Sleep state self-adaptive layered regulation and control method and platform based on environment regulation window

Abstract

The invention discloses a sleep state self-adaptive layered regulation and control method and platform based on an environment regulation window, and relates to the technical field of self-adaptive control, and the method comprises the steps: carrying out the sleep state analysis according to a physiological data sequence, obtaining a real-time sleep stage, carrying out the scheduling task distribution, and generating a function unit scheduling sequence; performing data acquisition priority management and control on the wearable device to obtain real-time sleep associated data; generating a sleep environment adjustment strategy by taking the real-time sleep stage as an environment adjustment constraint; after updating the sleep environment of the user, activating an environment adjusting window; and operating a hierarchical scheduling architecture by taking an environment adjustment window as a time constraint, and circularly performing adaptive control updating. The technical problem that in the prior art, due to the fact that the sleep environment is difficult to adjust accurately in real time according to the sleep state of the user, the sleep quality of the user is low is solved, and the technical effects that the sleep state of the user is accurately judged, the sleep environment is dynamically and adaptively adjusted, and the sleep quality is improved are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of adaptive control, in particular to a sleep state adaptive hierarchical regulation method and platform based on an environment regulation window. BACKGROUND

[0002] With the acceleration of modern life rhythm and the increasing emphasis on healthy sleep, the influence of sleep environment on sleep quality is increasingly concerned. Sleep is a complex physiological process, which is affected by a variety of factors, among which the sleep environment factor plays a key role. The traditional sleep environment regulation method is relatively single and fixed, and lacks real-time monitoring and dynamic response ability to individual sleep state of users. In the sleep environment regulation process, how to realize the coordinated optimization of multiple factors (such as temperature, noise, illumination, humidity, etc.) is also a problem to be solved. Different sleep stages have different requirements for environmental factors, and there are differences between individuals, and the traditional technology is difficult to balance the regulation of multiple factors while meeting the individual needs. At the same time, with the development of smart home technology, although some home appliances that can be remotely controlled have appeared, they lack unified scheduling and coordination mechanism, and cannot form an intelligent sleep environment regulation system.

[0003] The prior art has the technical problem that the sleep environment is difficult to be accurately regulated in real time according to the sleep state of the user, resulting in low sleep quality of the user. SUMMARY

[0004] The present application provides a sleep state adaptive hierarchical regulation method and platform based on an environment regulation window, which is used to solve the technical problem that the sleep environment is difficult to be accurately regulated in real time according to the sleep state of the user in the prior art, resulting in low sleep quality of the user.

[0005] In view of the above problems, the present application provides a sleep state adaptive hierarchical regulation method and platform based on an environment regulation window.

[0006] In a first aspect, the present application provides a sleep state adaptive hierarchical regulation method based on an environment regulation window, which comprises: According to the physiological data sequence uploaded by the wearable device, the sleep state is analyzed, the real-time sleep stage is obtained, the hierarchical scheduling architecture is scheduled according to the real-time sleep stage, and a function unit scheduling sequence is generated; the data acquisition priority of the wearable device is controlled by using the function unit scheduling sequence, and real-time sleep-related data is obtained; the real-time sleep stage is used as an environmental regulation constraint, and a sleep environment regulation strategy is generated according to the real-time sleep-related data; after updating the user sleep environment according to the sleep environment regulation strategy, the environmental regulation window is activated; the hierarchical scheduling architecture is run with the environmental regulation window as a time constraint and the function unit scheduling sequence as a scheduling constraint, and the sleep stage judgment and the adaptive control update of the user sleep environment are circularly performed; when the real-time sleep stage enters the wake-up stage and the duration meets the preset wake-up window, the user sleep environment is forcibly switched and updated according to the coverage scale of the preset wake-up period with respect to the wake-up stage timestamp.

[0007] In a second aspect of the present application, a sleep state adaptive hierarchical regulation platform based on an environmental regulation window is provided, and the platform comprises: A function unit scheduling sequence generation module is configured to analyze the sleep state according to the physiological data sequence uploaded by the wearable device, obtain the real-time sleep stage, and assign scheduling tasks to the hierarchical scheduling architecture according to the real-time sleep stage, thereby generating a function unit scheduling sequence; a real-time sleep-related data acquisition module is configured to control the data acquisition priority of the wearable device by using the function unit scheduling sequence, and obtain real-time sleep-related data; a sleep environment regulation strategy generation module is configured to use the real-time sleep stage as an environmental regulation constraint, and generate a sleep environment regulation strategy according to the real-time sleep-related data; an environmental regulation window activation module is configured to activate the environmental regulation window after updating the user sleep environment according to the sleep environment regulation strategy; an adaptive control update module is configured to run the hierarchical scheduling architecture with the environmental regulation window as a time constraint and the function unit scheduling sequence as a scheduling constraint, and circularly perform sleep stage judgment and adaptive control update of the user sleep environment; and a forced switching update module is configured to forcibly switch and update the user sleep environment according to the coverage scale of the preset wake-up period with respect to the wake-up stage timestamp when the real-time sleep stage enters the wake-up stage and the duration meets the preset wake-up window.

[0008] One or more technical solutions provided in the present application have at least the following technical effects or advantages: According to the physiological data sequence uploaded by the wearable device, the sleep state analysis is performed, the real-time sleep stage is obtained, the scheduling task allocation is performed, and the function unit scheduling sequence is generated; the data acquisition priority control of the wearable device is performed by using the function unit scheduling sequence, and real-time sleep-related data is obtained; the sleep environment adjustment strategy is generated by taking the real-time sleep stage as an environmental adjustment constraint; after updating the user sleep environment according to the sleep environment adjustment strategy, the environmental adjustment window is activated; the hierarchical scheduling architecture is run by taking the environmental adjustment window as a time constraint and the function unit scheduling sequence as a scheduling constraint, and the sleep stage judgment and adaptive control update are performed in a loop; when the real-time sleep stage enters the wake-up stage for a time period that satisfies a preset wake-up window, the user sleep environment is forcibly switched and updated according to the coverage scale of the wake-up stage timestamp on the preset wake-up period. The technical effect of realizing accurate judgment of the user sleep state and dynamic adaptive adjustment of the sleep environment to improve the sleep quality is achieved. BRIEF DESCRIPTION OF DRAWINGS

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0010] Figure 1 The sleep state adaptive hierarchical regulation method based on the environmental adjustment window provided by the embodiments of the present application is shown in the flowchart. Figure 2 The sleep state adaptive hierarchical regulation platform structure based on the environmental adjustment window provided by the embodiments of the present application is shown in the schematic diagram.

[0011] Explanation of reference signs: function unit scheduling sequence generation module 10, real-time sleep-related data acquisition module 20, sleep environment adjustment strategy generation module 30, environmental adjustment window activation module 40, adaptive control update module 50, and forced switching update module 60. DETAILED DESCRIPTION

[0012] The present application provides a sleep state adaptive hierarchical regulation method and platform based on an environmental adjustment window, which is used to solve the technical problem that the sleep environment in the prior art is difficult to be accurately adjusted in real time according to the user sleep state, resulting in low user sleep quality.

[0013] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0014] As shown in Embodiment One, Figure 1 The present application provides a sleep state adaptive hierarchical regulation method based on an environmental regulation window, which comprises the following steps: Step S100: After obtaining a real-time sleep stage according to sleep state analysis on a physiological data sequence uploaded by a wearable device, a scheduling task allocation is performed on a hierarchical scheduling architecture according to the real-time sleep stage, and a functional unit scheduling sequence is generated.

[0015] Specifically, first, a physiological data sequence uploaded by a wearable device is received, which contains information about changes in various physiological indicators of a user's body over time, and then a sleep state analysis model is used to deeply process the sequence. Through comprehensive analysis of multi-dimensional physiological data such as heart rate, respiratory rate, and body movement, the current sleep state of the user is accurately determined, and the real-time sleep stage is determined, such as light sleep, deep sleep, and rapid eye movement. When the real-time sleep stage is obtained, a scheduling task allocation is performed on the hierarchical scheduling architecture. At the scheduling level of the functional unit, according to the different needs of different sleep stages for various monitoring functional units, the activation, deactivation, and running priority of each functional unit are reasonably determined. For example, in the deep sleep stage, more emphasis is placed on the priority scheduling of monitoring functional units related to the stability of respiration and heart rate, to ensure close attention to the user's critical physiological state; and in the light sleep period, resources are appropriately allocated to body movement monitoring functional units to assist in determining the stability of sleep. At the same time, in the scheduling of monitoring tasks within the functional unit, according to the characteristics of the real-time sleep stage, the priority of each specific monitoring task within the functional unit is set. For example, in the rapid eye movement period, which is sensitive to the sleep environment, noise and light monitoring tasks are given higher priority to timely discover changes in environmental factors that may interfere with sleep. Through such detailed two-level scheduling task allocation, a functional unit scheduling sequence is finally generated, which clearly specifies the execution order and resource allocation strategy of each functional unit and its internal monitoring tasks in the current real-time sleep stage, laying a solid foundation for subsequent accurate data collection and environmental regulation.

