensity: if the exercise frequency is high, it indicates that the user is awake or in the early stage of falling asleep; if the exercise frequency is reduced, it indicates that the user has entered a light sleep or deep sleep state. The results of the determination are organized into the user's sleep state data, and the states of different stages are marked in the sleep data. For example: heart rate: 59 bpm, breathing rate: 15 times / minute, body movement: 0.2 times / minute (very little body movement). Based on these features, the preset expert knowledge engine or sleep knowledge graph is used to determine that the user is currently in a light sleep state, and the result is stored as the user's current sleep state data.

[0113] The expert knowledge engine is built based on the experience and knowledge accumulated by domain experts, and uses a set of rules and logical reasoning processes to analyze and judge the user's physiological data. Define clear threshold ranges and judgment rules for different sleep stages based on physiological parameters (heart rate, respiratory rate, body movement, etc.). The thresholds are determined through a large number of clinical experiments and data statistics. Awake state: heart rate: greater than 70bpm, respiratory rate: greater than 18 times / minute, body movement: more frequent (>1 time / minute). Light sleep state: heart rate: 60-70bpm, respiratory rate: 15-18 times / minute, body movement: less (0.1-1 times / minute). Deep sleep state: heart rate: less than 60bpm, respiratory rate: less than 15 times / minute, body movement: almost no (<0.1 times / minute). REM period: heart rate: large fluctuations, unstable, respiratory rate: irregular, sometimes short pauses, body movement: occasional eye movement, no obvious muscle activity. These threshold rules are constructed into an operational rule library for system calls. Each rule in the rule base is in the form of "if-then": Rule 1: If the heart rate > 70bpm and the body movement > 1 time / minute, it is judged to be awake. Rule 2: If the heart rate is between 60-70bpm and the body movement is between 0.1-1 times / minute, it is judged to be a light sleep state. Rule 3: If the heart rate is <60bpm and the breathing rate is <15 times / minute and the body movement is <0.1 times / minute, it is judged to be a deep sleep state. The system analyzes the physiological data at each moment, matches it with the conditions in the rule base, and gives the corresponding sleep state. If multiple rules match, set the priority. For example, the deep sleep state takes precedence over the judgment of the light sleep and REM states.

[0114] In one embodiment, the user's current sleep state is analyzed in combination with the user's heart rate, breathing rate, and body movement data. The sleep stage is determined based on specific rules, for example: if the heart rate is lower than 70bpm, the breathing rate is between 12-18 times / minute, and the body movement data is lower than 0.02g, then it is in light sleep. If the heart rate is lower than 60bpm, the breathing rate is lower than 12 times / minute, and there is almost no body movement (movement data is lower than 0.01g), then it is in deep sleep.

[0115] Preferably, the user's body movement data collection specifically includes:

[0116] Acceleration amplitude data is collected by using an acceleration sensor to obtain acceleration amplitude data;

[0117] In one embodiment, an acceleration sensor built into a smart bracelet or wearable device is used. The sensor can record acceleration values ​​in three directions: x, y, and z. The acceleration sensor samples multiple times per second and records acceleration values ​​in each direction. Sampling is performed 10 times in 1 second, and the following data are obtained: x-axis: 0.01g, 0.02g, 0.01g, 0.00g, 0.02g, 0.01g, 0.02g, 0.01g, 0.00g, 0.01g; y-axis: 0.02g, 0.03g, 0.02g, 0.01g, 0.03g, 0.02g, 0.03g, 0.02g, 0.01g, 0.02g; z-axis: 0.00g, 0.01g, 0.00g, 0.02g, 0.01g, 0.01g, 0.00g, 0.02g, 0.01g, 0.01g. Calculate the acceleration amplitude (total acceleration) at each moment. a is the acceleration amplitude data, x is the acceleration value collected on the x-axis, y is the acceleration value collected on the y-axis, and z is the acceleration value collected on the z-axis. The acceleration amplitude data corresponding to the ten samples are summed and averaged to obtain the total acceleration, such as an average amplitude of 0.015g.

[0118] Perform frequency analysis based on acceleration amplitude data to obtain acceleration frequency data;

[0119] In one embodiment, the frequencies of different movements in the acceleration data are analyzed to distinguish the types of movements. During sleep, low-frequency movements usually represent turning over or adjusting posture. The collected acceleration amplitude data is subjected to time series analysis to determine the frequency components in the motion signal. The time domain data is converted into frequency domain data using Fourier transform to identify movements of different frequencies. The frequency components obtained are as follows: Main frequency component: 0.2 Hz (low frequency), and movement components in a specific frequency range are filtered out from the frequency domain data, such as low-frequency movement features of 0.1 to 0.5 Hz, which indicate the movement frequency when the user turns over or slightly adjusts posture during sleep. The obtained acceleration frequency data = 0.2 Hz.

[0120] Perform sleep movement event detection according to the acceleration frequency data and the acceleration amplitude data to obtain the user's sleep movement event data;

[0121] In one embodiment, the presence of specific motion events (such as turning over, arm movement, etc.) is detected by analyzing the acceleration amplitude and frequency data, thereby identifying the user's sleep activity. Set the motion event threshold: Set the threshold based on the total acceleration amplitude and frequency range. If the acceleration amplitude exceeds 0.02g and the frequency is between 0.1Hz and 0.5Hz, it is judged as a turning over or posture adjustment event. Each signal that exceeds the threshold is marked as a motion event. Two qualified motion events are detected within 10 seconds. The time and characteristics of the motion event are recorded and stored as event data. For example, two events are detected within 10 seconds, and the duration of each event is 1 second. Output user sleep motion event data = 2 motion events.

[0122] The motion data is analyzed based on the user's sleep motion event data to obtain the user's body motion data.

