Sitting posture monitoring method, system and intelligent seat

Abstract

This invention relates to the fields of intelligent health management and artificial intelligence application technology, and particularly to a sitting posture monitoring method, system, and intelligent chair. First, it generates comprehensive posture data by collecting user data on pressure distribution, posture angle, electromyography signals, and temperature distribution. Next, it uses hybrid graph neural networks and topology analysis technology to perform multi-level analysis of the posture data and identify posture risks. Then, it generates a collaborative adjustment strategy for intelligent materials through a nonlinear dynamics model and adjusts the chair shape in real time to correct the user's posture. Finally, the system generates a personalized posture assessment based on the optimization strategy and historical data, and formulates a maintenance plan. This invention achieves precise monitoring and proactive adjustment of the user's sitting posture, effectively improving the problems of poor model accuracy and lack of dynamic adjustment capabilities in traditional technologies, and providing users with personalized health management services.

Description

Technical Field

[0001] The present invention relates to the field of intelligent health management and artificial intelligence application technology, and in particular to a sitting posture monitoring method, system and intelligent chair. Background Art

[0002] In modern society, as people sit for long periods of time in their daily work and study, the negative impact of bad sitting posture on health is becoming increasingly prominent. Long-term bad sitting posture will not only cause chronic pain in the cervical spine, lumbar spine and other parts, but may also cause more serious spinal problems. Therefore, effective monitoring and correction of sitting posture has become an urgent need.

[0003] The existing sitting posture monitoring technology (Chinese invention patent, publication number: CN113297938B, name: sitting posture monitoring method, device, electronic device and storage medium) mainly relies on visual sensors and distance sensors. By acquiring the user's sitting posture image and distance data, a sitting posture model is established and compared with the preset standard sitting posture model. When it is detected that the deviation between the user's sitting posture and the standard sitting posture exceeds the preset threshold, the system will issue an adjustment reminder. However, this method has several defects:

[0004] Existing technologies mainly rely on image processing and distance measurement to build sitting posture models, but external factors such as image quality, lighting conditions, and measurement angles may affect the accuracy of the model, leading to misjudgments and false alerts;

[0005] The existing system only reminds when the sitting posture deviation exceeds the threshold, and cannot dynamically adjust the support structure according to the user's real-time posture changes, making it difficult to achieve active correction;

[0006] Existing technologies usually only focus on static posture comparison, lack in-depth analysis of users' long-term posture changes and health trends, and are unable to provide personalized posture maintenance recommendations. Summary of the invention

[0007] In view of the many problems existing in the above-mentioned prior art, the present invention provides a sitting posture monitoring method, system and intelligent chair. The present invention monitors the user's sitting posture in real time through multimodal data fusion, hybrid graph neural network and topological analysis, nonlinear dynamic model and other technical means, and realizes active correction and long-term health management of the user's posture through dynamic adjustment of intelligent materials. The present invention not only improves the accuracy of sitting posture monitoring, but also can provide customized posture maintenance plans according to the personalized needs of users, significantly reducing the negative impact of bad sitting posture on health.

[0008] A sitting posture monitoring method comprises the following steps:

[0009] Collect the user's pressure distribution data, posture angle data, electromyographic signal data and temperature distribution data; perform time synchronization correction on the collected data, and perform multi-modal fusion to generate comprehensive posture data;

[0010] Based on the hybrid graph neural network and topological analysis, the comprehensive posture data is analyzed at multiple levels, the posture topological features are extracted, and the posture topological data is generated; the posture topological data is input into the multi-layer structure of the hybrid graph neural network, and the posture risk assessment is performed through a dynamic feedback loop to generate posture risk assessment data;

[0011] Based on the posture risk assessment data, the intelligent material collaborative adjustment strategy is generated through the nonlinear dynamics and adaptive shape coupling model; according to the intelligent material collaborative adjustment strategy, the shape of the liquid metal and programmable material in the smart seat is adjusted, the posture correction is implemented, and the adjustment feedback data is generated;

[0012] Generate dynamic virtual posture data using adjustment feedback data through generative adversarial networks, and generate simulated posture feedback data in the adaptive posture simulation module; optimize the adjustment strategy using simulated posture feedback data and adjustment feedback data through the collaborative reinforcement learning module, generate optimized adjustment strategy data, and implement further intelligent material adjustment;

[0013] Generate personalized posture evaluation data based on the optimization adjustment strategy data and historical adjustment data; formulate a personalized posture maintenance plan based on the personalized posture evaluation data, and perform dynamic adjustments and regularly generate posture reports based on the personalized posture maintenance plan.