[0016] Step S200: The functional unit scheduling sequence is used to control the data collection priority of the wearable device, and real-time sleep-related data is obtained.

[0017] Specifically, for functional units deemed crucial for sleep state assessment and environmental regulation during the current sleep stage, their corresponding data acquisition tasks will be assigned the highest priority. For example, if the current sleep stage is deep sleep, data collected by functional units related to respiration and heart rate monitoring is essential for assessing sleep quality and physiological state. The system will prioritize ensuring these functional units collect data stably and frequently to obtain the most accurate real-time physiological information. Meanwhile, for less important functional units, such as environmental temperature and humidity monitoring units (which are less important than respiration and heart rate monitoring during deep sleep but still valuable for overall sleep environment assessment), data acquisition will be performed according to a predetermined lower priority. For functional units that are less critical at the current stage, their data acquisition frequency may be appropriately reduced or even suspended to avoid excessive consumption of device resources and transmission bandwidth, while ensuring high-quality acquisition of key data. Through this strict data acquisition priority control based on the functional unit scheduling sequence, the wearable device can accurately acquire the most valuable data closely related to the current sleep stage. After integration and processing, this data forms real-time sleep-related data. Real-time sleep-related data not only covers key information about the user's physiological state, such as changes in physiological indicators like real-time heart rate and respiratory rate, but also includes relevant data on environmental factors, such as the temperature, humidity, and noise level of the current sleep environment. This provides a comprehensive and reliable data foundation for the subsequent accurate generation of sleep environment regulation strategies.

[0018] Step S300: Using the real-time sleep stage as an environmental regulation constraint, generate a sleep environment regulation strategy based on the real-time sleep correlation data.

[0019] Specifically, a thorough analysis of various information in real-time sleep data is conducted. If the data shows a slight increase in heart rate and more frequent body movement during a certain sleep stage, it's inferred that the user is in light sleep and may be disturbed by external factors. In this case, when generating a sleep environment adjustment strategy, priority will be given to reducing environmental noise and adjusting light intensity to a more suitable, gentle state to ensure a quiet and comfortable sleep environment and reduce the impact of external disturbances on the user's sleep. Simultaneously, regarding factors such as ambient temperature and humidity, based on the user's body temperature changes and humidity perception data in the real-time sleep data, combined with the ideal temperature and humidity requirements for each sleep stage, the optimal temperature and humidity range for promoting deep sleep is calculated. The operating parameters of devices such as air conditioners, humidifiers, or dehumidifiers are then adjusted to control the temperature and humidity of the sleep environment at optimal levels. For carbon dioxide and oxygen concentrations, if the data indicates that the concentrations deviate from the ideal range, air purification or ventilation equipment will be activated to ensure good air quality in the sleep environment. Furthermore, if real-time sleep data includes pillow height information (such as information obtained through sensors detecting changes in the relative position of the user's head and pillow), the pillow can be adjusted to the most comfortable height using a smart pillow height adjustment device based on the sleep stage and the user's personal habits. This maintains the user's natural spinal curve and improves sleep comfort. Even regarding fragrance, if sleep data suggests that the user may benefit from a specific fragrance to soothe emotions or aid sleep during their current sleep stage, the fragrance diffuser can be controlled to release an appropriate amount of fragrance, creating a more pleasant sleep atmosphere from an olfactory perspective. Through this comprehensive analysis method based on real-time sleep stage constraints and combined with real-time sleep data, highly accurate and personalized sleep environment adjustment strategies are generated, comprehensively creating a sleep environment most suitable for the user's current sleep state and maximizing sleep quality.

[0020] Step S400: After updating the user's sleep environment according to the sleep environment adjustment strategy, activate the environment adjustment window.

[0021] Specifically, following the instructions in the sleep environment regulation strategy, multiple elements of the sleep environment are updated. For example, the bedroom temperature is adjusted to a comfortable level for sleep, and a smart thermostat is used to stabilize the room temperature at the level most conducive to the current sleep stage; the brightness and color temperature of lighting are adjusted to match the body's biological clock and sleep needs at that stage, such as dimming the lights to near darkness during deep sleep to avoid light interference; noise reduction devices are controlled to reduce environmental noise and create a quiet sleep atmosphere; the operating parameters of air purifiers are adjusted to optimize indoor carbon dioxide and oxygen concentrations; the pillow height may also be adjusted to a comfortable level, and appropriate amounts of sleep-inducing fragrances may be released. After completing the above comprehensive sleep environment update, the environment regulation window is immediately activated. This window acts like a time-limited frame, setting a period for close monitoring of dynamic changes in the sleep environment and timely response and adjustment to any fluctuations in environmental parameters based on the environment regulation strategy. This time period is set based on research on sleep cycles and the body's ability to adapt to environmental changes, ensuring that the sleep environment remains optimal during key sleep periods. During the activation of the environmental regulation window, environmental data such as temperature, humidity, light intensity, and noise level are continuously monitored. Once these data are found to deviate from the target range set by the sleep environment regulation strategy, the corresponding regulation mechanism is quickly activated to correct them, thereby ensuring that users can enjoy a stable and suitable sleep environment throughout the sleep process and further improve sleep quality.

[0022] Step S500: Using the environmental adjustment window as a time constraint and the functional unit scheduling sequence as a scheduling constraint, run the hierarchical scheduling architecture to cyclically perform sleep stage judgment and adaptive control update of the user's sleep environment.

[0023] Specifically, using the activated environmental adjustment window as a strict time constraint and the functional unit scheduling sequence as a precise scheduling guide, the hierarchical scheduling architecture operates at full capacity, initiating a cyclical optimization process of sleep stage judgment and user sleep environment adaptive control updates. The environmental adjustment window clearly defines the specific time period during which focused efforts must be made to precisely control the sleep environment, ensuring that environmental factors consistently meet the user's sleep needs. Within this time frame, the hierarchical scheduling architecture begins to operate efficiently, using the functional unit scheduling sequence as the scheduling criterion. Regarding sleep stage judgment, physiological data collected from wearable devices is continuously utilized, combined with the previously constructed sleep stage judgment model, to perform in-depth analysis of the user's real-time sleep state, quickly and accurately determining the user's sleep stage, such as light sleep, deep sleep, or REM sleep. Based on the newly determined sleep stage, the hierarchical scheduling architecture immediately dynamically adjusts and prioritizes each monitoring functional unit and its internal monitoring tasks according to the functional unit scheduling sequence. For example, if a user enters deep sleep, the priority of the physiological data stability monitoring unit is increased to ensure that any subtle physiological changes are captured in a timely manner. Simultaneously, the environmental monitoring unit will allocate resources rationally based on the environmental requirements of deep sleep, increasing the monitoring frequency and accuracy of key environmental factors such as temperature and noise. Based on the new sleep stage judgment results and functional unit scheduling adjustments, a more precise sleep environment adjustment strategy is generated based on real-time sleep correlation data. This strategy is then quickly applied to the user's sleep environment, dynamically updating environmental elements such as temperature, humidity, light, noise, pillow height, and fragrance in real time. For example, if a slight drop in the user's body temperature is detected during deep sleep, the ambient temperature will be fine-tuned to maintain it within the most suitable sleep temperature range, while ensuring that other environmental factors are also in an ideal state. This process is repeated continuously within the time constraint of the environmental adjustment window, involving sleep stage judgment, functional unit scheduling adjustments, sleep environment adjustment strategy generation, and environmental updates, achieving continuous adaptive control and updates of the user's sleep environment. Each cycle makes the sleep environment more closely match the user's actual needs at the current sleep stage, providing users with a consistently high-quality sleep experience, effectively improving sleep quality and promoting the stability and health of the sleep cycle.

[0024] Step S600: When the duration of the real-time sleep stage entering the wakefulness stage meets the preset wake-up window, the user's sleep environment is forcibly switched and updated according to the coverage scale of the preset wakefulness period based on the wakefulness stage timestamp.

