Sleep position recognition and adaptive pillow height adjustment system, method and storage medium

Abstract

This application relates to a sleep posture recognition and adaptive pillow height adjustment system, method, and storage medium. The sleep posture recognition and adaptive pillow height adjustment system includes: a data acquisition module, a sleep posture recognition module, a pillow height adjustment module, and a user interaction module. The user interaction module acquires personalized prior data and historical sleep comfort scores from the user. The data acquisition module collects pressure datasets from a pressure sensor array in real time. The sleep posture recognition module constructs multiple time-domain data slices from the pressure dataset according to a preset time window, inputs them into a time-domain feature model, obtains multiple sets of low-dimensional pressure data feature parameters, and then identifies effective sleep postures. The pillow height adjustment module calculates and calibrates the pillow height based on the user's personalized prior data, effective sleep postures, and historical sleep comfort scores, and adjusts the pillow surface. This application solves the problems of high computational consumption and difficulty in meeting users' precise and personalized sleep needs in related technologies.

Description

Technical Field

[0001] This application relates to the field of intelligent sleep device technology, and in particular to a sleep posture recognition and adaptive pillow height adjustment system, method and storage medium. Background Technology

[0002] As public awareness of sleep health continues to rise, sleep aids such as smart pillows are gradually entering people's lives. Users are increasingly demanding more tailored sleeping postures and dynamically adjustable pillow heights. In practice, when users toss and turn at night, switching between supine and side-lying positions, a fixed pillow height can lead to insufficient neck support, potentially causing sleep interruptions and morning neck pain. Furthermore, users with different heights, weights, and other body types have significantly different basic pillow height requirements, making a single pillow height design insufficient to meet personalized sleep support needs. Current technologies often focus on sleep monitoring (such as heart rate and respiration monitoring), which suffers from high computational costs and a disconnect between the monitored data and users' core sleep needs, making it difficult to meet users' precise and personalized sleep requirements.

[0003] Currently, no effective solution has been proposed to address the issues of high computing power consumption and difficulty in meeting users' precise and personalized sleep needs in related technologies. Summary of the Invention

[0004] This application provides a sleeping posture recognition and adaptive pillow height adjustment system, method, and storage medium to at least solve the problems of high computing power consumption and difficulty in meeting users' precise and personalized sleep needs in related technologies.

[0005] In a first aspect, embodiments of this application provide a sleeping posture recognition and adaptive pillow height adjustment system, the system comprising: a data acquisition module, a sleeping posture recognition module, a pillow height adjustment module, and a user interaction module;

[0006] The user interaction module is used to obtain personalized prior data and historical sleep comfort scores of users;

[0007] The data acquisition module is used to acquire pressure datasets from a preset pressure sensor array in real time.

[0008] The sleeping posture recognition module is used to construct multiple sets of time-domain data slices from the pressure dataset according to a preset time window, and input the multiple sets of time-domain data slices into a preset time-domain feature model to obtain multiple sets of low-dimensional pressure data feature parameters; based on the multiple sets of low-dimensional pressure data feature parameters, the effective sleeping posture state is identified.

[0009] The pillow height adjustment module is used to obtain a calibrated pillow height based on the user's personalized prior data, the effective sleeping posture, and the historical sleep comfort score, and to adjust the pillow surface to the calibrated pillow height.

[0010] In some embodiments, the sleeping posture includes sleeping posture type and sleeping area; the sleeping posture type includes supine, lateral, and resting;

[0011] The sleeping posture recognition module is also used to obtain multiple sleeping posture states based on the multiple sets of low-dimensional pressure data feature parameters; and to determine the initially effective sleeping posture state based on the user's personalized prior data and the system's preset rules.

[0012] The sleeping posture recognition module is further configured to determine the effective sleeping posture state based on the sleeping posture type when the sleeping posture type in the multiple sleeping posture states is either supine or lateral.

[0013] The sleeping posture recognition module is also used to pause the determination of the current effective sleeping posture state and continuously acquire new pressure datasets when body movement occurs in the sleeping posture type among the multiple sleeping posture states; when the number of times the sleeping posture type corresponding to the new pressure dataset is determined to be supine or lateral reaches a first preset number, multiple sleeping posture states are reacquired, and the effective sleeping posture state is obtained based on the reacquired multiple sleeping posture states.

[0014] In some embodiments, the sleeping posture recognition module is further configured to count the number of times a new sleeping posture state that is different from the sleeping posture type or sleep area of ​​the initially effective valid sleeping posture state among a number of recently acquired second preset number of sleeping posture states; if the number of times the new sleeping posture state appears reaches a third preset number, then the new sleeping posture state is determined as a new effective sleeping posture state, and the initially effective valid sleeping posture state is updated to the new effective sleeping posture state.

[0015] In some embodiments, the user interaction module is further configured to obtain the number of days the user has used the service;

[0016] The pillow height adjustment module is also used to calculate the standard pillow height based on the user's personalized prior data and the effective sleeping posture.

[0017] The pillow height adjustment module is also used to adjust the standard pillow height based on the user's personalized prior data, the historical sleep comfort score and the number of days the user has used the pillow, to obtain the calibrated pillow height, and to adjust the pillow surface to the calibrated pillow height.

[0018] In some embodiments, the user-personalized prior data includes height, weight, and user pillow height preference;

[0019] The pillow height adjustment module is also used to obtain historical coefficients based on the user's pillow height preference, the historical sleep comfort score, and the number of days the user has used the device; and to obtain prior coefficients based on the user's pillow height preference.

[0020] The pillow height adjustment module is further configured to obtain historical feedback weights and prior data weights based on the number of days the user has used the product; and to adjust the standard pillow height based on the historical coefficients, the historical feedback weights, the prior coefficients, and the prior data weights to obtain a calibrated pillow height.

[0021] In some embodiments, the pillow height adjustment module is further configured to map the historical sleep comfort score to a base score coefficient based on the user's pillow height preference;

[0022] The pillow height adjustment module is also used to use the basic rating coefficient of the day as the historical coefficient when the number of days the user uses the device is less than the first preset number of days.

[0023] The pillow height adjustment module is also used to calculate the historical coefficient by weighting the basic rating coefficient of each day within the user's usage days according to a preset rule when the number of days the user uses the device is greater than the first preset number of days and less than the second preset number of days.

[0024] The pillow height adjustment module is also used to calculate the historical coefficient by weighting the basic rating coefficients of each day within the most recent second preset number of days according to the preset rules when the number of days the user uses the device is greater than the second preset number of days; wherein, the preset rules are rules for assigning corresponding weights based on the basic rating coefficients of different time periods.

[0025] In some embodiments, the pressure sensor array is a thin-film pressure sensor array; the pressure sensor array is arranged in M ​​columns along the length direction of the pillow surface and N rows along the width direction, forming an N×M array; where M and N are positive integers.

[0026] In some embodiments, the low-dimensional pressure data feature parameters include the centroid coordinates of the pressure distribution, the location of the pressure peak, and the standard deviation of the pressure value;

[0027] The sleeping posture recognition module is also used to input the feature parameters of the multiple sets of low-dimensional stress data into a pre-trained support vector machine algorithm model to obtain multiple sleeping posture states; the pre-trained support vector machine algorithm model is trained based on stress data of multiple users of different body types and genders in different sleeping postures during real sleep.

[0028] Secondly, embodiments of this application provide a method for sleep posture recognition and adaptive pillow height adjustment, the method comprising:

[0029] Obtain personalized prior data and historical sleep comfort scores from users;

[0030] The pressure dataset of a preset pressure sensor array is collected in real time. The pressure dataset is then used to construct multiple sets of time-domain data slices according to a preset time window. These multiple sets of time-domain data slices are then input into a preset time-domain feature model to obtain multiple sets of low-dimensional pressure data feature parameters.