[0123] In one embodiment, the overall movement situation is analyzed based on the motion event data to further determine the sleep state. Based on the collected motion event data, the frequency and intensity of the motion events are calculated. For example, it is detected that the frequency of motion events per minute is 2 times, which is a low-frequency motion. The relationship between motion events and sleep stages is further analyzed. For example, more frequent small movements indicate shallow sleep or turning over, and basically no movement indicates a deep sleep state. The analysis results are quantified into overall motion data, such as average motion intensity, event frequency, etc. Output motion intensity and motion frequency, such as motion intensity: 0.015g (average acceleration amplitude), motion frequency: 2 times / minute (motion event frequency), user body motion data = low-frequency motion (motion intensity is 0.015g, frequency is 2 times / minute).

[0124] Preferably, step S2 specifically includes:

[0125] Step S21: extracting sleep state change characteristics according to the sleep state data to obtain sleep state change characteristic data;

[0126] In one embodiment, the change rate of heart rate, respiratory rate and body movement is extracted from the sleep state data. For example, the change rate of heart rate per minute and the fluctuation amplitude of respiratory rate are calculated to evaluate the physiological response of the body. The mean and standard deviation within a specific time are calculated using the sliding window method to capture changes in the short term. For example, the heart rate change data is: 1st minute: 62bpm, 2nd minute: 64bpm, 3rd minute: 63bpm, 4th minute: 65bpm. Calculate the change rate, and calculate the change of heart rate per minute by difference, such as (64-62) / 1 minute = 2bpm / minute change rate. Integrate these change rates and fluctuation amplitudes into sleep state change characteristics. For example, the comprehensive change trend of heart rate change rate, respiratory change rate and movement event frequency is used as feature data. Sleep state change feature data = heart rate change rate 2bpm / minute, respiratory change rate 1 time / minute, movement change frequency 0.2 times / minute.

[0127] Step S22: extracting sleep state quality features according to the sleep state data to obtain sleep state quality feature data;

[0128] In one embodiment, the stability and depth of the sleep state are evaluated to determine the sleep quality, including the characteristics of light sleep, deep sleep, and rapid eye movement (REM) stages. The duration of each stage is extracted from the sleep state data. For example, light sleep time: 40 minutes, deep sleep time: 90 minutes, REM time: 30 minutes. The ratio of sleep stages is calculated, and the ratio of deep sleep to light sleep is evaluated as a measure of sleep quality. The deep sleep state is judged based on the average value of heart rate and breathing and their fluctuations. For example, during deep sleep, the heart rate is 58bpm and stable, and the breathing rate is 13 times / minute and stable. These values ​​are integrated into sleep quality feature data, including the depth index and stability index of sleep. Sleep state quality feature data = deep sleep accounts for 45%, the heart rate is highly stable, and the breathing is stable.

[0129] Step S23: performing sleep state characteristic fusion on the sleep state change characteristic data and the sleep state quality characteristic data to obtain sleep state characteristic data.

[0130] In one embodiment, by fusing the change characteristics and quality characteristics of the sleep state, the overall sleep state characteristics are obtained to more comprehensively reflect the user's sleep condition. The change characteristic data and the quality characteristic data are weighted averaged or spliced ​​according to different weights. For example, in the sleep change characteristics, the heart rate change rate, the breathing change rate and the movement change frequency are used as dynamic features; in the sleep quality characteristics, the deep sleep ratio and the stability index are used as static features. These features are standardized and all features are scaled to the same scale (for example, between 0 and 1) to ensure the balance during data fusion. All feature data are merged into a sleep state feature vector for subsequent steps. The fused features are as follows: sleep characteristics: heart rate change rate 0.5, breathing change rate 0.3, movement change frequency 0.2, deep sleep ratio 0.45, stability index 0.8. Sleep state feature data = [heart rate change rate 0.5, breathing change rate 0.3, movement change frequency 0.2, deep sleep ratio 0.45, stability index 0.8]

[0131] Preferably, the sleep state change feature extraction is specifically:

[0132] Segment the sleep cycle according to the sleep state data to obtain sleep cycle segmentation data;

[0133] In one embodiment, the sleep state data is divided into multiple cycles to facilitate the analysis of the characteristics of each cycle. Based on a fixed time window or detected physiological signal changes, the whole night of sleep is divided into multiple stages or cycles. For example, with 90 minutes as a cycle, the whole night of sleep is divided into 4 cycles: Cycle 1: falling asleep and light sleep (the first 90 minutes), Cycle 2: deep sleep (90 to 180 minutes), Cycle 3: light sleep and REM (180 to 270 minutes), Cycle 4: deep sleep (270 to 360 minutes). The start and end time and stage characteristics of each cycle (such as the mean of heart rate and respiratory rate) are recorded to obtain cycle segmentation data. Output sleep cycle segmentation data = [Cycle 1: light sleep, Cycle 2: deep sleep, Cycle 3: light sleep and REM, Cycle 4: deep sleep].

[0134] Perform sleep change recognition according to the sleep cycle segmentation data to obtain sleep change recognition data;

[0135] In one embodiment, significant changes within each cycle are detected to identify transitions between different stages of sleep. Within each cycle, changes in sleep state are identified by analyzing fluctuations in heart rate, respiratory rate, and motion data. For example, in cycle 2 (deep sleep), heart rate and respiration are relatively stable; in cycle 3 (light sleep and REM), heart rate and respiration fluctuate greatly, which is the transition from light sleep to REM. Change points are marked, such as when the heart rate fluctuation exceeds 5bpm, it is recorded as a change. The identified change data is recorded as change identification data, marking significant change points within the sleep stage. Sleep change identification data = cycle 1: 1 change, cycle 2: 0 changes, cycle 3: 2 changes, cycle 4: 1 change.