[0025] Specifically, once the real-time sleep stage transitions into the wakefulness stage, a timer begins. When this duration reaches a preset wake-up window (e.g., 10 minutes), the timestamp of the wakefulness stage is extracted and compared with the user's preset wakefulness time (e.g., the user's usual wake-up time of 6:00-7:00). The coverage ratio of the wakefulness stage timestamp to the preset wakefulness time is calculated. If the coverage meets the preset requirements (e.g., coverage of over 80%), a forced switch to the user's sleep environment is triggered. For example, the bedroom lights are gradually adjusted from low-brightness warm light at night to high-brightness white light simulating natural light, and the temperature is adjusted from the suitable 22℃ for sleep to 25℃ for daily activities. At the same time, white noise playback is turned off and replaced with gentle wake-up music. If the coverage does not meet the preset requirements, the current environment adjustment mode is maintained, and the sleep stage changes are monitored cyclically within the environment adjustment window until the forced switch conditions are met or the user re-enters the non-wakefulness sleep stage. This process combines the monitoring priority configuration of the waking phase in the hierarchical scheduling architecture with the judgment of the coverage scale of the preset waking period to ensure that users are naturally awakened at a time that conforms to their own work and rest habits, and avoids the impact of early or late environmental changes on sleep quality or wake-up state.

[0026] In one possible implementation, step S100 further includes: Step S110: Analyze the physiological data sequence uploaded by the wearable device to determine the user's status and output real-time status information.

[0027] Step S120: When the real-time status information is in a sleep state, initialize the wearable device and activate the hierarchical scheduling architecture.

[0028] Step S130: Determine the sleep stage based on the real-time physiological data collected by the wearable device after initialization, and obtain the real-time sleep stage.

[0029] Step S140: The hierarchical scheduling architecture allocates the priority of functional units according to the real-time sleep stage to obtain the scheduling sequence of the functional units.

[0030] Specifically, advanced signal processing technology and a meticulously designed data analysis model are used to conduct a detailed analysis of the physiological data sequences uploaded by wearable devices, thereby accurately determining the user's state and outputting real-time status information. The received heart rate data is filtered to remove noise interference, and then spectral analysis is used to extract heart rate variability features. For respiratory rate data, an adaptive threshold algorithm is used to accurately identify the respiratory cycle. For body movement data, pattern recognition algorithms are used to analyze patterns such as amplitude, frequency, and duration of body movements. By inputting this processed and feature-extracted data into a deep learning-based user state classification model, which has been trained on a large amount of sample data, this model can accurately distinguish different user states, such as awake, active, relaxed, or about to fall asleep, and thus output real-time status information, providing a reliable basis for subsequent operations.

[0031] The initialization process for the wearable device is initiated rapidly. This includes precise calibration of the device's sensors to ensure accurate measurement of physiological data; clearing the device's data cache to free up sufficient storage space for newly acquired data; and resetting key parameters such as sampling frequency and data transmission protocol to put the device in optimal working condition and guarantee the highest accuracy of subsequent data acquisition. Simultaneously, the hierarchical scheduling architecture is activated, preparing it to perform task scheduling and resource allocation operations based on subsequently acquired data.

[0032] By continuously collecting real-time physiological data using an initialized wearable device, this data reflects the dynamic physiological changes of the user during sleep. This real-time physiological data is then input into a specially constructed sleep stage determination model. This model was developed through in-depth analysis of a large amount of sleep experimental data and trained using machine learning algorithms. The model employs multivariate time series analysis, comprehensively considering the changes in multiple physiological variables such as heart rate, respiratory rate, and body movement over time. It also incorporates a hidden Markov model to model the potential states of sleep stages, thereby accurately identifying different sleep stages based on the characteristics of real-time physiological data, such as light sleep, deep sleep, and REM sleep. This crucial information, the real-time sleep stage, will provide important guidance for the subsequent precise regulation of the sleep environment.

[0033] The hierarchical scheduling architecture prioritizes functional units based on the newly determined real-time sleep stages. For different sleep stages, each functional unit (e.g., physiological data acquisition unit, environmental monitoring unit, etc.) is comprehensively evaluated to determine its importance in the current stage. For example, during light sleep, changes in environmental noise and light are more likely to disrupt sleep, so resource allocation is relatively more focused on environmental noise and light monitoring units. During deep sleep, maintaining the stability and accuracy of physiological data acquisition is crucial for accurately monitoring the user's sleep state, so physiological data acquisition units receive more attention. Based on these evaluation results, a dynamic programming algorithm is used to assign a corresponding priority to each functional unit, generating a functional unit scheduling sequence. This sequence guides subsequent data acquisition and environmental adjustments in an orderly manner, ensuring the efficient and accurate operation of the entire sleep environment adaptive control system.

[0034] In one possible implementation, step S100 further includes: Step S150: Interact to obtain multiple sets of sleep-related indicators for various sleep stages, wherein the multiple sets of sleep-related indicators have multiple sets of monitoring task priority identifiers; interactively obtain multiple historical transition information for the various sleep stages; solve for the stage transition probability based on the multiple historical transition information to obtain a stage transition correlation table; configure multiple monitoring functional units for the various sleep stages; after configuring multiple sets of physiological monitoring tasks in the multiple monitoring functional units based on the multiple sets of sleep-related indicators, configure the priority of the multiple sets of physiological monitoring tasks based on the priority of the multiple sets of monitoring tasks; load the stage transition correlation table into the functional unit allocation layer; connect the multiple monitoring functional units in parallel to obtain the functional unit scheduling layer; complete the construction of a hierarchical scheduling architecture by cascading the functional unit allocation layer and the functional unit scheduling layer.

[0035] Specifically, by interacting with professional sleep research databases, clinical data from medical institutions, and a large amount of actual sleep monitoring data from users, multiple sets of sleep-related indicators for various sleep stages (including light sleep, deep sleep, and REM sleep) are obtained, and these indicators are accompanied by multiple sets of monitoring task priority labels. For example, for the light sleep stage, the related indicators include heart rate variability within a certain range, a relatively fast and stable respiratory rate, and a relatively high number of body movements. Among these, the heart rate monitoring task is marked as high priority because heart rate changes during light sleep are crucial for judging sleep stability. For the deep sleep stage, the related indicators are enhanced slow-wave brain activity, increased and regular breathing depth, and almost no body movement. At this time, the priority of respiratory monitoring will be significantly increased.

[0036] Multiple historical transition information for various sleep stages is obtained through interaction with relevant data sources. This information records in detail the transitions between different sleep stages, such as the specific number of times, the timing, and the physiological state and environmental factors at the time of transitioning from light sleep to deep sleep, from deep sleep to REM sleep, or from REM sleep back to light sleep.

[0037] In constructing a hierarchical scheduling architecture, a crucial step is to calculate the stage switching probability based on multiple historical transition information to obtain a stage switching correlation table. First, detailed information on transitions between various sleep stages (such as light sleep, deep sleep, and REM sleep) is carefully selected from massive amounts of historical sleep data. This information includes multi-dimensional data such as the specific time each user transitions from one stage to another during different sleep periods, their physiological state characteristics at the time of transition, and the environmental conditions at that time. Then, advanced probabilistic statistical analysis methods are applied to this rich data. A Bayesian network model is used, treating sleep stages as nodes and transitions between stages as directed edges. By calculating the conditional probability of transitioning to the next sleep stage given the previous sleep stage, a probabilistic relationship model between stages is constructed. Next, a two-dimensional table of stage switching correlation numbers is constructed with different sleep stages as the horizontal and vertical axes. The calculated probability values ​​for transitioning from each sleep stage (stage A) to other sleep stages (stage B) are accurately filled into the corresponding table cells. In this way, during subsequent sleep monitoring and regulation, the probability of switching between different sleep stages can be quickly found based on this table, thereby predicting the changing trend of the user's sleep stages in advance. This provides a key basis for rationally allocating monitoring resources, optimizing the scheduling of functional units, and accurately adjusting the sleep environment, ensuring the efficient operation of the entire sleep environment adaptive control system.

[0038] Multiple monitoring units are configured for different sleep stages, such as light sleep, deep sleep, and REM sleep. For example, a high-precision heart rate monitoring unit, body movement sensing unit, and environmental noise monitoring unit are configured for the light sleep stage; a more sensitive breathing monitoring unit, EEG monitoring unit (if applicable), and environmental temperature monitoring unit are provided for the deep sleep stage; and a dedicated eye movement monitoring unit, muscle tone monitoring unit, and environmental light monitoring unit are set up for the REM sleep stage.