[0031] Based on the feature parameters of the multiple sets of low-dimensional pressure data, the effective sleeping posture is identified.

[0032] Based on the user's personalized prior data, the effective sleeping posture, and the historical sleep comfort score, a calibrated pillow height is obtained, and the pillow surface is adjusted to the calibrated pillow height.

[0033] Thirdly, embodiments of this application provide a storage medium storing a computer program thereon, which, when executed by a processor, implements the operation of the sleeping posture recognition and adaptive pillow height adjustment system as described in the first aspect above.

[0034] Compared to related technologies, the sleep posture recognition and adaptive pillow height adjustment system, method, and storage medium provided in this application embodiment include: a data acquisition module, a sleep posture recognition module, a pillow height adjustment module, and a user interaction module; the user interaction module is used to acquire personalized prior data and historical sleep comfort scores of the user; the data acquisition module is used to acquire pressure datasets from a preset pressure sensor array in real time; the sleep posture recognition module is used to construct multiple sets of time-domain data slices from the pressure dataset according to a preset time window, and input the multiple sets of time-domain data slices into a preset time-domain feature model to obtain multiple sets of low-dimensional pressure data feature parameters; based on the multiple sets of low-dimensional pressure data feature parameters, an effective sleep posture state is identified; the pillow height adjustment module is used to obtain a calibrated pillow height based on the user's personalized prior data, effective sleep posture state, and historical sleep comfort scores, and adjust the pillow surface to the calibrated pillow height. This solves the problems of high computational consumption and difficulty in meeting users' precise and personalized sleep needs.

[0035] Details of one or more embodiments of this application are set forth in the following drawings and description to make other features, objects and advantages of this application more readily apparent. Attached Figure Description

[0036] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0037] Figure 1 This is a hardware structure block diagram of a terminal of a sleeping posture recognition and adaptive pillow height adjustment system according to an embodiment of this application;

[0038] Figure 2This is a structural block diagram of a sleeping posture recognition and adaptive pillow height adjustment system according to an embodiment of this application;

[0039] Figure 3 This is a pressure sensor distribution diagram according to an embodiment of this application;

[0040] Figure 4 This is a flowchart of a sleeping posture recognition and adaptive pillow height adjustment method according to an embodiment of this application. Detailed Implementation

[0041] To make the objectives, technical solutions, and advantages of this application clearer, the application is described and illustrated below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the application. All other embodiments obtained by those skilled in the art based on the embodiments provided in this application without inventive effort are within the scope of protection of this application. Furthermore, it is understood that although the efforts made in such a development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this application, modifications to design, manufacturing, or production based on the technical content disclosed in this application are merely conventional technical means and should not be construed as insufficient disclosure of the content of this application.

[0042] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that is mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0043] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms “a,” “an,” “an,” “the,” and similar words used in this application do not indicate quantity limitation and may indicate singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units not listed, or may include other steps or units inherent to these processes, methods, products, or devices. The terms “connected,” “linked,” “coupled,” and similar words used in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. “Multiple” used in this application means two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. The terms “first,” “second,” “third,” etc., used in this application are merely to distinguish similar objects and do not represent a specific ordering of the objects.

[0044] The method embodiments provided in this example can be executed on a terminal, computer, or similar computing device. Taking running on a terminal as an example, Figure 1 This is a hardware structure block diagram of a terminal for a sleeping posture recognition and adaptive pillow height adjustment system according to an embodiment of this application. Figure 1 As shown, a terminal may include one or more ( Figure 1 Only one is shown in the diagram. A processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data are also shown. Optionally, the terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the terminal described above. For example, the terminal may also include components that are more... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0045] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the sleep posture recognition and adaptive pillow height adjustment system in this embodiment. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, thereby implementing the above-described method. The memory 104 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0046] The transmission device 106 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the terminal's communication provider. In one example, the transmission device 106 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet. In another example, the transmission device 106 may be a Radio Frequency (RF) module used for wireless communication with the Internet.

[0047] With increasing awareness of sleep health, sleep aids such as smart pillows are becoming more widespread. Users are increasingly demanding "sleep posture adaptability" and "dynamic pillow height adjustment." For example, when users turn over at night to switch between lying on their back and side, a fixed pillow height can easily lead to insufficient neck support, causing sleep interruptions or morning neck pain. Furthermore, users of different body types (height, weight) have significantly different basic needs for pillow height, making a single pillow height design insufficient to meet individual needs. Currently, most mainstream smart pillow products in the industry focus on "sleep monitoring" (such as heart rate and respiration monitoring), while integrated solutions for "sleep posture recognition + adaptive pillow height adjustment" are still in their early stages.

[0048] Current related technologies are mainly divided into two categories:

[0049] The first category is sleep posture recognition solutions: The mainstream technical approaches include two types. One is to use low-density pressure sensors or single vibration sensors to collect data, and then identify sleep postures (such as distinguishing between supine and lateral sleeping) through simple threshold judgment or feature matching. For example, some products use 3-5 discrete pressure sensors to detect the shift in the head's center of gravity and determine the change in sleep posture. The second type involves introducing wearable devices (such as smart bracelets) or recording devices to assist in monitoring, combining heart rate, body movement, or snoring signals to indirectly infer sleep state. Some advanced solutions attempt to improve accuracy through deep learning algorithms, but these mostly rely on frequency domain analysis (such as Fourier transform) to process data, requiring significant computing power.

[0050] The second category is pillow height adjustment solutions: these use an "airbag + air pump" as the core actuator. The adjustment logic can be divided into two types: one is a manual adjustment mode, where users switch between preset high, medium, and low pillow heights via an app or physical buttons, lacking automatic adaptation capabilities; the other is a semi-automatic adjustment mode, based on simple sleeping posture rules (e.g., a low pillow for supine sleeping and a high pillow for side sleeping), without incorporating individual user characteristics. For example, a certain brand of smart pillow uses vibration sensors to detect body movement and raises or lowers the airbag height by a fixed amount, with the adjustment area mostly covering the entire pillow surface, unable to adapt to specific head / neck positions. A few solutions attempt to link sleep data with pillow height parameters, but lack a dynamic feedback mechanism, making it difficult to optimize the adjustment strategy based on long-term user habits.

[0051] The existing solutions described above have the following core problems, making it difficult to meet users' needs for "accurate identification + dynamic adaptation":

[0052] First, the accuracy of sleep posture recognition is low, the computing power consumption is high, and the function is limited: low-density sensors are prone to data blind spots, leading to misjudgment of sleep posture (such as misjudging side sleeping as supine sleeping); deep learning related algorithms consume a lot of computing power and cannot run for a long time on low-computing-power embedded devices such as STM32; and it can only identify the type of sleep posture, but cannot locate the specific area of ​​the head / neck on the pillow surface, resulting in a lack of targeted adjustment of pillow height.

[0053] Secondly, the pillow height adjustment has poor adaptability: it does not integrate users' prior data (height, weight, pillow height preference) with real-time sleep status (sleeping position, sleep area), and only relies on a single rule for adjustment, which cannot meet personalized needs (such as the difference in pillow height needs between users who are 180cm tall and those who are 160cm tall); and there is no historical feedback calibration mechanism, so it is impossible to optimize pillow height parameters based on users' long-term usage habits, which easily leads to the problem of "the algorithm is optimal but the user feels uncomfortable".

[0054] Finally, the anti-interference ability is weak: the existing solution does not distinguish between "effective sleeping position switching" and "accidental body movement" (such as unconscious head movements at night). Accidental body movements can easily trigger unnecessary pillow height adjustments, leading to sleep interruption and reducing user experience.