[0136] Perform feature dimensionality reduction according to the sleep change recognition data to obtain sleep change recognition dimensionality reduction data;

[0137] In one embodiment, key features in the change identification data are extracted and redundant data is removed. Features such as the number of changes and the magnitude of changes are selected, and minor features are discarded. Information such as the number of changes and the magnitude of changes are converted into feature vectors. For example, the number of changes and the magnitude of changes are recorded as follows: number of changes: [1, 0, 2, 1], magnitude of changes: [4bpm, 0bpm, 6bpm, 3bpm]. The change identification data is converted into more concise dimensionality reduction data, retaining the number of changes and magnitude as the main features. Sleep change identification dimensionality reduction data = [number of changes 1, 0, 2, 1; magnitude of changes 4, 0, 6, 3bpm].

[0138] Perform frequency domain analysis on the dimensionality reduction data based on sleep change identification to obtain sleep change frequency feature data;

[0139] In one embodiment, the periodicity of the change characteristics is identified to understand the sleep change pattern at a specific frequency. The frequency domain transformation (such as fast Fourier transform) is applied to the data after dimensionality reduction to analyze the frequency distribution of the number and amplitude of changes. For example, the change frequency of cycles 1 and 3 is 0.1Hz, and that of cycle 4 is 0.05Hz. The distribution of high-frequency or low-frequency changes is identified to understand the stability of the changes. Assume that the number of changes is mainly concentrated in the low frequency of 0.05Hz. The frequency domain analysis results are recorded as frequency feature data to obtain the main change frequency of each cycle. Sleep change frequency feature data = main frequency 0.1Hz (cycles 1 and 3), low-frequency change 0.05Hz (cycle 4).

[0140] Perform sequence pattern mining based on the sleep change frequency feature data to obtain sleep change sequence pattern data;

[0141] In one embodiment, common patterns in periodic changes are discovered to detect recurring patterns of changes within sleep stages. Sequential pattern mining techniques are used to analyze patterns in frequency feature data, such as detecting a sequence of changes that occur when a high-frequency fluctuation is 0.1 Hz. Patterns are identified and marked. For example, the frequencies of change in cycles 1 and 3 are similar, representing the recurrence of light sleep and REM cycles. The pattern analysis results are organized into sequential pattern data to represent common patterns of sleep changes. For example, a change of 0.1 Hz between light sleep and REM is detected. Sleep change sequence pattern data = high-frequency fluctuation pattern 0.1 Hz (transition between light sleep and REM).

[0142] In the data sequence, some important events (such as rapid rise or fall of frequency, significant fluctuation, etc.) are marked as events. Events can represent changes in a specific frequency. In frequency feature data, patterns represent the frequency and order of occurrence of a certain frequency or event over a period of time. For example, "heart rate drops significantly when the high frequency fluctuates at 0.1Hz" is a pattern. The frequency of the pattern is evaluated by calculating the support and confidence. The support is the proportion of the number of times a pattern appears in the entire data sequence to the total sequence length, and the confidence is the probability that a certain preceding event leads to a subsequent event. Define the type of pattern to be mined, for example: frequency fluctuation pattern, such as identifying the change pattern of heart rate, breathing and other data when the fluctuation of a certain frequency (such as 0.1Hz) occurs frequently. Continuous event pattern, such as detecting the continuous occurrence of a certain type of event (such as high-frequency fluctuation) over a period of time. The pattern is expressed as: if the frequency is 0.1Hz at multiple time points and the heart rate drops continuously, then the pattern can be identified as a "heart rate drop pattern". Find frequently occurring frequency item sets from the event sequence, that is, the number of times certain frequency values ​​appear in the data exceeds the set support threshold. For example, if the support threshold is set to 30%, if the frequency fluctuation of 0.1Hz in a certain period of time exceeds 30% of the occurrence frequency, it is identified as a frequent item. Based on the frequent item sets, possible candidate sequence patterns are generated. For example, if the system processes the frequent item sets of 0.1Hz and heart rate drop, their combined sequences can be further mined. By verifying the support and confidence of each candidate pattern, meaningful patterns are screened out. Patterns with support below the threshold are cut off, and the remaining ones are high-support, high-confidence patterns. For example, if "0.1Hz frequency fluctuation" and "heart rate drop" often appear together and the confidence is high, the pattern is valid. Sort the patterns by support or confidence, and prioritize the analysis of high-confidence, high-support patterns. After identifying high-frequency patterns, the system marks the data according to these patterns. For example, the system marks "When the frequency fluctuates by 0.1Hz, the user's heart rate drops significantly" and records the pattern for analysis and intervention.

[0143] The change index is calculated based on the sleep change sequence pattern data to obtain the sleep state change characteristic data.

[0144] In one embodiment, the pattern data is converted into a variation index to quantify the dynamic fluctuation of the sleep state. According to the sequence pattern data, different weights are assigned to the high-frequency and low-frequency fluctuation patterns. For example, the high-frequency fluctuation 0.1Hz pattern is assigned a weight of 0.7, and the low-frequency fluctuation 0.05Hz pattern is assigned a weight of 0.3. The variation index is calculated as 0.085. The variation index represents the volatility and change trend of the sleep state. The sleep state change characteristic data = the variation index 0.085 (reflecting the overall change frequency and pattern).