[0039] After configuring the sleep-related indicator acquisition and monitoring units, the crucial physiological monitoring task configuration and priority setting phase begins. Based on the previously acquired multiple sets of sleep-related indicators, which accurately reflect the intrinsic connection between the human physiological state and the sleep environment at different sleep stages (light sleep, deep sleep, and REM sleep), multiple sets of physiological monitoring tasks are carefully configured on the corresponding monitoring units. For example, for the light sleep stage, given the relatively high number of body movements and active heart rate variability among its related indicators, a high-frequency body movement amplitude and frequency monitoring task is configured in the body movement sensing monitoring unit, while the heart rate monitoring unit focuses on real-time heart rate data acquisition and heart rate variability analysis. For the deep sleep stage, based on its EEG activity characteristics, respiratory depth, and regularity, a high-precision slow-wave EEG activity monitoring task is set in the EEG monitoring unit, while the respiratory monitoring unit conducts deep respiratory monitoring and respiratory rhythm stability assessment tasks. After completing these targeted task configurations, the multiple sets of physiological monitoring tasks within each monitoring unit are prioritized according to the priority identifiers of the multiple monitoring tasks associated with each set of sleep-related indicators. Taking the light sleep stage as an example, if sleep-related indicators show that body movement is important for judging sleep quality and stage transitions, then the body movement amplitude and frequency monitoring tasks in the body movement sensing monitoring unit will be given the highest priority, while other auxiliary monitoring tasks will be ranked according to their importance. For the deep sleep stage, if the stability of breathing depth and rhythm is a key correlation indicator, the deep breathing monitoring and rhythm stability assessment tasks in the breathing monitoring unit will become the highest priority tasks. This ensures that the system prioritizes the collection and processing of the most critical data for judging sleep state within limited resources and time, providing the most valuable information support for subsequent accurate sleep stage judgment and environmental regulation strategy formulation, thereby achieving the precise operation of the sleep environment adaptive control system.

[0040] The calculated stage transition correlation table is loaded into the functional unit allocation layer. This table is derived from in-depth analysis of the transition information between different sleep stages in a large amount of historical sleep data. It uses sleep stages as the horizontal and vertical axes, presenting the probability of switching from one stage to another. Once loaded into the functional unit allocation layer, it can quickly and intelligently determine how to allocate resources reasonably when the user's sleep stage changes, based on the probability information in the table. It clarifies the key areas that each monitoring functional unit should focus on during different stage transitions, ensuring that resource allocation is highly adapted to the dynamic changes in sleep stages.

[0041] Multiple monitoring units were operated in parallel to successfully establish a functional unit scheduling layer. These monitoring units encompass various types closely related to sleep monitoring, such as heart rate monitoring, respiration monitoring, body movement monitoring, and environmental monitoring (including temperature, noise, and light). Through parallel operation, each monitoring unit can work independently while simultaneously coordinating under unified scheduling commands. Each monitoring unit acts as an independent information acquisition channel; their parallel operation efficiently acquires physiological and environmental data from various sources, providing a rich data source for comprehensively understanding the user's sleep state and sleep environment. Furthermore, they possess the ability to quickly respond to system scheduling needs, ready to adjust their operating mode and data acquisition frequency according to system instructions.

[0042] A hierarchical scheduling architecture was successfully constructed through a cascaded functional unit allocation layer and a functional unit scheduling layer. In this architecture, the functional unit allocation layer is at the upper level, providing macro-level guidance and resource allocation to the lower functional unit scheduling layer based on the probability information in the stage switching correlation table. The functional unit scheduling layer, on the other hand, is at the lower level, receiving scheduling instructions from the upper level and directly controlling the specific operation of each parallel monitoring functional unit. The two work closely together to form an organic whole. When a user's sleep state changes, the functional unit allocation layer predicts possible stage transitions based on the stage switching correlation table and adjusts its resource allocation strategy for the functional unit scheduling layer. The functional unit scheduling layer, in turn, adjusts the data acquisition task priority, working frequency, and other parameters of the parallel monitoring functional units in a timely manner according to instructions from the upper level, ensuring dynamic adaptation to changes in sleep stages. This enables precise monitoring and regulation of the sleep environment and the user's physiological state, creating the optimal sleep environment for the user.

[0043] In one possible implementation, step S130 further includes: Step S131: Using the multiple sets of sleep-related indicators as constraints, collect multiple sample sleep-related datasets.

[0044] Step S132: Construct a sleep stage judgment model using the multiple sample sleep association datasets, wherein the sleep stage judgment model includes multiple sleep stage judgment branches in parallel, and the input end of the multiple sleep stage judgment branches is configured with multiple sleep data filtering engines constructed based on the multiple sets of sleep association indicators.

[0045] Step S133: Align the data collected by the initialized wearable device to acquire the real-time physiological data.

[0046] Step S134: Analyze the real-time physiological data through the sleep stage judgment model and output the real-time sleep stage.

[0047] Specifically, multiple sets of sleep-related indicators were used as strict constraints to collect sample sleep-related datasets. These sleep-related indicators were derived through in-depth analysis of a large amount of sleep research and actual monitoring data, and are closely related to different sleep stages (such as light sleep, deep sleep, REM sleep, etc.), covering specific ranges or patterns of change of physiological parameters (such as heart rate, respiratory rate, and electroencephalogram activity) and environmental factors (such as temperature, noise, and light). Based on these indicators, data were accurately collected from multiple data sources (such as past sleep monitoring records and clinical sleep research data) to obtain multiple sample sleep-related datasets. Each dataset contains a series of physiological and environmental data samples under the constraints of specific sleep-related indicators, providing rich and targeted basic data for subsequent model construction.

[0048] When constructing a sleep stage judgment model using multiple sleep association datasets, a decision tree algorithm is chosen. First, comprehensive data preprocessing is performed on the collected sleep association datasets. Missing values ​​in the datasets are carefully handled; for example, for a small number of missing heart rate values, the average heart rate of the sample within a specific time period is used to fill in the gaps. For outliers, the mean and standard deviation of the data are calculated, and data exceeding a certain multiple of the standard deviation range are considered outliers and corrected or deleted. Simultaneously, all data are normalized, mapping physiological data such as heart rate and respiratory rate, as well as data such as ambient temperature and noise levels, to specific intervals to ensure the numerical comparability of various features. Based on multiple sets of sleep association indicators, features crucial for sleep stage judgment are selected, such as specific statistical indicators of heart rate variability, patterns of respiratory depth changes, frequency of body movement within a specific duration, and changes in the decibel range of ambient noise. These features are used as splitting attributes for constructing the decision tree. For each sleep stage—light sleep, deep sleep, and REM sleep—an independent sleep stage judgment branch is constructed. Taking the light sleep stage branch as an example, the decision tree begins at its root node and splits based on heart rate variability characteristics. If the heart rate variability falls within a specific range corresponding to the light sleep stage, the data continues along that branch, further judging based on respiratory depth change patterns at the next node; otherwise, the data flows to other branches. At each split node, a scientifically reasonable threshold is set by deeply analyzing the distribution of corresponding features in the sample data. For example, for the respiratory depth change pattern judgment node, a highly discriminative threshold is determined based on the statistical regularity of respiratory depth changes in light sleep stage samples, ensuring that the data can be accurately divided into different sub-nodes. Simultaneously, a sleep data filtering engine is built at the input end of each sleep stage judgment branch. Based on the feature ranges and change patterns of each sleep stage determined by multiple sets of sleep-related indicators, the input data is efficiently filtered. For instance, in the light sleep stage filtering engine, if the heart rate variability of the input data significantly deviates from the typical range of the light sleep stage, the data will be marked as low priority or directly filtered out, avoiding subsequent complex light sleep stage judgment branch calculations, thereby optimizing computational resource allocation and improving overall judgment efficiency. After the model was built, a subset of sample data was used as a validation set to rigorously evaluate the decision tree-based sleep stage judgment model. Key metrics such as accuracy and recall were calculated to carefully observe the model's accuracy in judging different sleep stages. If biases were found in the model's judgment of certain sleep stages, such as lower accuracy in REM sleep, the depth of the decision tree was adjusted to prevent overfitting or underfitting; feature selection was re-examined, considering whether to introduce other more representative features; and the threshold settings of each split node were optimized to enable the model to more accurately capture the feature differences of sleep stages.After repeated optimization and adjustments, a high-performance model was finally obtained that can accurately determine sleep stages based on real-time physiological data, providing a solid and reliable basis for subsequent adaptive control of the sleep environment.

[0049] Data alignment is performed using an initialized wearable device to acquire real-time physiological data. While the initialized device can collect user physiological data stably and accurately, various factors (such as sensor latency and data transmission time differences) may cause some deviation in the timeline of the collected data. Advanced data alignment algorithms are used to synchronize and calibrate real-time physiological data from different sensors (such as heart rate and respiration sensors), ensuring data consistency and accuracy over time. This results in high-quality real-time physiological data, providing reliable data input for accurate sleep stage assessment.