[0055] To address the aforementioned issues, this embodiment provides a sleeping posture recognition and adaptive pillow height adjustment system. Figure 2 This is a structural block diagram of a sleeping posture recognition and adaptive pillow height adjustment system according to an embodiment of this application, such as... Figure 2 As shown, the system includes: a data acquisition module 21, a sleeping posture recognition module 22, a pillow height adjustment module 23, and a user interaction module 24;

[0056] User interaction module 24 is used to obtain personalized prior data and historical sleep comfort scores of users;

[0057] Data acquisition module 21 is used to acquire pressure datasets from a preset pressure sensor array in real time;

[0058] The sleeping posture recognition module 22 is used to construct multiple sets of time-domain data slices from the pressure dataset according to a preset time window, and input the multiple sets of time-domain data slices into a preset time-domain feature model to obtain multiple sets of low-dimensional pressure data feature parameters; based on the multiple sets of low-dimensional pressure data feature parameters, the effective sleeping posture state is identified.

[0059] The pillow height adjustment module 23 is used to obtain a calibrated pillow height based on the user's personalized prior data, effective sleeping posture and historical sleep comfort score, and adjust the pillow surface to the calibrated pillow height.

[0060] The user interaction module 24 serves as the entry point for interaction between the system and the user. Its core function is to acquire personalized prior data and historical sleep comfort scores, providing a data foundation for subsequent personalized adaptation. Personalized prior data includes physiological parameters such as height (cm) and weight (kg), as well as sleep habit parameters such as sleeping posture preferences (e.g., "prefers side sleeping," "prefers supine sleeping") and pillow height preferences (e.g., "prefers high pillow," "prefers low pillow," "prefers normal height"). Users can input this data through a mobile app or mini-program. For existing users, the system will automatically load their historical data. Historical sleep comfort scores are obtained through a "proactive feedback + automatic completion" mechanism. After waking up, users can rate their sleep comfort on a scale of 1-5 via the app or touchscreen. If no rating is actively given, the system will automatically generate a score based on the sleep quality assessment model. The score data will be stored in association with sleeping posture and pillow height parameters for subsequent pillow height optimization.

[0061] The core of the data acquisition module 21 is a distributed pressure sensor array, whose function is to collect pressure data from the array in real time, providing raw sensor data for sleep posture recognition. This pressure sensor array is an array structure composed of thin-film sensors, with corresponding numbers of sensors deployed along the length and width of the smart pillow surface to form a sensor network covering the sleep support area of ​​the pillow surface. The module controls the sensors to collect pressure data at a fixed frequency (e.g., 10Hz) through embedded software (e.g., based on an STM32F407 microcontroller). During the acquisition process, the raw data undergoes preliminary filtering (e.g., using a moving average filtering algorithm) to eliminate instantaneous noise interference. The processed pressure dataset is then transmitted to the sleep posture recognition module via the SPI (Serial Peripheral Interface) protocol to ensure data continuity and accuracy.

[0062] The sleeping posture recognition module 22 is the core of the system's decision-making, achieving accurate identification of effective sleeping postures through multi-step processing. The module first receives the pressure dataset from the data acquisition module. After data preprocessing, it constructs multiple sets of time-domain data slices according to a preset time window, such as 10 seconds per slice, transforming continuous data into structured data that meets feature extraction requirements. Next, the multiple sets of time-domain data slices are input into a preset time-domain feature model to extract multiple sets of low-dimensional pressure data feature parameters. These low-dimensional pressure data feature parameters refer to feature parameters whose data dimension is lower than a preset dimension threshold. For example, high-dimensional raw data (such as a single 10-second slice containing 1500 data points) can be compressed into 20-40 feature parameters to significantly reduce computational power consumption. Subsequently, based on the multiple sets of low-dimensional pressure data feature parameters, the effective sleeping posture is identified. Finally, the effective sleeping posture information is transmitted to the pillow height adjustment module via UART (Universal Asynchronous Receiver / Transmitter) to provide a basis for judging pillow height adaptation. The effective sleeping posture is either supine or lateral. The data preprocessing process is as follows: receiving the pressure data output by the data acquisition module, firstly removing outliers (such as data exceeding the range caused by sensor failure) using the 3σ criterion, and then performing normalization to map the pressure value to the 0-1 range. The formula is: Normalized value = (original value - minimum value) / (maximum value - minimum value), eliminating errors caused by differences in the sensitivity of different sensors.

[0063] The pillow height adjustment module 23 is responsible for the dynamic adaptation of pillow height. Its core logic is to calculate and calibrate the pillow height based on multi-dimensional data and then complete the adjustment. The module first receives the user's personalized prior data, historical sleep comfort scores, and effective sleeping posture status output by the sleeping posture recognition module provided by the user interaction module 21. Then, it integrates these data through a preset algorithm to calculate the calibrated pillow height that matches the current user and current sleeping posture. Finally, it converts the calibrated pillow height parameters into control signals for the actuator, driving the "multi-zone electric push rod" inside the smart pillow to adjust the pillow surface height of the corresponding area to the calibrated pillow height. For example, if the user is lying on their side with their head in the left area, only the push rod in the left area is driven to rise to the target height, while the middle and right areas maintain an appropriate height (the height of the middle and right areas is finely adjusted to ensure that the neck curve is completely in contact with the pillow surface, and the adjustment of the pillow surface is a gradual process, without being too steep). This ensures that the neck curve is completely in contact with the pillow surface, achieving the adaptation of the pillow height to the user's sleep needs. After adjustment, the sleeping posture is continuously monitored. If the sleeping posture changes, the adjustment process is repeated, with an adjustment response time of ≤3 seconds.

[0064] The sleep posture recognition and adaptive pillow height adjustment system provided in the above embodiments achieves multi-dimensional technical effects through the collaborative work of four modules: Firstly, the user interaction module 24 accurately acquires personalized data and feedback; secondly, the data acquisition module 21 uses a thin-film sensor array and filtering processing to ensure accurate and continuous pressure data, solving the problem of weak data foundation in traditional systems. Thirdly, the sleep posture recognition module 22 utilizes temporal slicing and low-dimensional feature extraction technology to achieve accurate recognition of effective sleep postures while reducing computational power consumption, avoiding interference from accidental body movements, and achieving a sleep posture (supine, lateral, and moving) recognition accuracy of ≥95%. Simultaneously, it locates the specific sleeping area of ​​the head / neck on the smart pillow, with a recognition response time of ≤0.5 seconds. It also integrates multi-dimensional data through the pillow height adjustment module 23 to calculate and calibrate the pillow height and drive the adjustment, achieving dynamic adaptation of the pillow height to the user's body shape, sleeping posture, preferences, and historical experience. Ultimately, it effectively improves the accuracy of sleeping posture recognition and pillow height adaptability, generating pillow height parameters with a fit of ≥80% to the user's neck curve, realizing personalized adaptation of "a thousand pillows for a thousand people", ensuring sleep continuity, solving the technical pain points of existing smart pillows such as high computing power consumption and insufficient personalization and dynamic adaptation capabilities, and improving the user's sleep comfort and sleep quality.

[0065] In some embodiments, the sleeping posture includes sleeping posture type and sleeping area; the sleeping posture type includes supine, lateral, and resting;

[0066] The sleeping posture recognition module is also used to obtain multiple sleeping posture states based on the multiple sets of low-dimensional pressure data feature parameters; and to determine the initially effective sleeping posture state based on the user's personalized prior data and the system's preset rules.

[0067] The sleeping posture recognition module is further configured to determine the effective sleeping posture state based on the sleeping posture type when the sleeping posture type in the multiple sleeping posture states is either supine or lateral.