[0145] Preferably, the sleep state quality feature extraction is specifically:

[0146] Extract sleep cycle features according to the sleep state data to obtain sleep state cycle feature data;

[0147] In one embodiment, the main features of the sleep cycle are identified, such as the distribution and duration of light sleep, deep sleep and REM. The sleep state data is segmented, and the mean and standard deviation of the heart rate, respiratory rate and body movement in each cycle are extracted to describe the overall characteristics of sleep. The stage ratio (light sleep, deep sleep, REM) of each cycle is calculated. The characteristic data of each cycle are as follows: Cycle 1: light sleep accounts for 70%, deep sleep accounts for 30%, and there is no REM. Cycle 2: light sleep accounts for 40%, deep sleep accounts for 50%, and REM accounts for 10%. Cycle 3: light sleep accounts for 50%, deep sleep accounts for 20%, and REM accounts for 30%. Cycle 4: light sleep accounts for 20%, deep sleep accounts for 70%, and there is no REM. The characteristic data of each cycle is sorted out and expressed as sleep cycle characteristic data. Sleep state cycle characteristic data = [Cycle 1: light sleep 70%, deep sleep 30%; Cycle 2: light sleep 40%, deep sleep 50%, REM 10%; Cycle 3: light sleep 50%, deep sleep 20%, REM 30%; Cycle 4: light sleep 20%, deep sleep 70%].

[0148] Performing sleep stability assessment based on sleep state cycle characteristic data to obtain sleep state stability assessment data;

[0149] In one embodiment, the stability within each sleep cycle is evaluated, including the fluctuation amplitude of heart rate and breathing and the continuity of sleep stages. The standard deviation of the heart rate, breathing and body movement data within each cycle is calculated to quantify the fluctuation. The higher the stability, the smaller the fluctuation amplitude. The stability evaluation results are: cycle 1: heart rate standard deviation 4bpm, breathing standard deviation 2 times / minute, cycle 2: heart rate standard deviation 3bpm, breathing standard deviation 1 time / minute, cycle 3: heart rate standard deviation 5bpm, breathing standard deviation 3 times / minute, cycle 4: heart rate standard deviation 2bpm, breathing standard deviation 1 time / minute. The stability score of each cycle is calculated based on the fluctuation amplitude and integrated into sleep stability evaluation data. Sleep state stability evaluation data = [Cycle 1: heart rate fluctuation 4bpm, breathing fluctuation 2 times / minute; cycle 2: heart rate fluctuation 3bpm, breathing fluctuation 1 time / minute; cycle 3: heart rate fluctuation 5bpm, breathing fluctuation 3 times / minute; cycle 4: heart rate fluctuation 2bpm, breathing fluctuation 1 time / minute].

[0150] Perform wakefulness detection based on sleep state stability assessment data to obtain sleep wakefulness event data;

[0151] In one embodiment, wakefulness events in a sleep cycle are identified to understand sleep interruptions. In the stability assessment data, abnormal increases in heart rate and body movement are detected to determine wakefulness events. For example, if the heart rate increases dramatically in a short period of time and is accompanied by an increase in body movement, it is marked as a wakefulness event. The wakefulness detection results are as follows: Cycle 1: no wakefulness events, Cycle 2: no wakefulness events, Cycle 3: 1 wakefulness event (sharp increase in heart rate and body movement), Cycle 4: no wakefulness events. The number of wakefulness events in each cycle is recorded and represented as sleep wakefulness event data. Sleep wakefulness event data = [Cycle 1: None, Cycle 2: None, Cycle 3: 1 time, Cycle 4: None].

[0152] Performing sleep stage continuity analysis based on sleep state stability assessment data to obtain sleep stage continuity feature data;

[0153] In one embodiment, the continuity of each sleep stage is evaluated to determine the stability of sleep. In the sleep cycle characteristic data, the continuous time periods of the light sleep, deep sleep and REM stages are analyzed. The higher the continuity, the more stable the sleep. The continuity analysis results are: cycle 1: light sleep for 40 minutes, deep sleep for 30 minutes, cycle 2: light sleep for 20 minutes, deep sleep for 45 minutes, REM for 10 minutes, cycle 3: light sleep for 30 minutes, deep sleep for 10 minutes, REM for 25 minutes, cycle 4: deep sleep for 60 minutes. The continuous duration of each sleep stage is summarized and expressed as a continuity score to obtain the sleep stage continuity characteristic data. Sleep stage continuity characteristic data = [cycle 1: light sleep for 40 minutes, deep sleep for 30 minutes; cycle 2: light sleep for 20 minutes, deep sleep for 45 minutes, REM for 10 minutes; cycle 3: light sleep for 30 minutes, deep sleep for 10 minutes, REM for 25 minutes; cycle 4: deep sleep for 60 minutes].

[0154] Each stage is assigned a continuity score by combining the average duration and volatility. The longer the duration and the smaller the volatility, the higher the score. Scoring criteria: Average duration is long (more than 30 minutes): score 4-5 points. Average duration is moderate (20-30 minutes): score 3-4 points. Average duration is short (less than 20 minutes): score 1-2 points. Small volatility (difference is less than 20 minutes): increase by 1 point. Large volatility (difference is more than 30 minutes): decrease by 1 point. According to this standard, the scores are as follows: Light sleep score: average duration is 30 minutes (moderate), with fluctuations of 15 minutes (small), so the score is 4 points. Deep sleep score: average duration is 36.25 minutes (higher), with fluctuations of 33.33 minutes (larger), so the score is 3.5 points. REM score: average duration is 17.5 minutes (shorter), with fluctuations of 15 minutes (smaller), so the score is 3 points.