[0050] When real-time physiological data is transmitted to the sleep stage determination model, it first enters the root node of the decision tree. From there, based on pre-selected key features (such as heart rate variability, respiratory depth variation patterns, etc.) and thresholds set during the construction of the decision tree, the data is branched into different paths. For example, if the heart rate variability in the real-time physiological data falls within a specific range, the data will continue to be transmitted down the corresponding branch, further determined based on features set by the next layer of nodes (such as respiratory rate characteristics). At each node, the decision tree continuously refines the determination range of the sleep stage to which the data belongs by comparing the real-time physiological data with the node's splitting conditions. In this process, each sleep stage determination branch in the decision tree model (for light sleep, deep sleep, REM sleep, etc.) works independently. For the light sleep stage determination branch, as the data flows along the branch, the algorithm within the branch continuously evaluates the fit between the data and the characteristics of the light sleep stage. By comprehensively judging a series of physiological data features (such as heart rate fluctuations within a certain range, relatively frequent but small body movements, etc.), it calculates the probability or score of the data belonging to the light sleep stage. Similarly, the deep sleep and REM sleep stages are each analyzed based on their unique combination of features and judgment logic. Ultimately, the decision tree model accurately outputs the user's current real-time sleep stage based on the judgment results of each branch, combined with preset decision rules (such as selecting the sleep stage with the highest probability or score from multiple branch results as the final output, or using weighted voting to comprehensively consider the influence of each branch's judgment). This accurate judgment result will serve as a crucial basis to guide the subsequent sleep environment adaptive control system in making targeted dynamic adjustments to environmental factors (such as temperature, light, and noise) to optimize the user's sleep environment to the greatest extent and improve sleep quality.

[0051] In one possible implementation, step S140 further includes: Step S141: Using the real-time sleep stage, traverse the stage switching association table in the functional unit allocation layer of the hierarchical scheduling architecture to obtain the stage switching decision tree.

[0052] Step S142: Perform multi-level sleep stage retention on the stage switching decision tree according to the stage switching probability to obtain the sleep stage cycle sequence.

[0053] Step S143: The functional unit scheduling layer of the hierarchical scheduling architecture receives and allocates the priority of the multiple monitoring functional units according to the sleep stage cycle sequence to obtain the functional unit scheduling sequence.

[0054] Specifically, once the real-time sleep stage is determined, this information is used to traverse the previously constructed stage switching association table within the functional unit allocation layer of the hierarchical scheduling architecture. The stage switching association table uses the sleep stage as the horizontal and vertical axes, clearly presenting the probability of switching from one sleep stage to another. Using the real-time sleep stage as a key index, all related switching probability information is quickly located and extracted from the table, thus constructing a stage switching decision tree. For example, if the current real-time sleep stage is light sleep, starting from light sleep, the system obtains the probability of switching to deep sleep, REM sleep, and other possible stages, and uses this as the basis to construct the nodes and branches of the decision tree. The root node of the decision tree is the current real-time sleep stage (light sleep), while the branches represent the possible next sleep stage and their corresponding probabilities. The extension and expansion of these branches form a complete decision framework to guide subsequent scheduling decisions.

[0055] Based on the stage switching probability, a multi-level sleep stage retention operation is performed on the stage switching decision tree to obtain a sleep stage cyclic sequence. Starting from the root node of the decision tree, each branch is evaluated sequentially according to its probability. Sleep stages pointed to by branches with higher probabilities are retained first, and the next level of sleep stages is explored along these branches. For example, if the probability of switching from light sleep to deep sleep is high, then deep sleep will be retained as the next possible sleep stage. Then, starting from deep sleep, the above probability evaluation and sleep stage retention process is repeated to continue exploring subsequent possible sleep stage changes, such as the probability of switching from deep sleep to REM sleep. Through this progressive, probability-based screening and retention mechanism, a cyclic sequence containing multiple sleep stages is constructed. This cyclic sequence reflects a series of sleep stage change paths that the user is most likely to experience under the current real-time sleep stage, providing a forward-looking basis for the subsequent priority allocation of functional units.

[0056] The hierarchical scheduling architecture's functional unit scheduling layer receives the sleep stage cycle sequence and prioritizes multiple monitoring functional units based on this sequence, thus obtaining a functional unit scheduling sequence. For each sleep stage in the cycle sequence, the functional unit scheduling layer assigns different priorities to the associated monitoring functional units according to pre-defined rules and strategies. For example, during light sleep, changes in environmental noise and light have a significant impact on sleep; therefore, functional units related to environmental monitoring (such as noise monitoring units and light monitoring units) will be given higher priority to promptly capture changes in environmental factors and make corresponding adjustments. During deep sleep, the stability monitoring of physiological data is more critical, and the priority of functional units such as heart rate and respiration monitoring will be increased accordingly. In this way, the functional unit scheduling layer continuously adjusts the working priority of monitoring functional units based on the dynamic changes in the sleep stage cycle sequence, ensuring that the most critical monitoring tasks are focused on in each sleep stage. This generates the most suitable functional unit scheduling sequence for the current sleep state, achieving precise monitoring and regulation of the sleep environment and the user's physiological state, ensuring that the user is always in an optimal sleep environment.

[0057] In one possible implementation, step S300 further includes: Step S310: Interact to obtain multiple sample sleep quantization parameter sets of the multiple sample sleep association datasets.

[0058] Step S320: Perform multivariate regression analysis on the multiple sample sleep association datasets and multiple sample sleep quantification parameter sets to obtain multiple sleep quantification functions.

[0059] Step S330: Associate and store the various sleep stages and multiple sleep quantification functions to complete the construction of the sleep quantification model.

[0060] Step S340: After obtaining the real-time sleep quantization function from the sleep quantization model using the real-time sleep stage as the function call constraint, the real-time sleep associated data is synchronized to the real-time sleep quantization function to obtain the real-time sleep quantization parameters.

[0061] Step S350: Using the real-time sleep stage as the environmental regulation constraint, after scheduling the associated environmental regulation table, the associated environmental regulation table is traversed using the real-time sleep quantification parameters, and the sleep environment regulation strategy is output.

[0062] Specifically, by interacting with professional sleep research databases, a large number of past sleep monitoring records, and clinical sleep experimental data, multiple sample sleep quantitative parameter sets were obtained from multiple sample sleep association datasets. These sample sleep association datasets cover the numerical changes of various physiological indicators (such as heart rate, respiratory rate, and electroencephalogram activity) and environmental factors (such as temperature, humidity, and noise) at different sleep stages. The corresponding sample sleep quantitative parameter sets are the sets of parameters obtained after quantifying these data; for example, heart rate variability is quantified into a specific numerical range, and environmental noise intensity is quantified into decibel levels. These quantitative parameters can more intuitively reflect the relationship between sleep state and environmental factors, providing fundamental data support for subsequent model construction.

[0063] Performing multiple regression analysis on multiple sleep-related datasets and multiple sets of sleep quantification parameters is a crucial step in constructing a sleep quantification model. The multiple sleep-related datasets contain a wealth of diverse information, such as the range of heart rate changes, respiratory rate patterns, body movement, and environmental factors like temperature, humidity, and noise levels for different individuals at different sleep stages. The multiple sets of sleep quantification parameters are the result of quantifying these sleep-related data, such as quantifying heart rate changes into specific numerical values, representing respiratory rate stability with specific numerical values, and quantifying environmental noise into decibel levels. Next, these sleep-related datasets are used as the independent variable matrix, and the sleep quantification parameter sets are used as the dependent variable vector, employing multiple regression analysis. This method aims to find the optimal linear or nonlinear relationship model between the independent and dependent variables. During the analysis, the influence of each independent variable (i.e., various sleep-related data characteristics) on the dependent variable (sleep quantification parameters) is calculated, and the corresponding regression coefficients are determined. For example, calculations revealed that heart rate variability has a significant positive impact on sleep stability, a key parameter in sleep quantification, with a positive and relatively large regression coefficient. Conversely, environmental noise has a significant negative impact on sleep depth quantification within a certain range, with a negative regression coefficient. Based on these results, sleep quantification functions were constructed for each sleep stage (e.g., light sleep, deep sleep, REM sleep). Taking light sleep as an example, the constructed sleep quantification function might be: Sleep Quantification Parameter (Light Sleep) = a × Heart Rate Variability (Light Sleep) + b × Respiratory Rate (Light Sleep) + c × Environmental Noise (Light Sleep) + ... + d, where a, b, c, etc., are regression coefficients determined through multiple regression analysis, and d is a constant term. This function can calculate the corresponding sleep quantification parameter values ​​for the light sleep stage based on input real-time sleep-related data (e.g., real-time heart rate variability, real-time respiratory rate, real-time environmental noise), thus providing a quantitative basis for accurately assessing sleep quality and environmental requirements during the light sleep stage. By repeating the above process for each sleep stage, multiple sleep quantification functions applicable to different sleep stages are obtained, laying a solid foundation for building a complete sleep quantification model and further promoting the precise operation of the sleep environment adaptive control system.