[0068] The sleeping posture recognition module is also used to pause the determination of the current effective sleeping posture state and continuously acquire new pressure datasets when body movement occurs in the sleeping posture type among the multiple sleeping posture states; when the number of times the sleeping posture type corresponding to the new pressure dataset is determined to be supine or lateral reaches a first preset number, multiple sleeping posture states are reacquired, and the effective sleeping posture state is obtained based on the reacquired multiple sleeping posture states.

[0069] First, the sleep posture recognition module generates multiple sleep posture states based on multiple sets of low-dimensional pressure data feature parameters. These low-dimensional pressure data feature parameters are the core results of the module's processing of the pressure dataset, covering key information such as the centroid coordinates of pressure distribution, the location of pressure peaks, and the standard deviation of pressure values. Through these parameters, the module can determine the current user's sleep posture type (e.g., relatively uniform pressure distribution when lying on their back, and pressure concentrated on one side when lying on their side), and also locate the specific sleep area of ​​the head / neck on the pillow surface (e.g., the middle of the pillow surface, the left edge, etc.), thus forming complete sleep posture state data including "sleep posture type + sleep area". At the same time, the module combines the user's personalized prior data (e.g., the user's preferred sleep posture) with the system's preset rules (e.g., the initial state defaults to the basic parameters of lying on the back) to determine the initially effective sleep posture state, providing a benchmark for subsequent state updates and judgments.

[0070] Secondly, for scenarios where multiple sleeping positions involve either supine or lateral lying, the sleeping position recognition module uses the sleeping position type as the core criterion. It filters and determines the final valid sleeping position from these positions and the initially effective sleeping position. The core of this logic is to eliminate body movement interference and focus on stable sleep postures. Since supine and lateral lying are both stable sleeping positions during actual sleep, the module compares the consistency of different sleeping positions (e.g., multiple recognitions all show lateral lying with similar sleep areas) and combines this with the adaptability of the initial effective sleeping position to ensure that the final determined valid sleeping position accurately reflects the user's current stable sleep state, providing a reliable basis for subsequent pillow height adjustments.

[0071] Finally, if a body movement-related sleeping posture appears among multiple sleeping positions, the sleeping posture recognition module will immediately pause the determination process for the current valid sleeping posture and continuously acquire new pressure datasets from the data acquisition module. This is because body movement is a momentary action during sleep (such as unconscious turning over at night or slight head movements), not a stable sleeping posture. If the sleeping posture is determined based on body movement, it can easily lead to false triggering of subsequent pillow height adjustments, affecting the continuity of the user's sleep. The module will continuously monitor and determine the newly acquired pressure dataset. When the number of consecutive determinations that the sleeping posture type corresponding to the new pressure dataset is supine or lateral reaches a first preset number (such as 5 consecutive determinations of non-body movement), it is considered that the user has returned to a stable sleep state. At this time, the module will reacquire multiple sleeping postures and complete the determination of the valid sleeping posture based on the new sleeping posture data, ensuring that the valid sleeping posture always matches the user's stable sleep posture.

[0072] In the above embodiments, by clearly defining the sleeping posture state as including a dual dimension of "sleeping posture type + sleep area" and classifying it into three types of sleeping postures: supine, lateral, and body movement, the sleeping posture recognition module can generate accurate sleeping posture state data based on low-dimensional pressure data feature parameters. Simultaneously, the module combines user-personalized prior data with system preset rules to determine the initial valid sleeping posture state. When all sleeping posture types are supine or lateral, it accurately filters valid states. When body movement occurs, the judgment is paused and re-judged after a preset number of consecutive detections of stable sleeping postures. This achieves accurate recognition of stable sleeping postures while avoiding misjudgments of valid sleeping postures caused by instantaneous movements such as body movement. This prevents unnecessary actions triggered by misjudgments during subsequent pillow height adjustments, ensuring the continuity of user sleep and providing the pillow height adjustment module with accurate and reliable valid sleeping posture state data, thus facilitating precise adaptation of pillow height to the user's stable sleep needs.

[0073] In some embodiments, the sleeping posture recognition module is further configured to count the number of times a new sleeping posture state that is different from the sleeping posture type or sleep area of ​​the initially effective valid sleeping posture state among a number of recently acquired second preset number of sleeping posture states; if the number of times the new sleeping posture state appears reaches a third preset number, then the new sleeping posture state is determined as a new effective sleeping posture state, and the initially effective valid sleeping posture state is updated to the new effective sleeping posture state.

[0074] Specifically, to further improve the accuracy and stability of effective sleep posture recognition and avoid misjudgments due to fluctuations in sleep posture in a single instance or a few instances, the sleep posture recognition module also sets up a dynamic confidence feedback mechanism. The module first determines a statistical range of multiple sleep postures obtained in the second preset number (e.g., 10 times). This can be understood as 10 recognition results corresponding to 10 sleep postures generated from 10 time-domain data slices. Within this range, the module compares the difference between each sleep posture and the initially effective sleep posture, focusing on the sleep posture type (e.g., initially supine, new state is side-lying) or sleep area. If there is a difference in any dimension, the new sleeping position is determined, and the total number of times this new sleeping position appears within a second preset number (e.g., 10 times) is counted. When the number of times the new sleeping position appears reaches a third preset number (e.g., 8 times), the module determines that the user has completed a stable sleeping position or sleep area switch. At this time, the new sleeping position is officially determined as a new valid sleeping position, and the initially effective valid sleeping position is updated to this new state as the benchmark for subsequent judgments of valid sleeping positions. For example, if the initial valid sleeping position is determined to be supine, although there may be a few occasional recognition results of side-lying in the following half hour, the number of times the new sleeping position of side-lying appears within any consecutive 10 recognitions never reaches 8 times, and the switching condition of "the number of times the new sleeping position appears reaches the third preset number" is not met, then the module determines that the user's sleeping position has not changed stably, and the valid sleeping position for this half hour will remain supine, further ensuring the stability of the valid sleeping position recognition and avoiding misjudgments caused by short-term fluctuations.

[0075] In the above embodiments, by setting a dynamic confidence feedback mechanism for the sleeping posture recognition module, a second preset number (e.g., 10 times) of sleeping posture states is used as the statistical range, and a third preset number (e.g., 8 times) is used as the state update threshold. The number of new sleeping posture states that differ from the initial effective sleeping posture state in terms of sleeping posture type or sleeping area is counted. The effective sleeping posture state is only updated when the number of new sleeping posture states reaches the threshold. This effectively filters out false judgments of effective sleeping postures caused by single or a few instantaneous data fluctuations (e.g., occasional head movements), reducing the false judgment rate by more than 80%. It ensures that the update of the effective sleeping posture state is based only on the user's stable sleeping posture switching or sleeping area change, thus improving the accuracy and stability of the effective sleeping posture state recognition. It also provides a reliable state basis for the subsequent pillow height adjustment module, avoiding unnecessary pillow height adjustments triggered by invalid state updates, thereby ensuring the continuity of the user's sleep and helping to achieve accurate adaptation of pillow height to the user's actual stable sleep needs.

[0076] In some embodiments, the user interaction module is further configured to obtain the number of days the user has used the service;

[0077] The pillow height adjustment module is also used to calculate the standard pillow height based on the user's personalized prior data and the effective sleeping posture.

[0078] The pillow height adjustment module is also used to adjust the standard pillow height based on the user's personalized prior data, the historical sleep comfort score and the number of days the user has used the pillow, to obtain the calibrated pillow height, and to adjust the pillow surface to the calibrated pillow height.