[0014] Preferably, the steps of performing multi-level analysis on the comprehensive posture data based on the hybrid graph neural network and topological analysis include:

[0015] Perform topological data analysis on the comprehensive posture data, extract the shape features in the user's sitting posture through persistent coherence calculation, and generate posture topological feature data; the persistent coherence calculation includes calculating the coherence group of a single sitting posture and calculating its life cycle to identify features with a long life cycle. The persistent coherence calculation formula is as follows:

[0016]

[0017] in, Indicates dimensional homology group; Indicates Boundary operators of layers; represents the kernel space of the boundary operator; represents the image space of the boundary operator of the previous layer;

[0018] The posture topology feature data is used as input and sent to the first layer of the hybrid graph neural network. The first layer of the hybrid graph neural network is used to analyze the local posture features and generate the first layer of posture analysis data containing local posture information;

[0019] The first layer of posture analysis data is input into the second layer of the hybrid graph neural network, and the second layer of the hybrid graph neural network is used to analyze the user's whole body posture characteristics and generate the second layer of posture analysis data containing the whole body posture information;

[0020] The second-layer posture analysis data is input into the third layer of the hybrid graph neural network. The third-layer hybrid graph neural network is used to analyze the overall coordination of the postures of various parts of the user's body, and adjust the data of the first and second layers through a dynamic feedback loop to finally generate posture risk assessment data.

[0021] Preferably, the step of generating posture risk assessment data comprises:

[0022] Compare the third-layer posture analysis data with the user's historical posture data to identify potential posture problems and long-term health risks by comparing the difference between the user's current posture and historical data;

[0023] Based on the results of the comparative analysis, posture risk assessment data including the user's posture health risk level and recommended posture adjustment measures are generated to guide the adjustment of smart materials.

[0024] Preferably, the step of generating a smart material collaborative adjustment strategy through a nonlinear dynamics and adaptive shape coupling model based on posture risk assessment data includes:

[0025] Based on the posture risk assessment data, a nonlinear dynamics model is used to calculate the potential change trend of the user's posture, and dynamically adjusted prediction data is generated. The dynamically adjusted prediction data is used to predict the possible posture changes of the user in the next period of time. The calculation formula of the dynamically adjusted prediction data is as follows:

[0026]

[0027] in, Indicates time The state vector at the moment; represents the initial state vector; Kinetic equations representing changes of state; Indicates at time The state vector at the moment; Indicates at time The control input vector at time t; represents the integral variable, which means from the initial time To current time any time;

[0028] The dynamic adjustment prediction data is input into the adaptive shape coupling model, and the adaptive shape coupling model generates an intelligent material collaborative adjustment strategy according to the dynamic adjustment prediction data, wherein the strategy includes adjustment instructions for the liquid metal morphology and adjustment instructions for the hardness and morphology of the programmable material.

[0029] Preferably, the smart material collaborative adjustment strategy includes:

[0030] According to the smart material collaborative adjustment strategy, the liquid metal in the smart seat is adjusted to make its shape meet the user's current and predicted posture requirements, forming a support structure that adapts to the user's posture;

[0031] At the same time, the programmable material in the smart seat is adjusted, including its hardness and shape, to enhance the coordination with the change of liquid metal shape, thereby optimizing the overall support effect;

[0032] Generate adjustment feedback data, which is used to evaluate the actual effect of the smart material adjustment and provide a reference for subsequent adjustments.

[0033] Preferably, the step of generating dynamic virtual posture data by adjusting feedback data through a generative adversarial network includes:

[0034] Using the adjustment feedback data and the user's historical posture data to perform adversarial training through a generative adversarial network to generate dynamic virtual posture data that is highly matched with the actual situation, and the dynamic virtual posture data is used to simulate the user's possible reactions under different posture adjustments;

[0035] The dynamic virtual posture data is input into the adaptive posture simulation module to generate simulated posture feedback data, which is used to correct and optimize the further adjustment strategy of the smart material.

[0036] Preferably, the step of optimizing the adjustment strategy by using the simulated posture feedback data and the adjustment feedback data through the collaborative reinforcement learning module includes:

[0037] The collaborative reinforcement learning module realizes the collaborative optimization of multiple agents in the adjustment strategy of each part of the smart seat through joint training of the simulated posture feedback data and the adjustment feedback data. The agents include the liquid metal adjustment module, the programmable material adjustment module and the posture sensing module in the seat. The agents of each part coordinate and cooperate with each other during the execution process to generate optimized adjustment strategy data.

[0038] The optimized adjustment strategy data is applied to further adjust the smart materials, so that the smart seat can achieve optimal support and posture correction for the user under complex posture conditions.