[0064] By associating and storing multiple sleep stages with their corresponding sleep quantification functions, a sleep quantification model is constructed, storing a precise relationship model between physiological and environmental factors and quantification parameters under different sleep stages. When real-time sleep stage information is obtained, the corresponding sleep quantification function can be quickly retrieved from this model, preparing for subsequent real-time sleep quantification parameter calculations.

[0065] Then, using the real-time sleep stage as a function call constraint, the real-time sleep quantification function is precisely obtained from the sleep quantification model. Subsequently, the real-time collected sleep-related data is synchronized to this real-time sleep quantification function for calculation. The real-time sleep-related data includes the latest physiological and environmental information during the user's current sleep process, such as real-time heart rate, respiratory rate, current ambient temperature, and noise level. By substituting this data into the real-time sleep quantification function, real-time sleep quantification parameters are calculated. These parameters provide a precise quantitative description of the current sleep state, comprehensively considering the real-time status of physiological and environmental factors, providing crucial evidence for assessing current sleep quality and determining environmental adjustment needs.

[0066] Finally, using real-time sleep stages as environmental regulation constraints, a related environmental regulation table is scheduled. This table stores target values ​​for various environmental factors required to achieve the optimal sleep environment at different sleep stages. Then, the real-time sleep quantification parameters calculated earlier are used to traverse the related environmental regulation table. By comparing the quantification parameters with the target values ​​in the table, the direction and magnitude of adjustment for each environmental factor (such as temperature, humidity, light, noise, etc.) are determined. Based on this information, a precise sleep environment regulation strategy is output, such as adjusting the air conditioner temperature to a specific degree, controlling the light brightness to a suitable level, and activating noise cancellation equipment to reduce environmental noise. This achieves precise dynamic regulation of the sleep environment, providing users with the most suitable sleep environment and improving sleep quality.

[0067] In one possible implementation, step S110 further includes: Step S111: Interact to obtain multiple sample physiological datasets of various sample physiological states.

[0068] Step S112: Perform physiological state correlation analysis on the multiple sample physiological datasets to obtain multiple sample correlation index sets.

[0069] Step S113: After preprocessing the multiple sample physiological datasets using the multiple sample association index sets to obtain multiple sample update datasets, the multiple sample update datasets are used as training data to construct and obtain multiple physiological state judgment branches.

[0070] Step S114: Construct multiple data filtering engines based on the multiple sample association index sets, and load the multiple data filtering engines at the front end of the multiple physiological state judgment branches.

[0071] Step S115: Connect the multiple physiological state judgment branches in parallel to complete the construction of the user state judgment model.

[0072] Step S116: Analyze the physiological data sequence uploaded by the wearable device through the user status judgment model, and output the real-time status information.

[0073] Specifically, by interacting with multi-source data, multiple sample physiological datasets of various physiological states are obtained. These data sources include large-scale sleep research project databases, physiological data from different populations collected by professional medical institutions, and physiological data recorded by numerous volunteers wearing monitoring devices in their daily lives (covering various states such as sleep, exercise, and rest). For example, in the sleep state, the dataset includes data such as the subjects' heart rate, respiratory rate, and body movement; in the exercise state, it records information such as exercise intensity, heart rate variation range, and amount of sweating; in the resting state, there are also corresponding data such as heart rate and respiratory stability. This rich and diverse sample physiological dataset provides a sufficient data foundation for subsequent accurate analysis.

[0074] The collected massive physiological datasets encompass a wide range of physiological data from different individuals under various physiological states (such as sleep, movement, and rest), including but not limited to changes in indicators such as heart rate, respiratory rate, blood pressure, body movement amplitude, and skin conductance. For the sleep data, in-depth analysis revealed that heart rate fluctuations throughout the night exhibited different characteristics in different sleep stages (such as light sleep, deep sleep, and REM sleep). For instance, heart rate remained relatively stable and at a lower level during deep sleep, a stability characteristic that serves as an important correlation indicator. Similarly, respiratory rate also exhibits specific patterns during sleep, with its depth and rhythm closely related to sleep stages. The stability and specific range of variation in respiratory rate are also key correlation indicators. Furthermore, body movement also displays typical characteristics during sleep, with relatively more movement during light sleep and very little movement during deep sleep. The frequency and amplitude of body movement changes provide important references for assessing sleep state. In the sample data of exercise states, there is a significant positive correlation between heart rate and exercise intensity. The higher the exercise intensity, the faster the heart rate rises and the greater the fluctuation range. The sensitivity of heart rate to changes in exercise intensity and the change pattern under different types of exercise (such as aerobic and anaerobic exercise) become important correlation indicators. Simultaneously, respiratory rate increases and deepens during exercise. Its matching relationship with exercise intensity and type, such as the rapid rise and depth of respiratory rate during strenuous exercise, is also a key correlation indicator. Furthermore, skin conductance changes during exercise due to factors such as sweating, and the correlation between its amplitude and rate of change and exercise intensity and duration is also included in the analysis. For the resting state, heart rate and respiratory rate are relatively stable within a narrow range. The degree of stability and its relationship with the individual's baseline physiological state become important characteristics. Through comprehensive and in-depth analysis of various physiological indicators under different physiological states, representative features with high numerical stability under specific physiological states are extracted, and these features are combined to form multiple sample correlation indicator sets.

[0075] Multiple sample physiological datasets are preprocessed using multiple sets of sample association indicators. For each sample physiological dataset, data cleaning, filtering, and feature enhancement operations are performed based on its corresponding sample association indicator set. For example, if a certain association indicator shows the stable range of heart rate during sleep, then during the preprocessing of the sleep sample physiological dataset, data points with abnormal heart rate fluctuations are removed, and data that conform to the characteristics of sleep state are weighted to highlight key features. After preprocessing, multiple sample update datasets are obtained. These datasets are more targeted and representative, and are then used as training data to construct multiple physiological state judgment branches. Each physiological state judgment branch focuses on identifying a specific physiological state, such as a branch specifically for judging sleep state, a branch for judging movement state, and a branch for judging rest state. Each branch learns the feature patterns in the sample update dataset through a neural network algorithm, thereby gaining the ability to distinguish different physiological states.

[0076] The user status assessment model is constructed by parallelizing multiple physiological status assessment branches. This structure allows each branch to independently assess the data filtered by the data filtering engine, while also collaborating with each other to comprehensively evaluate the user's status.

[0077] Once the wearable device uploads a sequence of physiological data, the user state assessment model based on a neural network algorithm begins operation. This neural network model consists of multiple neuron layers, including an input layer, hidden layers, and an output layer. The input layer receives the physiological data sequence from the wearable device, which contains various features such as heart rate, respiratory rate, and body movement amplitude. The data first enters each neuron node in the input layer, with each node corresponding to a physiological data feature. The data is then passed from the input layer to the hidden layer. The neurons in the hidden layer perform deep feature extraction and pattern recognition on the input data through complex connections and weight allocation. During training, the model continuously adjusts the connection weights between neurons in the hidden layer based on a large sample physiological dataset to learn the complex relationships between physiological data features under different physiological states. For example, in the sleep state, the hidden layer neurons learn to recognize feature combinations such as heart rate variability within a certain range, slow and regular respiratory rate, and small and low body movement amplitude; in the exercise state, they learn feature patterns such as a rapid increase in heart rate, increased respiratory rate and depth, and large and frequent body movement amplitude. After processing by the hidden layer, the data finally reaches the output layer. The output layer typically has multiple nodes, each corresponding to a different possible user state (such as sleep, movement, or rest). Each output node calculates the probability or score of the data belonging to that state based on information passed from the hidden layer. For example, the sleep state node in the output layer comprehensively considers various feature information extracted from the hidden layer to calculate the probability value of the current physiological data sequence representing the sleep state. Finally, based on the probabilities or scores of each node in the output layer, the model determines the user's current real-time state information through a preset decision mechanism (such as selecting the state with the highest probability as the final judgment result, or setting a threshold, and determining that state when the probability of a certain state exceeds the threshold), thereby accurately determining whether the user is in a sleep, movement, or rest state, providing crucial information for subsequent system operations.

[0078] In one possible implementation, step S600 further includes: Step S610: If the duration of the real-time sleep stage entering the wakefulness stage meets the preset wake-up window, and the coverage scale of the wakefulness stage timestamp to the preset wakefulness period meets the preset coverage rate, then the adaptive closed-loop control of the user's sleep environment is terminated and sleep wake-up is initiated.