[0079] In addition to its core functions of acquiring personalized prior data and receiving historical sleep comfort scores, the user interaction module has added the function of acquiring the number of days the user has used the system. This module automatically records the time when the user first uses the system and calculates the cumulative number of days the user has used the system based on the difference between the current usage time and the first usage time, forming key time dimension data. This usage day data will be synchronized to the pillow height adjustment module in real time, serving as the core basis for weight allocation in the subsequent pillow height calibration process, and providing support for the system to switch from relying on prior data to integrating historical feedback.

[0080] The pillow height adjustment module forms a complete processing flow around "standard pillow height calculation - calibration pillow height generation - pillow surface adjustment execution": The first step is to calculate the standard pillow height. The module first calls the user's personalized prior data (including height, weight, preferred pillow height, preferred sleeping position, etc.) provided by the user interaction module, and combines it with the effective sleeping position status output by the sleeping position recognition module (including sleeping position type and sleep area). This data is then fused through a preset formula: "Standard pillow height = (height × 0.05 + weight × 0.02) × sleeping position coefficient (the sleeping position coefficient is 1.0 when the sleeping position type is supine)". The system calculates a standard pillow height that meets the user's basic needs based on the following formula: (Sleeping position type: side sleeping position coefficient 1.2) × pillow height preference coefficient (preference coefficient 1.1 for a high pillow, 0.9 for a low pillow, and 1.0 for a normal pillow) × sleep zone coefficient (coefficient 1.0 for a central sleep zone and 1.1 for a peripheral sleep zone). This ensures the initial accuracy of pillow height matching. In addition, the system supports user customization of pillow height; users can directly input their preferred back sleeping height and side sleeping height, and the standard pillow height will be the height input by the user. The second step is to generate a calibrated pillow height. The module incorporates three types of data: the user's personalized prior data, historical sleep comfort scores (1-5 points), and the number of days the user has used the pillow. Through dynamic weight allocation and coefficient calibration, the standard pillow height is optimized, ultimately resulting in a calibrated pillow height accurate to 0.1cm. The third step is to perform pillow surface adjustment. The module converts the calibrated pillow height parameters into control signals for the actuator, driving the multi-zone electric push rods inside the smart pillow to adjust the pillow height only for the sleep area corresponding to the effective sleeping posture, ensuring that the pillow surface fits the neck curve and the adjustment response time is ≤3 seconds. Ultimately, this achieves a personalized pillow height adaptation that is dynamically optimized based on the user's usage cycle and conforms to long-term sleep habits.

[0081] In the above embodiments, the user interaction module is enhanced with the function of obtaining the number of days the user has used the product, providing time-dimensional data support for the pillow height adjustment module. The pillow height adjustment module first calculates a standard pillow height that meets basic requirements based on the user's personalized prior data and effective sleeping posture. Then, it dynamically optimizes the standard pillow height by combining the user's personalized prior data, historical sleep comfort scores, and the number of days the user has used the product, obtaining a calibrated pillow height accurate to 0.1cm. This calibrated pillow height is then adjusted by driving a multi-zone electric push rod (response time ≤ 3 seconds). This achieves a dynamic evolution of pillow height adaptation from relying on initial prior data to incorporating long-term usage feedback, ensuring accurate matching of pillow height with user body type, sleeping posture, preferences, and historical experience at different usage stages. Furthermore, the multi-zone adjustment ensures that the pillow surface fits the neck curve, avoiding unnecessary overall adjustments. Ultimately, this improves the personalization and accuracy of pillow height adaptation, further ensuring user sleep comfort and sleep quality.

[0082] In some embodiments, the user-personalized prior data includes height, weight, and user pillow height preference;

[0083] The pillow height adjustment module is also used to obtain historical coefficients based on the user's pillow height preference, the historical sleep comfort score, and the number of days the user has used the device; and to obtain prior coefficients based on the user's pillow height preference.

[0084] The pillow height adjustment module is further configured to obtain historical feedback weights and prior data weights based on the number of days the user has used the product; and to adjust the standard pillow height based on the historical coefficients, the historical feedback weights, the prior coefficients, and the prior data weights to obtain a calibrated pillow height.

[0085] Specifically, on the one hand, the pillow height adjustment module builds an adaptation foundation around coefficient calculation, focusing on generating two core parameters: "historical coefficient" and "prior coefficient." For the "historical coefficient" calculation, the module uses the user's pillow height preference as a benchmark, first mapping historical sleep comfort scores (1-5 points) to a basic score coefficient, and then further optimizing it based on the number of days the user has used the product, calculating the historical coefficient according to a preset weighting rule. For the "prior coefficient" calculation, the module directly determines it based on the user's pillow height preference: a preference for a high pillow corresponds to a prior coefficient of 1.1, a preference for a low pillow corresponds to a prior coefficient of 0.9, and a preference for a normal pillow height corresponds to a prior coefficient of 1.0, ensuring a high match between the prior coefficient and the user's initial sleep preferences.

[0086] On the other hand, the pillow height adjustment module allocates weights based on the number of days the user has used the device, dynamically balancing the proportion of "historical feedback weight" and "prior data weight": when the user has used the device for less than 10 days, there is no sufficient historical data by default, so the prior data weight is set to 100% and the historical feedback weight is set to 0%; when the user has used the device for 10-20 days, the historical feedback weight increases to 30% and the prior data weight decreases to 70%; when the user has used the device for 20-30 days, the historical feedback weight further increases to 40% and the prior data weight decreases to 60%; after the user has used the device for more than 30 days, the historical feedback weight stabilizes at 50% and the prior data weight is 50% accordingly. Through this weight rule that dynamically adjusts with the number of days of use, a smooth transition from the system relying on initial prior data to integrating long-term sleep feedback is achieved.

[0087] Finally, the pillow height adjustment module substitutes the historical coefficients, historical feedback weights, prior coefficients, and prior data weights calculated above into the preset calibration formula (calibrated pillow height = standard pillow height × (prior data weights × prior coefficients + historical feedback weights × historical coefficients)) to optimize and adjust the calculated standard pillow height, obtaining a calibrated pillow height accurate to 0.1cm. This ensures that the calibrated pillow height not only matches the user's initial body shape and preferences but also continues to optimize with long-term usage habits, achieving personalized adaptation for each user.

[0088] The above embodiment uses a pillow height adjustment module based on user-personalized prior data. First, it calculates historical coefficients by combining user pillow height preferences, historical sleep comfort scores, and the number of days of use. Then, it determines prior coefficients based on pillow height preferences and dynamically allocates historical feedback weights and prior data weights according to the number of days of use. Finally, it substitutes these into a formula to optimize the standard pillow height and obtain a calibrated pillow height accurate to 0.1cm. This achieves a smooth transition from relying on initial preferences to incorporating long-term sleep feedback in pillow height adaptation. It ensures that the calibrated pillow height not only fits the user's initial body shape and sleep preferences but also continues to optimize with usage habits. This effectively solves the problem that traditional fixed pillow height or single-rule adjustments cannot meet personalized and dynamic adaptation needs, achieving a precise adaptation effect of "a thousand pillows for a thousand people," and further improving user sleep comfort and sleep quality.

[0089] In some embodiments, the pillow height adjustment module is further configured to map the historical sleep comfort score to a base score coefficient based on the user's pillow height preference;

[0090] The pillow height adjustment module is also used to use the basic rating coefficient of the day as the historical coefficient when the number of days the user uses the device is less than the first preset number of days.

[0091] The pillow height adjustment module is also used to calculate the historical coefficient by weighting the basic rating coefficient of each day within the user's usage days according to a preset rule when the number of days the user uses the device is greater than the first preset number of days and less than the second preset number of days.

[0092] The pillow height adjustment module is also used to calculate the historical coefficient by weighting the basic rating coefficients of each day within the most recent second preset number of days according to the preset rules when the number of days the user uses the device is greater than the second preset number of days; wherein, the preset rules are rules for assigning corresponding weights based on the basic rating coefficients of different time periods.