[0039] Preferably, the step of generating personalized posture assessment data according to the optimization adjustment strategy data and the historical adjustment data includes:

[0040] Summarize optimization adjustment strategy data and historical adjustment data, and identify users' long-term posture changes and potential health problems by analyzing users' long-term posture change patterns and health trends;

[0041] Based on the analysis results, personalized posture assessment data is generated, where the personalized posture assessment data includes the user's current posture health status, possible health risks, and personalized posture improvement suggestions.

[0042] A sitting posture monitoring system, used to implement the sitting posture monitoring method, comprising:

[0043] A multimodal sensing module, including a pressure sensor, an inertial measurement unit sensor, an electromyographic sensor and a temperature sensor, wherein the multimodal sensing module is used to collect pressure distribution data, posture angle data, electromyographic signal data and temperature distribution data of the user, and to perform time synchronization correction on the collected data to generate comprehensive posture data;

[0044] A data processing module, comprising a topological data analysis unit and a hybrid graph neural network unit, wherein the topological data analysis unit is used to extract topological features from the comprehensive posture data to generate posture topological feature data, and the hybrid graph neural network unit is used to perform multi-level posture analysis based on the posture topological feature data to generate posture risk assessment data;

[0045] A nonlinear dynamics and adaptive shape coupling module, the module is used to generate dynamic adjustment prediction data through a nonlinear dynamics model according to the posture risk assessment data, and generate an intelligent material collaborative adjustment strategy based on the dynamic adjustment prediction data, the intelligent material collaborative adjustment strategy includes adjustment instructions for the liquid metal morphology and adjustment instructions for the hardness and morphology of the programmable material;

[0046] An intelligent material adjustment module, comprising a liquid metal adjustment unit and a programmable material adjustment unit, wherein the liquid metal adjustment unit is used to adjust the shape of the liquid metal, and the programmable material adjustment unit is used to adjust the hardness and shape of the programmable material to implement posture correction and generate adjustment feedback data;

[0047] A generative adversarial network and adaptive posture simulation module, the module is used to generate dynamic virtual posture data using the adjustment feedback data through a generative adversarial network, and to generate simulated posture feedback data through adaptive posture simulation;

[0048] A collaborative reinforcement learning module, which is used to achieve multi-agent collaborative optimization by jointly training the simulation posture feedback data and the adjustment feedback data. The intelligent agent includes a liquid metal adjustment module, a programmable material adjustment module and a posture sensing module, generates optimization adjustment strategy data, and applies the optimization adjustment strategy data to further adjustment of the intelligent material;

[0049] A personalized posture evaluation and maintenance module is used to summarize the optimization adjustment strategy data and historical adjustment data, generate personalized posture evaluation data, and formulate a personalized posture maintenance plan based on the personalized posture evaluation data, while performing dynamic adjustments and generating regular posture reports.

[0050] A smart chair is provided with the sitting posture monitoring system.

[0051] Compared with the prior art, the advantages and beneficial effects of the present invention are:

[0052] Through the multimodal data fusion technology, accurate perception and dynamic adjustment of user posture are achieved; the present invention adopts multiple data sources such as pressure distribution data, posture angle data, electromyographic signal data and temperature distribution data, and generates comprehensive posture data through time synchronization correction and multimodal fusion, thereby improving the accuracy of the sitting posture model;

[0053] Through the hybrid graph neural network and topological analysis technology, the comprehensive analysis and dynamic risk assessment of user posture are achieved; the present invention introduces a hybrid graph neural network to perform multi-level analysis on posture data, and extracts posture features in combination with topological analysis, which not only improves the accuracy of data processing, but also can dynamically evaluate the user's posture and identify potential health risks;

[0054] Through the technical means of nonlinear dynamics and adaptive shape coupling model, the autonomous adjustment and posture correction effects of intelligent materials are realized; the present invention predicts the user's posture change trend through a nonlinear dynamic model, and generates a collaborative adjustment strategy for intelligent materials, so that the seat can adjust its shape according to the user's real-time needs, thereby achieving the effect of active posture correction. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a schematic diagram of the process of the present invention;

[0056] Figure 2 This is a schematic diagram of adjusting the smart material in the present invention;

[0057] Figure 3 Schematic diagram of the generative adversarial network and adaptive posture simulation in the present invention

[0058] Figure 4 It is a structural block diagram of the system of the present invention. DETAILED DESCRIPTION

[0059] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is obvious that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0060] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0061] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0062] like Figure 1 As shown, a sitting posture monitoring method comprises the following steps:

[0063] Collect the user's pressure distribution data, posture angle data, electromyographic signal data and temperature distribution data; perform time synchronization correction on the collected data, and perform multi-modal fusion to generate comprehensive posture data;

[0064] The collection of pressure distribution data mainly relies on the built-in pressure sensors of the smart chair, which can sense the pressure distribution of the part where the user contacts the seat. By analyzing this data, it is possible to determine whether the user's posture on the seat is uniform and identify the pressure concentration areas that may cause bad posture. For example, when it is detected that the user has exerted too much pressure on a certain part for a long time, the system can identify the abnormal pressure distribution in this part and use it as an early signal of bad sitting posture.