[0079] Specifically, when the duration of the transition from the real-time sleep stage to the wake-up stage reaches the preset wake-up window (e.g., 10 minutes), and the data collection and analysis by the functional units of the hierarchical scheduling architecture shows that the coverage ratio of the wake-up stage's timestamp to the user's preset wake-up time (e.g., 6:00-7:00 on weekdays) reaches the preset coverage rate (e.g., 80%), the system will terminate the previous adaptive closed-loop control process based on the dynamic adjustment of the environment according to the sleep stage, stop the dynamic adjustment of data collection priority by the functional unit scheduling sequence and the cyclical update of the sleep environment adjustment strategy, and instead start the sleep wake-up program: At this time, the environment adjustment module will perform a forced switch according to the preset wake-up scheme, such as gradually increasing the indoor light from the low-brightness warm light of the sleep mode to the high-brightness natural light simulating morning light, and raising the temperature from the suitable 22℃ for sleep to the 25℃ for daily activities. At the same time, the sleep-aiding white noise will be turned off and the preset gentle wake-up audio will be played. The wearable device will also stop high-frequency physiological data collection and only retain basic status monitoring to ensure that the user transitions naturally from the sleep state to the wake-up state, which is in line with their work and rest habits and improves the comfort of wake-up.

[0080] Example 2 is based on the same inventive concept as the sleep state adaptive hierarchical control method based on the environmental adjustment window in the previous examples, such as... Figure 2 As shown, this application provides a sleep state adaptive hierarchical control platform based on an environmental adjustment window. The platform and method embodiments in this application are based on the same inventive concept. The platform includes: The functional unit scheduling sequence generation module 10 is used to analyze the sleep state based on the physiological data sequence uploaded by the wearable device, obtain the real-time sleep stage, and then allocate scheduling tasks to the hierarchical scheduling architecture according to the real-time sleep stage to generate a functional unit scheduling sequence.

[0081] The real-time sleep-related data acquisition module 20 is used to manage the data acquisition priority of the wearable device using the functional unit scheduling sequence to obtain real-time sleep-related data.

[0082] The sleep environment regulation strategy generation module 30 is used to generate a sleep environment regulation strategy based on the real-time sleep stage as the environmental regulation constraint and the real-time sleep correlation data.

[0083] The environment adjustment window activation module 40 is used to activate the environment adjustment window after updating the user's sleep environment according to the sleep environment adjustment strategy.

[0084] The adaptive control update module 50 is used to run the hierarchical scheduling architecture with the environmental adjustment window as the time constraint and the functional unit scheduling sequence as the scheduling constraint, and to perform sleep stage judgment and adaptive control update of the user's sleep environment in a loop.

[0085] The forced switching update module 60 is used to perform a forced switching update of the user's sleep environment based on the coverage scale of the preset wake-up time period according to the wake-up time stamp after the duration of the real-time sleep stage entering the wake-up stage meets the preset wake-up window.

[0086] Furthermore, the functional unit scheduling sequence generation module 10 also includes: A real-time status information output unit is used to analyze the physiological data sequence uploaded by the wearable device to determine the user's status and output real-time status information.

[0087] A hierarchical scheduling architecture activation unit is used to initialize the wearable device and activate the hierarchical scheduling architecture when the real-time status information is in a sleep state.

[0088] A real-time sleep stage acquisition unit is used to determine the sleep stage based on the real-time physiological data collected by the wearable device after initialization, and to obtain the real-time sleep stage.

[0089] A functional unit scheduling sequence acquisition unit is used by the hierarchical scheduling architecture to allocate functional unit priorities according to the real-time sleep stage and obtain the functional unit scheduling sequence.

[0090] Furthermore, the functional unit scheduling sequence generation module 10 also includes: A multi-set sleep-related indicator acquisition unit is used to interactively obtain multiple sets of sleep-related indicators for various sleep stages, wherein the multiple sets of sleep-related indicators have multiple sets of monitoring task priority identifiers.

[0091] A historical transition information acquisition unit is used to interactively obtain multiple historical transition information of the various sleep stages.

[0092] A stage switching association table acquisition unit is used to solve the stage switching probability based on the multiple historical connection switching information to obtain a stage switching association table.

[0093] A monitoring function configuration unit is used to configure multiple monitoring function units for the multiple sleep stages.

[0094] A priority configuration unit is configured to configure the priority of the multiple sets of physiological monitoring tasks according to the priority of the multiple sets of monitoring tasks after configuring multiple sets of physiological monitoring tasks in the multiple monitoring function units according to the multiple sets of sleep-related indicators.

[0095] Functional unit allocation layer loading unit, which is used to load the stage switching association table to the functional unit allocation layer.

[0096] A functional unit scheduling layer acquisition unit is used to connect the multiple monitoring functional units in parallel to obtain the functional unit scheduling layer.

[0097] A hierarchical scheduling architecture construction unit is used to construct a hierarchical scheduling architecture by cascading the functional unit allocation layer and the functional unit scheduling layer.

[0098] Furthermore, the hierarchical scheduling architecture activation unit also includes: A sample sleep association dataset collection unit is used to collect multiple sample sleep association datasets by using the multiple sets of sleep association indicators as constraints.

[0099] A sleep data filtering engine construction unit is used to construct a sleep stage judgment model using the multiple sample sleep association datasets. The sleep stage judgment model includes multiple sleep stage judgment branches in parallel, and the input ends of the multiple sleep stage judgment branches are configured with multiple sleep data filtering engines constructed based on the multiple sets of sleep association indicators.

[0100] A real-time physiological data acquisition unit is used to perform data alignment through the initialized wearable device in order to acquire the real-time physiological data.

[0101] A real-time sleep stage output unit is used to analyze the real-time physiological data through the sleep stage judgment model and output the real-time sleep stage.

[0102] Furthermore, the functional unit scheduling sequence acquisition unit also includes: A stage switching decision tree acquisition unit is used to traverse the stage switching association table in the functional unit allocation layer of the hierarchical scheduling architecture using the real-time sleep stage to obtain the stage switching decision tree.

[0103] A sleep stage cycle sequence acquisition unit is used to retain sleep stages in a multi-level manner based on the stage switching decision tree according to the stage switching probability, thereby obtaining a sleep stage cycle sequence.

[0104] A functional unit scheduling sequence acquisition unit is used by the functional unit scheduling layer of the hierarchical scheduling architecture to receive and prioritize the multiple monitoring functional units according to the sleep stage cycle sequence, thereby obtaining the functional unit scheduling sequence.

[0105] Furthermore, the sleep environment regulation strategy generation module 30 also includes: A sample sleep quantization parameter set acquisition unit is used to interactively obtain multiple sample sleep quantization parameter sets from the multiple sample sleep associated datasets.

[0106] The sleep quantification function acquisition unit is used to perform multiple regression analysis on the multiple sample sleep association datasets and multiple sample sleep quantification parameter sets to obtain multiple sleep quantification functions.

[0107] A sleep quantification model construction unit is used to associate and store the various sleep stages and multiple sleep quantification functions to complete the construction of the sleep quantification model.

[0108] A real-time sleep quantization parameter acquisition unit is used to obtain a real-time sleep quantization function from the sleep quantization model by calling the real-time sleep stage as a function call constraint, and then synchronize the real-time sleep associated data to the real-time sleep quantization function to obtain real-time sleep quantization parameters.

[0109] A sleep environment regulation strategy output unit is used to use the real-time sleep stage as an environmental regulation constraint, schedule the associated environmental regulation table, traverse the associated environmental regulation table using the real-time sleep quantification parameters, and output the sleep environment regulation strategy.

[0110] Furthermore, the real-time status information output unit also includes: A sample physiological dataset acquisition unit is used to interactively obtain multiple sample physiological datasets for various sample physiological states.

[0111] A sample association index set acquisition unit is used to perform physiological state association analysis on the multiple sample physiological datasets to obtain multiple sample association index sets.

[0112] A physiological state judgment branch construction unit is used to construct multiple physiological state judgment branches by using the multiple sample association index sets to preprocess the multiple sample physiological datasets and obtain multiple sample updated datasets, and then using the multiple sample updated datasets as training data.

[0113] A data filtering engine construction unit is used to construct multiple data filtering engines based on the multiple sample association indicator sets, and load the multiple data filtering engines at the front end of the multiple physiological state judgment branches.

[0114] The user state judgment model construction unit is used to connect the multiple physiological state judgment branches in parallel to complete the construction of the user state judgment model.