[0093] First, the module maps historical sleep comfort scores (1-5 points) to base rating coefficients based on the user's pillow height preference. This step is a prerequisite for calculating historical coefficients, and its core purpose is to match comfort scores with user sleep preferences. The mapping rules differ for users with different pillow height preferences: for example, for users who "prefer low pillows," 5 and 4 points correspond to a base rating coefficient of 1, 3 points to 0.9, and 2 and 1 points to 0.8; for users who "prefer high pillows," 5 and 4 points correspond to a base rating coefficient of 1, 3 points to 1.1, and 2 and 1 points to 1.2. This mapping ensures that the base rating coefficients accurately convey the matching relationship between "comfort score - pillow height preference."

[0094] The module then calculates historical coefficients based on the different stages of the user's usage days, categorized into three scenarios. Each scenario uses preset rules (assigning corresponding weights to different time periods) as the weighting basis. Combined with time-dimensional adaptation logic, the first preset number of days is typically set to 10 days, and the second preset number of days is typically set to 30 days. Specific scenarios are as follows:

[0095] When the number of days a user uses the system is less than the first preset number of days (e.g., 10 days), the system determines that the user's historical sleep data accumulation is insufficient and cannot form an effective long-term feedback reference. At this time, the module directly uses the basic score coefficient of the day as the historical coefficient. This design avoids the bias of the weighted result caused by the small amount of data and ensures that the initial historical coefficient can reflect the user's current immediate comfort feedback.

[0096] When a user's usage period exceeds the first preset number of days (e.g., 10 days) but is less than the second preset number of days (e.g., 30 days), the module collects the daily basic rating coefficients from the user's first use to the current day, and then calculates the historical coefficients by weighted averaging according to preset rules. These preset rules clearly assign corresponding weights to the basic rating coefficients for different time periods, typically with more recent data having a higher weight than earlier data. For example, the basic rating coefficient weight for the most recent 3 days is 5, and the weight for the basic rating coefficients from 4 to 10 days is 3. The weighted average is calculated by first calculating "(sum of basic rating coefficients for the most recent 3 days × 5 + sum of basic rating coefficients for 4 to 10 days × 3)," and then dividing by the total weight (e.g., for 15 days of use, the total weight is 3 × 5 + 7 × 3 = 36). The result is the historical coefficient for that period.

[0097] When a user's usage period exceeds the second preset number of days (e.g., 30 days), the module determines that earlier data (over 30 days) may no longer reflect changes in the user's current sleep habits (such as subtle adjustments in sleeping posture preferences or neck condition). Therefore, it only selects the daily basic score coefficients within the most recent second preset number of days (e.g., 30 days) and performs a weighted average according to the aforementioned preset rules (e.g., weight 5 for the most recent 3 days, weight 3 for days 4-10, weight 2 for days 11-20, and weight 1 for days 21-30). For example, the calculation first calculates "sum of basic score coefficients for the most recent 3 days × 5 + sum of basic score coefficients for days 4-10 × 3 + sum of basic score coefficients for days 11-20 × 2 + sum of basic score coefficients for days 21-30 × 1", then divides by the fixed weighted sum (5 × 3 + 3 × 7 + 2 × 10 + 1 × 10 = 66) to finally obtain the historical coefficients. This design ensures the timeliness of historical data and covers recent and slightly earlier feedback through multi-time period weighting, avoiding the influence of randomness in data from a single point in time.

[0098] The above embodiment first obtains a basic score coefficient by mapping historical sleep comfort scores based on user pillow height preferences using the pillow height adjustment module. Then, it calculates historical coefficients by categorizing users into three scenarios based on the number of days they have used the product, and combining this with a preset rule that recent data has a higher weight than earlier data. This achieves both a precise match between the basic score coefficient and the user's pillow height preference, ensuring that the coefficient accurately conveys the correlation between comfort feedback and preference. Furthermore, the phased and time-sensitive weighted calculation logic avoids deviations caused by insufficient data and interference from early invalid data. This allows the historical coefficients to reflect both immediate user feedback and long-term stable sleep needs, providing an accurate and reliable quantitative basis for subsequent pillow height calibration calculations. This further ensures the personalization and adaptability of pillow height adjustment, helping to achieve the effect of personalized pillows for each user and dynamic optimization over the usage cycle.

[0099] In some embodiments, the pressure sensor array is a thin-film pressure sensor array; the pressure sensor array is arranged in M ​​columns along the length direction of the pillow surface and N rows along the width direction, forming an N×M array; where M and N are positive integers.

[0100] Specifically, the sensor selected is a thin-film pressure sensor (model: FSR402), with a range of 0-100N, accuracy ±2%FS, and response time ≤10ms, ensuring continuous and accurate data acquisition. The deployment scheme can be as follows: Figure 3 As shown, M columns (e.g., 3 columns) are arranged along the length of the pillow surface (the direction of user head support) and N rows (e.g., 5 rows) are arranged along the width (the direction of left and right side lying), for a total of N×M pressure sensors, forming an N×M (e.g., 5×3) array. The data (1, 2, 3) in the figure are distance ratios. The sensor spacing is dense in the middle and sparse on both sides, covering the entire sleep support area of ​​the smart pillow (e.g., 60cm long × 30cm wide), with no data blind spots.

[0101] The above embodiment utilizes a thin-film pressure sensor array of N×M size, with a denser spacing in the middle and sparser spacing on both sides to cover the sleep support area. This design ensures the continuity and accuracy of pressure data acquisition thanks to the high precision and fast response of the sensor. Furthermore, the reasonable row and column layout and spacing design achieve blind-spot-free coverage of the pillow support area. The dense layout in the middle enhances the pressure perception resolution of the core support area of ​​the head and neck, providing high-fidelity, full-range raw data support for the subsequent sleep posture recognition module to extract key feature parameters such as the center of gravity and peak position of pressure distribution. This, in turn, helps improve the accuracy of sleep posture recognition and the adaptability of subsequent pillow height adjustment, ensuring user sleep comfort.

[0102] In some embodiments, the low-dimensional pressure data feature parameters include the centroid coordinates of the pressure distribution, the location of the pressure peak, and the standard deviation of the pressure value;

[0103] The sleeping posture recognition module is also used to input the feature parameters of the multiple sets of low-dimensional stress data into a pre-trained support vector machine algorithm model to obtain multiple sleeping posture states; the pre-trained support vector machine algorithm model is trained based on stress data of multiple users of different body types and genders in different sleeping postures during real sleep.

[0104] Specifically, the characteristic parameters of low-dimensional pressure data include pressure distribution centroid coordinates, pressure peak position, pressure value standard deviation, pressure mean, pressure median, pressure range, pressure skewness, pressure kurtosis, and pressure activation ratio. Among them, the pressure distribution centroid coordinates refer to the weighted center coordinates of all effective pressure values ​​in the N×M pressure sensor array, directly reflecting the position of the head's center of gravity on the pillow surface. This parameter is the core basis for determining sleep area and sleeping posture type. The pressure peak position refers to the coordinates of the sensor with the highest pressure value in the sensor array, which usually corresponds to the core area of ​​contact between the head and the pillow surface. This parameter can help verify the sleeping posture trend reflected by the pressure distribution centroid coordinates and improve the reliability of the features. The standard deviation of pressure values ​​quantifies the uniformity of pressure distribution by calculating the deviation of all effective pressure values ​​from the average value. This parameter serves as an important supplementary feature to distinguish between supine and lateral lying positions, and can also filter out abnormal data interference caused by minor sensor errors. The pressure mean is the arithmetic mean of all effective pressure sensor data. The pressure median is the value in the middle after all effective pressure data are sorted from smallest to largest (if the data volume is even, the average of the two middle values ​​is taken). The pressure range is the difference between the maximum and minimum values ​​of all effective pressure data. The pressure skewness measures the asymmetry of the pressure data distribution, directly reflecting the unilateral concentration trend of the pressure distribution. The pressure kurtosis measures the peak of the pressure data distribution, reflecting the concentration density of pressure. The pressure activation ratio is the proportion of the number of sensors with pressure values ​​greater than a preset threshold (which can be calibrated according to sensor sensitivity) to the total number of sensors, reflecting the contact area between the head and the pillow surface.