[0155] In one embodiment, according to the sleep cycle characteristic data, the continuous time period information of each sleep cycle is extracted, and the duration of the light sleep, deep sleep and REM stages is classified. The total duration of each sleep stage (light sleep, deep sleep, REM) is calculated. The average duration of each stage in a single cycle is evaluated to measure continuity. The longer the continuous time, the more stable the stage is. The average continuous time = the total time of each stage / the number of cycles that occur. The continuity fluctuations of each stage are evaluated, that is, the difference in continuous time between stages. High fluctuations mean that sleep is not stable enough, and low fluctuations mean a more stable sleep stage. For fluctuation calculations, the standard deviation of the continuous time difference of each stage is used to quantify. The frequency of changes in the light sleep, deep sleep, and REM stages is analyzed, that is, the number of times each stage switches in the entire sleep cycle. Frequent switching means lower sleep stability. The conversion of light sleep-deep sleep-REM is relatively frequent, indicating that the fluctuations of the sleep stage are large in cycles 1-3. In cycle 4, deep sleep dominates, indicating that this cycle is relatively stable. In order to facilitate overall evaluation, the continuity of each stage is usually quantified by a scoring system. For example, a scoring standard can be set with 5 points as the highest continuity (no obvious fluctuations, long continuous time) and 1 point as the lowest continuity (frequent switching, short continuous time and large fluctuations). Scoring is based on the following factors: Average continuous time: longer times have higher scores. Continuity fluctuations: the smaller the fluctuations, the higher the score. Light sleep score: average continuous time of 30 minutes (high), fluctuations of 15 minutes (medium), medium switching frequency → score of 3.5 points. Deep sleep score: average continuous time of 36.25 minutes (high), fluctuations of 33.33 minutes (large), low switching frequency → score of 4 points. REM score: average continuous time of 17.5 minutes (low), fluctuations of 15 minutes (medium), high switching frequency → score of 3 points. The continuity scores of each sleep stage are integrated into an overall continuity score, which reflects the stability of the entire sleep cycle. The overall score can be evaluated by weighted average, according to the importance of each stage (deep sleep contributes more to sleep quality and is given greater weight). Deep sleep is weighted 40%, light sleep 30%, and REM 30%: overall score = (3.5 × 0.3) + (4 × 0.4) + (3 × 0.3) = 3.55 points.

[0156] Calculate the sleep depth fluctuation frequency according to the sleep state cycle characteristic data to obtain sleep fluctuation frequency characteristic data;

[0157] In one embodiment, the fluctuation frequency of sleep depth is calculated to reflect the switching frequency between deep sleep and light sleep. The switching frequency between deep sleep and light sleep stages is analyzed using a frequency domain analysis method, such as fast Fourier transform (FFT). The switching frequency between light sleep and REM in cycle 3 is 0.1 Hz, and the switching frequency between light sleep and deep sleep in cycle 1 is 0.05 Hz. The fluctuation frequencies of each cycle are integrated and used as sleep fluctuation frequency feature data. Sleep fluctuation frequency feature data = [Cycle 1: switching frequency 0.05 Hz, cycle 3: switching frequency 0.1 Hz].

[0158] The sleep quality index is calculated based on the sleep state stability assessment data, sleep wake event data, sleep stage continuity characteristic data and sleep fluctuation frequency characteristic data to obtain the sleep state quality characteristic data.

[0159] In one embodiment, stability, wake events, continuity and fluctuation frequency are weighted averaged according to certain weights, such as stability weight 0.4, wake events weight 0.2, stage continuity weight 0.2, and fluctuation frequency weight 0.2. According to the example data, the sleep quality index is calculated, such as stability score: 8 (low volatility high score), wake score: 7 (few wake events high score), continuity score: 9 (long stage high score), fluctuation frequency score: 6 (stable frequency high score), sleep quality index = 0.4 × 8 + 0.2 × 7 + 0.2 × 9 + 0.2 × 6 = 7.6. Sleep state quality feature data = sleep quality index 7.6 (reflecting higher sleep quality).

[0160] Preferably, the sleep state feature fusion is specifically:

[0161] Generate a feature fusion matrix based on the sleep state change feature data and the sleep state quality feature data to obtain sleep feature fusion matrix data;

[0162] In one embodiment, sleep state change characteristic data (such as change rate, frequency, etc.) and sleep state quality characteristic data (such as quality index, stability, etc.) are taken and constructed into a two-dimensional matrix. Change characteristic data = [heart rate change rate 1.5bpm / minute, breathing change rate 0.8 times / minute, fluctuation frequency 0.05Hz], quality characteristic data = [sleep quality index 7.6, stability index 8.2, wakefulness event score 9]. The sleep feature fusion matrix data is [(1.5, 0.8, 0.05), (7.6, 8.2, 9)].

[0163] Performing feature weighting on the sleep state change feature data and the sleep state quality feature data according to the sleep feature fusion matrix data to obtain sleep state change feature weighted data and sleep state quality feature weighted data;

[0164] In one embodiment, a different weight is given to each feature to emphasize its importance in the overall sleep state. The weighting coefficient is set according to the importance of the feature. For example, the sleep quality feature is more important, so it is given a higher weight. The weighting coefficients are as follows: For change features: heart rate change rate weight 0.3, breathing change rate weight 0.2, fluctuation frequency weight 0.1. For quality features: sleep quality index weight 0.4, stability weight 0.3, wakefulness event score weight 0.3 By applying these weights to each feature of the fusion matrix, weighted data is obtained: change feature weight = [1.5×0.3, 0.8×0.2, 0.05×0.1] = [0.45, 0.16, 0.005], quality feature weight = [7.6×0.4, 8.2×0.3, 9×0.3] = [3.04, 2.46, 2.7]. The weighted data is stored as sleep state change feature weighted data and sleep state quality feature weighted data. The weighted data of sleep state change characteristics = [0.45, 0.16, 0.005]; the weighted data of sleep state quality characteristics = [3.04, 2.46, 2.7].