[0115] A physiological data sequence judgment unit is used to analyze the physiological data sequence uploaded by the wearable device through the user status judgment model and output the real-time status information.

[0116] Furthermore, the forced switch update module 60 also includes: The sleep wake-up unit is configured to terminate the adaptive closed-loop control of the user's sleep environment and initiate sleep wake-up if the duration of the real-time sleep stage entering the wake-up stage meets a preset wake-up window and the coverage scale of the wake-up stage timestamp to the preset wake-up time period meets a preset coverage rate.

[0117] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0118] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0119] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A sleep state adaptive hierarchical control method based on an environmental adjustment window, characterized in that, The method includes: Sleep state analysis is performed based on the physiological data sequence uploaded by the wearable device. After obtaining the real-time sleep stage, the hierarchical scheduling architecture is used to allocate scheduling tasks according to the real-time sleep stage, and a functional unit scheduling sequence is generated. The functional unit scheduling sequence is used to manage the data acquisition priority of the wearable device to obtain real-time sleep-related data. Using the real-time sleep stage as an environmental regulation constraint, a sleep environment regulation strategy is generated based on the real-time sleep correlation data; After updating the user's sleep environment according to the aforementioned sleep environment adjustment strategy, the environment adjustment window is activated; Using the environmental adjustment window as a time constraint and the functional unit scheduling sequence as a scheduling constraint, the hierarchical scheduling architecture is run to cyclically perform sleep stage judgment and adaptive control update of the user's sleep environment. Once the duration of the real-time sleep phase transitioning into the wakefulness phase meets the preset wake-up window, the user's sleep environment is forcibly switched and updated based on the coverage scale of the preset wakefulness period according to the wakefulness phase timestamp.

2. The sleep state adaptive hierarchical control method based on environmental adjustment window as described in claim 1, characterized in that, The method involves analyzing sleep states based on physiological data sequences uploaded by wearable devices to obtain real-time sleep stages, allocating tasks to a hierarchical scheduling architecture based on these stages, and generating functional unit scheduling sequences. The user's status is determined by analyzing the physiological data sequence uploaded by the wearable device, and real-time status information is output. When the real-time status information indicates a sleep state, the wearable device is initialized and the hierarchical scheduling architecture is activated. The sleep stage is determined by analyzing the real-time physiological data collected by the wearable device after initialization. The hierarchical scheduling architecture allocates functional unit priorities based on the real-time sleep stage to obtain the functional unit scheduling sequence.

3. The sleep state adaptive hierarchical control method based on environmental adjustment window as described in claim 2, characterized in that, The method includes: Multiple sets of sleep-related indicators for various sleep stages are obtained interactively, wherein the multiple sets of sleep-related indicators have multiple sets of monitoring task priority identifiers; Interactively obtain multiple historical transition and switching information of the various sleep stages; Based on the multiple historical connection and switching information, the stage switching probability is solved to obtain a stage switching association table; Multiple monitoring function units are configured for the various sleep stages; After configuring multiple sets of physiological monitoring tasks in the multiple monitoring functional units based on the multiple sets of sleep-related indicators, the priority of the multiple sets of physiological monitoring tasks is configured according to the priority of the multiple sets of monitoring tasks. Load the stage switching association table into the functional unit allocation layer; The multiple monitoring functional units are connected in parallel to obtain a functional unit scheduling layer; By cascading the functional unit allocation layer and the functional unit scheduling layer, a hierarchical scheduling architecture is constructed.

4. The sleep state adaptive hierarchical control method based on environmental adjustment window as described in claim 3, characterized in that, The method involves determining the real-time sleep stage based on real-time physiological data collected by the initialized wearable device, and obtaining the real-time sleep stage. Multiple sleep association datasets were collected using the aforementioned multiple sets of sleep association indicators as constraints. A sleep stage judgment model is constructed using the multiple sample sleep association datasets. The sleep stage judgment model includes multiple sleep stage judgment branches connected in parallel. The input of the multiple sleep stage judgment branches is configured with multiple sleep data filtering engines constructed based on the multiple sets of sleep association indicators. The wearable device is initialized and aligned with the collected data to acquire the real-time physiological data. The real-time physiological data is analyzed by the sleep stage judgment model, and the real-time sleep stage is output.

5. The sleep state adaptive hierarchical control method based on environmental adjustment window as described in claim 4, characterized in that, The hierarchical scheduling architecture allocates functional unit priorities based on the real-time sleep stage to obtain the functional unit scheduling sequence, and the method includes: The real-time sleep stage is used to traverse the stage switching association table in the functional unit allocation layer of the hierarchical scheduling architecture to obtain the stage switching decision tree. Based on the stage switching probability, the stage switching decision tree is used to retain multiple sleep stages at different levels to obtain a sleep stage cycle sequence. The hierarchical scheduling architecture's functional unit scheduling layer receives and prioritizes the multiple monitoring functional units according to the sleep phase cycle sequence, thereby obtaining the functional unit scheduling sequence.

6. The sleep state adaptive hierarchical control method based on environmental adjustment window as described in claim 5, characterized in that, Using the real-time sleep stage as an environmental regulation constraint, and generating a sleep environment regulation strategy based on the real-time sleep correlation data, the method includes: Interactively obtain multiple sets of sample sleep quantization parameters from the multiple sample sleep association datasets; Multiple regression analysis was performed on the multiple sample sleep association datasets and multiple sample sleep quantification parameter sets to obtain multiple sleep quantification functions; The various sleep stages and multiple sleep quantification functions are associated and stored to complete the construction of the sleep quantification model; After obtaining the real-time sleep quantization function from the sleep quantization model using the real-time sleep stage as the function call constraint, the real-time sleep associated data is synchronized to the real-time sleep quantization function to obtain the real-time sleep quantization parameters. Using the real-time sleep stage as an environmental regulation constraint, after scheduling the associated environmental regulation table, the real-time sleep quantification parameters are used to traverse the associated environmental regulation table, and the sleep environment regulation strategy is output.

7. The sleep state adaptive hierarchical control method based on environmental adjustment window as described in claim 2, characterized in that, The method involves analyzing the physiological data sequence uploaded by the wearable device to determine the user's status and outputting real-time status information. Interactively obtain multiple sample physiological datasets with various sample physiological states; Physiological state correlation analysis was performed on the multiple sample physiological datasets to obtain multiple sample correlation index sets; After preprocessing the multiple sample physiological datasets using the multiple sample association index sets to obtain multiple sample updated datasets, the multiple sample updated datasets are used as training data to construct and obtain multiple physiological state judgment branches. Multiple data filtering engines are constructed based on the multiple sample association index sets, and the multiple data filtering engines are loaded at the front end of the multiple physiological state judgment branches; By connecting the multiple physiological state judgment branches in parallel, the user state judgment model is constructed. The user status judgment model analyzes the physiological data sequence uploaded by the wearable device and outputs the real-time status information.

8. The sleep state adaptive hierarchical control method based on environmental adjustment window as described in claim 1, characterized in that, If the duration of the real-time sleep phase transitioning to the wake-up phase meets the preset wake-up window, and the coverage scale of the wake-up phase timestamp to the preset wake-up time period meets the preset coverage rate, then the adaptive closed-loop control of the user's sleep environment is terminated and sleep wake-up is initiated.

9. A sleep state adaptive hierarchical control platform based on an environmental adjustment window, characterized in that, The platform is used to execute the sleep state adaptive hierarchical control method based on an environmental adjustment window as described in any one of claims 1-8, and the platform includes: The functional unit scheduling sequence generation module is used to analyze the sleep state based on the physiological data sequence uploaded by the wearable device, obtain the real-time sleep stage, allocate scheduling tasks to the hierarchical scheduling architecture based on the real-time sleep stage, and generate a functional unit scheduling sequence. A real-time sleep-related data acquisition module is used to manage the data acquisition priority of the wearable device using the functional unit scheduling sequence to obtain real-time sleep-related data. A sleep environment regulation strategy generation module is used to generate a sleep environment regulation strategy based on the real-time sleep stage as the environmental regulation constraint and the real-time sleep correlation data. An environment adjustment window activation module is used to activate the environment adjustment window after updating the user's sleep environment according to the sleep environment adjustment strategy. An adaptive control update module is used to run the hierarchical scheduling architecture with the environmental adjustment window as the time constraint and the functional unit scheduling sequence as the scheduling constraint, and to perform sleep stage judgment and adaptive control update of the user's sleep environment in a loop. The forced switch update module is used to force switch update the user's sleep environment according to the coverage scale of the preset wake-up time period based on the wake-up time stamp after the real-time sleep stage enters the wake-up stage for a duration that meets the preset wake-up window.

Images