[0105] When performing the sleep posture recognition task, the sleep posture recognition module first preprocesses the collected low-dimensional stress data feature parameters, including data standardization (mapping parameters of different dimensions to a unified numerical range to eliminate the influence of units) and outlier filtering (removing invalid data caused by sensor instantaneous errors based on the 3σ criterion). Then, the preprocessed low-dimensional stress data feature parameters are input into a pre-trained support vector machine (SVM) algorithm model. During the training phase, this pre-trained SVM algorithm model has incorporated over 10,000 users of different body types and genders (e.g., covering common body types such as height 150-190cm and weight 40-100kg). The model collects massive amounts of pressure data from users in various sleeping positions, including supine, lateral, and active positions, during their actual sleep. Through feature learning and classification boundary optimization of this data, the model can accurately identify the differences in low-dimensional pressure data features of users of different body types and genders in different sleeping positions. It then outputs the corresponding sleeping position judgment result for each set of low-dimensional pressure data feature parameters, ultimately obtaining multiple sleeping position states. This provides a reliable algorithmic basis for selecting effective sleeping position states from multiple sleeping position states. At the same time, based on the training of data from users of different body types, the model ensures that it has high accuracy in recognizing the sleeping positions of users of different body types, avoiding recognition bias caused by differences in body type.

[0106] The above embodiments, by clearly defining the low-dimensional pressure data feature parameters as including the centroid coordinates of pressure distribution, the position of pressure peak, and the standard deviation of pressure value, and having the sleeping posture recognition module first perform standardization and outlier filtering preprocessing on multiple sets of these feature parameters, then inputting them into a support vector machine algorithm model pre-trained based on pressure data from multiple users of different body types in different sleeping postures, not only ensures the comprehensiveness and reliability of pressure data feature extraction through the complementary effect of the three types of feature parameters, but also improves data quality through preprocessing, and ensures high accuracy and generalization ability for recognizing sleeping postures of different body types by using a model trained on multiple body types, ultimately achieving stable output of multiple sleeping posture states, achieving a sleeping posture (supine, side-lying, body movement) recognition accuracy of ≥95%, while locating the specific sleep area of ​​the head / neck on the smart pillow, with a recognition response time of ≤0.5 seconds, providing a reliable basis for subsequent selection of effective sleeping posture states, avoiding sleeping posture recognition deviations caused by single features, data interference, or body type differences, and helping to improve the accuracy and stability of the entire system's sleeping posture recognition process.

[0107] This application provides a method for sleep posture recognition and adaptive pillow height adjustment. Figure 4 This is a flowchart of a sleeping posture recognition and adaptive pillow height adjustment method according to an embodiment of this application, such as... Figure 4 As shown, the process includes the following steps:

[0108] Step S401: Obtain the user's personalized prior data and historical sleep comfort scores;

[0109] Step S402: Real-time acquisition of pressure dataset from a preset pressure sensor array; construction of multiple time-domain data slices from the pressure dataset according to a preset time window; input of the multiple time-domain data slices into a preset time-domain feature model to obtain multiple sets of low-dimensional pressure data feature parameters.

[0110] Step S403: Identify effective sleeping postures based on multiple sets of low-dimensional pressure data feature parameters;

[0111] Step S404: Based on the user's personalized prior data, effective sleeping posture and historical sleep comfort score, obtain the calibrated pillow height and adjust the pillow surface to the calibrated pillow height.

[0112] Specifically, the system first obtains personalized prior data (including physiological parameters such as height and weight, as well as preference parameters such as pillow height and common sleeping posture) and historical sleep comfort scores (1-5 point scores provided by users after each day's sleep, forming a historical dataset in chronological order) through the system interaction module, providing a basis for subsequent adjustments;

[0113] Next, during the user's sleep, a pressure dataset is collected in real time at a preset frequency (usually 10Hz) using a preset thin-film pressure sensor array (such as an N×M distributed array). The continuous pressure data is then segmented according to a preset time window to construct multiple sets of time-domain data slices. These slices are input into a preset time-domain feature model (an optimized feature extraction algorithm model) to extract multiple sets of low-dimensional pressure data feature parameters, including the centroid coordinates of the pressure distribution, the position of the pressure peak, and the standard deviation of the pressure value. This achieves dimensionality reduction of the original data while retaining key information.

[0114] Subsequently, the sleeping posture recognition module inputs multiple sets of low-dimensional stress data feature parameters into a pre-trained support vector machine algorithm model (trained based on multi-body type user data) to obtain multiple preliminary sleeping posture states. Combined with the user's common sleeping posture tendencies and preset rules (such as excluding body movement states and verifying state stability), the module filters out the effective sleeping posture states that include sleeping posture type and sleep area.

[0115] Finally, the pillow height adjustment module first calculates the standard pillow height based on the user's personalized prior data and effective sleeping posture. Then, it incorporates historical sleep comfort scores and the number of days the user has used the device to calculate historical and prior coefficients and assign corresponding weights. Substituting these into a preset formula yields a calibrated pillow height accurate to 0.1cm. Ultimately, this drives a multi-zone electric push rod to adjust the pillow surface of the corresponding sleep area to the calibrated pillow height. After adjustment, it continuously collects pressure data to monitor whether the user's sleeping posture changes: if the sleeping posture remains stable, the current pillow height is maintained; if the sleeping posture changes, the "identification → calculation → adjustment" process is repeated. After waking up, the user scores their sleep comfort (1-5 points) through the interactive module or the sleep quality assessment model. The pillow height adjustment module stores the "current sleeping posture - pillow height - comfort score" in a historical database and updates the weights of historical feedback data (recent days' data have a higher weight than earlier data). The next time the device is used, the updated historical data is automatically retrieved to optimize the pillow height calculation parameters, achieving long-term adaptability improvement.

[0116] The above steps establish a foundation for adaptation by first acquiring personalized prior data and historical sleep comfort scores from the user. Then, the raw pressure dataset collected by the pressure sensor array is processed by constructing multiple time-domain data slices according to a preset time window and inputting them into a time-domain feature model to extract multiple sets of low-dimensional pressure data feature parameters. This process significantly reduces the amount of data processed by subsequent algorithms through data dimensionality reduction, avoiding excessive computational power consumption from the massive amount of raw pressure data. Subsequently, effective sleeping postures are identified based on the low-dimensional feature parameters, eliminating the need for complex calculations on the high-dimensional raw data and further reducing computational power consumption. Finally, the multi-dimensional data is integrated to calculate and calibrate the pillow height and adjust the pillow surface. The overall process reduces the computational power consumption of the algorithm through low-dimensional feature extraction and targeted data processing, and automates the entire process from data collection to pillow height adjustment. It achieves dynamic adaptation of pillow height to user body shape, sleeping posture, preferences and historical experience, ultimately effectively improving the accuracy of sleeping posture recognition and pillow height adaptability. It generates pillow height parameters with a fit of ≥80% to the user's neck curve, achieving personalized adaptation of "one pillow for one thousand people", ensuring sleep continuity. It solves the technical pain points of existing smart pillows, such as high computational power consumption and insufficient personalization and dynamic adaptation capabilities, improves user sleep comfort and sleep quality, and ultimately achieves the dual effects of high efficiency, low consumption and personalized and accurate adaptation.