[0165] Performing deep feature weighting according to the weighted data of sleep state change characteristics and the weighted data of sleep state quality characteristics to obtain preliminary sleep state characteristic data;

[0166] In one embodiment, the change feature and the quality feature are further integrated into a preliminary sleep state feature to enhance the overall expressiveness of the feature. The change feature weighted data and the quality feature weighted data are combined to generate a depth-weighted feature vector. For example, the combination method is simple addition: preliminary feature data = [0.45 + 3.04, 0.16 + 2.46, 0.005 + 2.7] = [3.49, 2.62, 2.705].

[0167] Feature rebalancing is performed based on the preliminary sleep state feature data to obtain sleep state feature data.

[0168] In one embodiment, the preliminary feature data is normalized and rebalanced to ensure that the influence of each feature is on the same scale, thereby generating a final sleep state feature. The preliminary sleep state feature data is normalized, for example, each value is scaled to the interval of 0 to 1. For example, the maximum value is 3.49: rebalanced feature data = [1, 0.75, 0.775]. The rebalanced feature data is represented in the form of a feature vector.

[0169] Preferably, step S3 is specifically:

[0170] Step S31: mapping according to a preset ambient sound type library and sleep state characteristic data to obtain primary ambient sound type data;

[0171] In one embodiment, a suitable initial ambient sound type is selected according to the current sleep state characteristics of the user to assist sleep. Referring to the ambient sound type library (including audio types suitable for different sleep stages, such as rain, ocean waves, white noise, etc.), matching audio is selected according to the sleep state characteristic data. The sleep state characteristic data indicates that the user is in a light sleep state and has a relatively stable heart rate, which is suitable for playing soft white noise. According to the mapping of sleep state characteristic data (such as light sleep and stability) to soft white noise, the primary ambient sound type is obtained. Primary ambient sound type data = soft white noise.

[0172] Step S32: Acquire environmental noise data;

[0173] In one embodiment, the noise in the current environment is detected so as to adapt to the current ambient sound level during the simulated playback. A microphone or a noise detection sensor is used to capture the ambient noise level and characteristics (e.g., the frequency and intensity of the noise). The ambient noise data is: 50 decibels, and the frequency range is concentrated from 200 Hz to 500 Hz. The detected ambient noise data is sorted into noise intensity and frequency information. Ambient noise data = 50 dB, 200 Hz to 500 Hz.

[0174] Step S33: performing simulated playback according to the primary ambient sound type data and the ambient noise data to obtain ambient sound playback effect simulation data;

[0175] In one embodiment, the ambient sound playback effect is simulated in advance to detect its adaptability in the current environment. The primary ambient sound (soft white noise) is combined with the ambient noise data and audio superposition is performed to simulate the final sound effect heard by the user. The white noise volume is set to 45 decibels and is integrated with the ambient noise in the frequency range of 200Hz to 500Hz. Test the ambient sound playback effect. For example, the simulated effect is that the 45-decibel white noise has a high degree of integration with the ambient noise and is suitable for playback. Ambient sound playback effect simulation data = 45dB white noise, well integrated.

[0176] Step S34: extracting playback effect features according to the ambient sound playback effect simulation data to obtain playback effect feature data;

[0177] In one embodiment, the simulated playback effect is evaluated and relevant features are extracted to further optimize the ambient sound. Features such as volume, frequency coverage, and sound quality are extracted from the simulated effect. For example, observe whether there are harsh frequencies or insufficient volume. The volume is 45 decibels, the frequency covers 200Hz to 500Hz smoothly, there are no harsh frequencies, and the sound quality is clear. The extracted feature data such as volume and frequency are sorted for use in adjusting the sleep state characteristics. Playback effect feature data = volume 45dB, frequency 200Hz to 500Hz, and clear sound quality.

[0178] Step S35: adjusting the sleep state characteristic data according to the playback effect characteristic data to obtain sleep state characteristic adjustment data;

[0179] In one embodiment, the sleep state characteristics are fine-tuned in combination with the playback effect characteristic data to optimize the response to the ambient sound. The playback effect characteristic data is fed back into the sleep state characteristic data for subtle adjustments. For example, the volume is adjusted appropriately to avoid irritation to sleep. The volume is adjusted from 45 decibels to 40 decibels, and the frequency range is maintained at 200Hz to 500Hz to avoid over-stimulation. The adjusted sleep state characteristics are recorded to adapt to the ambient sound playback effect. Sleep state characteristic adjustment data = volume 40dB, frequency 200Hz to 500Hz.

[0180] Step S36: mapping according to the preset ambient sound type library and the sleep state characteristic adjustment data to obtain secondary ambient sound type data;

[0181] In one embodiment, according to the adjusted sleep state characteristic data, a more optimized audio type is selected from the ambient sound type library. The sleep state characteristic adjustment data is used to map to the ambient sound type library again, and a more suitable ambient sound is selected. For example, a slightly different white noise frequency is selected to achieve a better effect. The adjusted volume and frequency are more suitable for low-frequency white noise. The secondary ambient sound type data is determined to be low-frequency white noise. Secondary ambient sound type data = low-frequency white noise.

[0182] Step S37: Perform audio fusion according to the primary ambient sound type data and the secondary ambient sound type data to obtain the ambient sound type data.

[0183] In one embodiment, the data of the primary and secondary ambient sound types are combined to generate an ambient sound type for playback. The primary and secondary audio types are mixed, for example, the primary ambient sound is soft white noise, and the secondary ambient sound is low-frequency white noise, and they are mixed into a gradual sound effect. The volume and frequency range of the final ambient sound type are adjusted so that it smoothly covers 200Hz to 400Hz at a volume of 40 decibels. The fused ambient sound type data is stored for actual playback. Ambient sound type data = low-frequency white noise (volume 40dB, frequency 200Hz to 400Hz).