[0117] Furthermore, in conjunction with the sleeping posture recognition and adaptive pillow height adjustment system in the above embodiments, this application embodiment can provide a storage medium for implementation. This storage medium stores a computer program; when executed by a processor, the computer program implements the operation of any of the sleeping posture recognition and adaptive pillow height adjustment systems in the above embodiments.

[0118] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0119] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0120] Those skilled in the art should understand that the technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments have been described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0121] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A sleep position recognition and adaptive pillow height adjustment system, characterized in that, The system comprises a data acquisition module, a sleep posture recognition module, a pillow height adjustment module, and a user interaction module; The user interaction module is configured to obtain user individualized prior data and a historical sleep comfort score; The data acquisition module is configured to acquire a pressure data set of a preset pressure sensor array in real time; The sleep posture recognition module is configured to construct a plurality of time domain data slices from the pressure data set according to a preset time window, input the plurality of time domain data slices into a preset time domain feature model, and obtain a plurality of low-dimensional pressure data feature parameters; and identify an effective sleep posture state based on the plurality of low-dimensional pressure data feature parameters; The sleep posture recognition module is further configured to obtain a plurality of sleep posture states based on the plurality of low-dimensional pressure data feature parameters, and determine an initial effective sleep posture state based on the user individualized prior data and a system preset rule; the sleep posture state comprises a sleep posture type and a sleep area; the sleep posture type comprises supine, lateral recumbency, and body movement; The sleep posture recognition module is further configured to, in a case where the sleep posture type in the plurality of sleep posture states is supine or lateral recumbency, determine the effective sleep posture state from the plurality of sleep posture states and the initial effective sleep posture state based on the sleep posture type; The sleep posture recognition module is further configured to, in a case where body movement appears in the sleep posture type in the plurality of sleep posture states, pause the determination of the current effective sleep posture state, and continuously acquire a new pressure data set; when the number of times of continuously determining that the sleep posture type of the new pressure data set is supine or lateral recumbency reaches a first preset number, reacquire a plurality of sleep posture states, and obtain an effective sleep posture state based on the reacquired plurality of sleep posture states; The pillow height adjustment module is configured to obtain a calibrated pillow height based on the user individualized prior data, the effective sleep posture state, and the historical sleep comfort score, and adjust a pillow surface to the calibrated pillow height.

2. The sleep position identification and adaptive pillow height adjustment system of claim 1, wherein, The sleep posture recognition module is further configured to count the number of times of appearance of a new sleep posture state different from the sleep posture type or the sleep area of the initial effective sleep posture state in a second preset number of the plurality of sleep posture states acquired most recently; If the number of times of appearance of the new sleep posture state reaches a third preset number, the new sleep posture state is determined as a new effective sleep posture state, and the initial effective sleep posture state is updated to the new effective sleep posture state.

3. The sleep position identification and adaptive pillow height adjustment system of claim 1, wherein, The user interaction module is further configured to acquire a user use day number; The pillow height adjustment module is further configured to calculate a standard pillow height based on the user individualized prior data and the effective sleep posture state; The pillow height adjustment module is further configured to adjust the standard pillow height based on the user individualized prior data, the historical sleep comfort score, and the user use day number, obtain the calibrated pillow height, and adjust the pillow surface to the calibrated pillow height.

4. The sleep position identification and adaptive pillow height adjustment system of claim 3, wherein, The user individualized prior data comprises height, weight, and user pillow height preference; The pillow height adjustment module is further configured to obtain a historical coefficient based on the user pillow height preference, the historical sleep comfort score, and the user use day number, and obtain a prior coefficient based on the user pillow height preference. The pillow height adjustment module is further configured to obtain a historical feedback weight and a priori data weight based on the number of days of use of the user; The standard pillow height is adjusted based on the historical coefficient, the historical feedback weight, the priori coefficient, and the priori data weight to obtain a calibrated pillow height.

5. The sleep position identification and adaptive pillow height adjustment system of claim 4, wherein, The pillow height adjustment module is further configured to map the historical sleep comfort score to a base score coefficient based on the user pillow height preference. The pillow height adjustment module is further configured to, when the number of days of use of the user is less than a first preset number of days, take the base score coefficient of the day as the historical coefficient. The pillow height adjustment module is further configured to, when the number of days of use of the user is greater than the first preset number of days and less than a second preset number of days, perform weighted average calculation on the base score coefficients of each day within the number of days of use of the user according to a preset rule to obtain the historical coefficient. The pillow height adjustment module is further configured to, when the number of days of use of the user is greater than the second preset number of days, only perform weighted average calculation on the base score coefficients of each day within the most recent second preset number of days according to the preset rule to obtain the historical coefficient; wherein the preset rule is a rule of assigning corresponding weights to base score coefficients of different time periods.

6. The sleep position identification and adaptive pillow height adjustment system of claim 1, wherein, The pressure sensor array is a thin-film pressure sensor array; the pressure sensor array is arranged in M columns along the length direction of the pillow surface and N rows along the width direction, forming an N×M array; wherein M and N are positive integers.

7. The sleep position identification and adaptive pillow height adjustment system of claim 1, wherein, The low-dimensional pressure data feature parameters include pressure distribution barycentric coordinates, pressure peak position, and pressure value standard deviation. The sleep posture recognition module is further configured to input the plurality of sets of low-dimensional pressure data feature parameters into a pre-trained support vector machine algorithm model to obtain a plurality of sleep posture states; the pre-trained support vector machine algorithm model is trained based on pressure data of different sleep postures of a plurality of users of different body types and genders in real sleep states.

8. A sleep position recognition and adaptive pillow height adjustment method, characterized in that, The method is applied to the sleep posture recognition and adaptive pillow height adjustment system of any one of claims 1 to 7, and the method comprises: obtaining user individualized priori data and historical sleep comfort scores; real-time collection of pressure data sets of a preset pressure sensor array, construction of a plurality of sets of time domain data slices from the pressure data sets according to a preset time window, input of the plurality of sets of time domain data slices into a preset time domain feature model, and obtaining of a plurality of sets of low-dimensional pressure data feature parameters; based on the plurality of sets of low-dimensional pressure data feature parameters, recognition of an effective sleep posture state, comprising: based on the plurality of sets of low-dimensional pressure data feature parameters, obtaining of a plurality of sleep posture states; and based on the user individualized priori data and a system preset rule, determining of an initial effective sleep posture state; the sleep posture state comprises a sleep posture type and a sleep area; the sleep posture type comprises supine, lateral recumbency, and body movement; in a case where the sleep posture type in the plurality of sleep posture states is all supine or lateral recumbency, based on the sleep posture type, determining the effective sleep posture state from the plurality of sleep posture states and the initial effective sleep posture state. In a case where body movement occurs in the sleep position type in the plurality of sleep position states, the determination of the current effective sleep position state is suspended, and new pressure data sets are continuously acquired; when the number of times of continuously determining that the sleep position type corresponding to the new pressure data sets is supine or lateral reaches a first preset number, the plurality of sleep position states are re-acquired, and an effective sleep position state is obtained based on the re-acquired plurality of sleep position states; Based on the user individualized prior data, the effective sleep position state, and the historical sleep comfort score, a calibrated pillow height is obtained, and a pillow surface is adjusted to the calibrated pillow height.

9. A storage medium, characterized by The storage medium has a computer program stored therein, wherein the computer program is configured to execute the work of the sleep position recognition and adaptive pillow height adjustment system of any one of claims 1 to 7 when running.

Images