[0184] Preferably, step S4 is specifically:

[0185] Step S41: adjusting the ambient sound frequency according to the ambient sound type data and the sleep state characteristic data to obtain ambient sound frequency adjustment data;

[0186] In one embodiment, the frequency of the ambient sound is adjusted according to the user's current sleep state characteristic data to make it more in line with the current sleep stage requirements. Based on the ambient sound type (e.g., low-frequency rain sound) and sleep state characteristic data (e.g., light sleep, deep sleep, etc.), the audio frequency range is adjusted. The current ambient sound type is low-frequency rain sound, the target frequency range is 300Hz to 500Hz, and the frequency is adjusted to 350Hz to 400Hz according to the sleep state characteristic data to enhance the deep sleep effect. The adjusted frequency is recorded as ambient sound frequency adjustment data. Ambient sound frequency adjustment data = 350Hz to 400Hz.

[0187] Step S42: Acquire real-time sleep status data;

[0188] In one embodiment, the user's current sleep state is monitored, and the changes in sleep stages are fed back in real time to facilitate adjustment of ambient sound. Use sleep monitoring sensors (such as heart rate sensors, respiratory rate sensors, etc. in wearable devices) to collect heart rate, respiratory rate and body movement data in real time. The real-time sleep state data is as follows: heart rate 60bpm, respiratory rate 14 times / minute, and little body movement, indicating that the user is in deep sleep. The collected real-time data is sorted for use in subsequent steps. Real-time sleep state data = heart rate 60bpm, respiratory rate 14 times / minute, and little body movement.

[0189] Step S43: updating and iterating the sleep state characteristic data according to the real-time sleep state data to obtain the real-time sleep state characteristic data;

[0190] In one embodiment, the current sleep state feature is updated according to the real-time sleep state data to reflect the latest sleep situation. The sleep state feature is iteratively updated according to the real-time data. If the heart rate decreases and the body movement decreases, it is marked as a deep sleep state and the feature data is updated. After the feature is updated, the current state is deep sleep, and the updated feature data is as follows: the duration of deep sleep increases to 40 minutes, and the stability score is improved. The updated data is stored as real-time sleep state feature data, and the real-time sleep state feature data = deep sleep state, heart rate 60bpm, and high stability score.

[0191] Step S44: adjusting the ambient sound volume according to the ambient sound type data and the real-time sleep state characteristic data to obtain ambient sound volume adjustment data;

[0192] In one embodiment, the volume of the ambient sound is adjusted according to the current sleep stage characteristics to ensure that the volume adapts to the user's deep sleep or light sleep state. If the real-time characteristics show deep sleep, the volume of the ambient sound is lowered to ensure that the sound effects do not interfere with deep sleep. Assume that the current volume is 50 decibels, adjust it to 40 decibels. If the characteristic data is updated to the light sleep stage, slightly increase the volume (such as to 45 decibels) to promote relaxation. The adjusted volume is recorded as the ambient volume adjustment data. Ambient volume adjustment data = 40dB (adapt to deep sleep).

[0193] Step S45: performing a sleep-improving ambient sound control auxiliary operation according to the ambient sound type data, the ambient sound frequency adjustment data, and the ambient sound volume adjustment data.

[0194] In one embodiment, the frequency and volume of the ambient sound are adjusted in real time to synchronize with the user's sleep state to continuously optimize sleep quality. The ambient sound type (low-frequency rain sound), frequency adjustment (350Hz to 400Hz) and volume adjustment (40 decibels) data are combined, and the playback parameters are adjusted in real time through the playback device. The audio output is adjusted regularly according to the changes in the user's sleep stages. For example, the volume is gradually lowered to match deep sleep or the volume is moderately increased to promote the transition from light sleep to deep sleep. The system automatically adjusts the volume and frequency of the ambient sound to meet the user's current sleep needs and outputs the optimized ambient sound effect. Sleep improvement ambient sound control auxiliary operation completed = low-frequency rain sound, frequency 350Hz to 400Hz, volume 40dB (adapting to deep sleep).

[0195] Preferably, a sleep-improving ambient sound control assistance system is used to execute the sleep-improving ambient sound control assistance method as described above, and the sleep-improving ambient sound control assistance system includes:

[0196] A sleep state acquisition module, used to acquire sleep state data;

[0197] A sleep state feature extraction module, used to extract sleep state features according to the sleep state data to obtain sleep state feature data;

[0198] An ambient sound type selection module is used to select an ambient sound type according to the sleep state characteristic data to obtain ambient sound type data;

[0199] The ambient sound frequency adjustment and ambient sound volume adjustment modules are used to adjust the ambient sound frequency and ambient sound volume according to the ambient sound type data and the sleep state characteristic data, and obtain the ambient sound frequency adjustment data and the ambient sound volume adjustment data respectively, so as to perform auxiliary operations of sleep improvement ambient sound control.

[0200] Therefore, from any point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the attached application documents rather than the above description, and it is intended that all changes falling within the meaning and scope of equivalent elements of the application documents are included in the present invention.

[0201] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A sleep-improving ambient sound control auxiliary method, characterized in that: The following steps are involved: Step S1: Obtaining sleep state data; Step S2: extracting sleep state features according to the sleep state data to obtain sleep state feature data; Step S3: Selecting an ambient sound type according to the sleep state characteristic data to obtain ambient sound type data; Step S4: adjusting the ambient sound frequency and the ambient sound volume according to the ambient sound type data and the sleep state characteristic data, and obtaining ambient sound frequency adjustment data and ambient sound volume adjustment data, respectively, to perform a sleep-improving ambient sound control auxiliary operation.

2. The method according to claim 1, characterized in that Step S1 is specifically as follows: Collect user heart rate data through a heart rate sensor to obtain user heart rate data; The user's respiratory frequency data is collected through a respiratory sensor to obtain the user's respiratory frequency data; Collecting user body motion data through an acceleration sensor to obtain user body motion data; The user's sleep state is determined based on the user's heart rate data, the user's breathing rate data, and the user's body movement data to obtain sleep state data.

3. The method according to claim 2, characterized in that The user's body movement data collection is specifically as follows: Acceleration amplitude data is collected by using an acceleration sensor to obtain acceleration amplitude data; Perform frequency analysis based on acceleration amplitude data to obtain acceleration frequency data; Perform sleep movement event detection based on the acceleration frequency data and the acceleration amplitude data to obtain the user's sleep movement event data; The motion data is analyzed based on the user's sleep motion event data to obtain the user's body motion data.

4. The method according to claim 1, characterized in that Step S2 is specifically as follows: Extracting sleep state change features according to the sleep state data to obtain sleep state change feature data; Extracting sleep state quality features according to the sleep state data to obtain sleep state quality feature data; The sleep state change characteristic data and the sleep state quality characteristic data are fused to obtain sleep state characteristic data.

5. The method according to claim 4, characterized in that The sleep state change feature extraction is specifically as follows: Segment the sleep cycle according to the sleep state data to obtain sleep cycle segmentation data; Perform sleep change recognition according to the sleep cycle segmentation data to obtain sleep change recognition data; Perform feature dimensionality reduction according to the sleep change recognition data to obtain sleep change recognition dimensionality reduction data; Perform frequency domain analysis on the dimensionality reduction data based on sleep change identification to obtain sleep change frequency feature data; Perform sequence pattern mining based on the sleep change frequency feature data to obtain sleep change sequence pattern data; The change index is calculated based on the sleep change sequence pattern data to obtain the sleep state change characteristic data.

6. The method according to claim 4, characterized in that The sleep state quality feature extraction is specifically as follows: Extract sleep cycle features according to the sleep state data to obtain sleep state cycle feature data; Performing sleep stability assessment based on sleep state cycle characteristic data to obtain sleep state stability assessment data; Perform wakefulness detection based on sleep state stability assessment data to obtain sleep wakefulness event data; Performing sleep stage continuity analysis based on sleep state stability assessment data to obtain sleep stage continuity feature data; Calculate the sleep depth fluctuation frequency according to the sleep state cycle characteristic data to obtain sleep fluctuation frequency characteristic data; The sleep quality index is calculated based on the sleep state stability assessment data, sleep wake event data, sleep stage continuity characteristic data and sleep fluctuation frequency characteristic data to obtain the sleep state quality characteristic data.

7. The method according to claim 4, characterized in that The sleep state feature fusion is as follows: Generate a feature fusion matrix based on the sleep state change feature data and the sleep state quality feature data to obtain sleep feature fusion matrix data; Performing feature weighting on the sleep state change feature data and the sleep state quality feature data according to the sleep feature fusion matrix data to obtain sleep state change feature weighted data and sleep state quality feature weighted data; Performing deep feature weighting according to the weighted data of sleep state change characteristics and the weighted data of sleep state quality characteristics to obtain preliminary sleep state characteristic data; Feature rebalancing is performed based on the preliminary sleep state feature data to obtain sleep state feature data.

8. The method according to claim 1, characterized in that: Step S3 is specifically as follows: Mapping is performed according to a preset ambient sound type library and sleep state characteristic data to obtain primary ambient sound type data; Obtain environmental noise data; Perform simulated playback according to the primary ambient sound type data and the ambient noise data to obtain ambient sound playback effect simulation data; Extracting playback effect features based on the ambient sound playback effect simulation data to obtain playback effect feature data; Performing characteristic adjustment on the sleep state characteristic data according to the playback effect characteristic data to obtain sleep state characteristic adjustment data; Mapping is performed according to a preset ambient sound type library and sleep state characteristic adjustment data to obtain secondary ambient sound type data; Audio fusion is performed according to the primary ambient sound type data and the secondary ambient sound type data to obtain the ambient sound type data.

9. The method according to claim 1, characterized in that: Step S4 is specifically as follows: Adjust the ambient sound frequency according to the ambient sound type data and the sleep state characteristic data to obtain ambient sound frequency adjustment data; Get real-time sleep status data; The sleep state characteristic data is updated and iterated according to the real-time sleep state data to obtain the real-time sleep state characteristic data; Adjust the ambient sound volume according to the ambient sound type data and the real-time sleep state characteristic data to obtain ambient sound volume adjustment data; Perform sleep improvement ambient sound control auxiliary operations based on ambient sound type data, ambient sound frequency adjustment data, and ambient sound volume adjustment data.

10. A sleep-improving ambient sound control auxiliary system, characterized in that: The sleep-improving environmental sound control assistance system is used to execute the sleep-improving environmental sound control assistance method as claimed in claim 1, and the sleep-improving environmental sound control assistance system comprises: A sleep state acquisition module, used to acquire sleep state data; A sleep state feature extraction module, used to extract sleep state features according to the sleep state data to obtain sleep state feature data; An ambient sound type selection module is used to select an ambient sound type according to the sleep state characteristic data to obtain ambient sound type data; The ambient sound frequency adjustment and ambient sound volume adjustment modules are used to adjust the ambient sound frequency and ambient sound volume according to the ambient sound type data and the sleep state characteristic data, and obtain the ambient sound frequency adjustment data and the ambient sound volume adjustment data respectively, so as to perform auxiliary operations of sleep improvement ambient sound control.

Images