[0065] Posture angle data is collected by the inertial measurement unit (IMU) sensor, which can capture the three-dimensional posture angle information of the user's upper body, such as forward leaning, backward leaning, left and right tilting, etc. These angle data are used to determine whether the user is in the correct sitting posture. For example, excessive forward leaning may indicate that the user is working with his head down. If this posture lasts for too long, it may cause health problems in the cervical spine and back. By continuously collecting and analyzing these angle data, the system can adjust the user's posture in real time to prevent potential health risks.

[0066] The EMG signal data is obtained through the EMG sensor integrated in the seat. The EMG signal reflects the activity status of the user's muscles and can provide information related to the user's muscle tension and fatigue. When the user maintains a certain posture for a long time, specific muscles may become fatigued, resulting in posture deviation. By monitoring the EMG signal, the system can identify the tension of the muscles, warn the user in advance and suggest posture adjustments, thereby avoiding muscle fatigue and pain.

[0067] Temperature distribution data is collected by temperature sensors in the seat, which can detect temperature changes in the area where the user's body contacts the seat. Temperature distribution can indirectly reflect the user's posture stability and the state of the body contact surface. For example, when the temperature of a certain part rises significantly, it may indicate that the contact pressure in that area is too high, or that the user has not changed his posture for a long time. The system can combine temperature distribution data with other data to further accurately locate bad postures.

[0068] After completing the collection of the above data, the system will perform time synchronization correction on these data. The purpose of time synchronization correction is to ensure that the data collected by all sensors can be compared and fused on the same time basis. Since different sensors have different sampling frequencies and response times, time synchronization is a prerequisite for fusing different data sources. For example, the sampling time of pressure distribution data and attitude angle data may not be synchronized. After correction, the system can uniformly process the attitude information at the same time point, thereby improving the temporal and spatial consistency of the data.

[0069] Next, the system will perform multimodal fusion, that is, integrate multi-source data such as pressure distribution, posture angle, electromyographic signal and temperature distribution into a comprehensive posture data through a fusion algorithm. The principle of this fusion is to integrate data from different modes through weighting, feature extraction, statistical analysis and other technologies to form a comprehensive data model that can fully reflect the user's current sitting posture. The comprehensive posture data not only contains geometric information such as the user's pressure distribution and posture angle, but also combines physiological signals such as electromyography and temperature, thereby more comprehensively reflecting the user's sitting posture.

[0070] Based on the hybrid graph neural network and topological analysis, the comprehensive posture data is analyzed at multiple levels, the posture topological features are extracted, and the posture topological data is generated; the posture topological data is input into the multi-layer structure of the hybrid graph neural network, and the posture risk assessment is performed through a dynamic feedback loop to generate posture risk assessment data;

[0071] Preferably, the steps of performing multi-level analysis on the comprehensive posture data based on the hybrid graph neural network and topological analysis include:

[0072] Perform topological data analysis on the comprehensive posture data, extract the shape features in the user's sitting posture through persistent coherence calculation, and generate posture topological feature data; the persistent coherence calculation includes calculating the coherence group of a single sitting posture and calculating its life cycle to identify features with a long life cycle. The persistent coherence calculation formula is as follows:

[0073]

[0074] in, Indicates dimensional homology group; Indicates Boundary operators of layers; represents the kernel space of the boundary operator; Represents the image space of the boundary operator of the previous layer; the core role of this calculation is to identify features with long life cycles in the sitting posture data, that is, those structures that remain stable at different scales, which are often closely related to the user's posture maintenance mode. For example, if the persistent homogeneous features of a posture mode are detected at multiple scales, the system can determine that the posture may be a state maintained by the user for a long time, and then analyze the health risks of this state.

[0075] The posture topology feature data is used as input and sent to the first layer of the hybrid graph neural network. The first layer of the hybrid graph neural network is used to analyze the local posture features and generate the first layer of posture analysis data containing local posture information;

[0076] The first layer of posture analysis data is input into the second layer of the hybrid graph neural network, and the second layer of the hybrid graph neural network is used to analyze the user's whole body posture characteristics and generate the second layer of posture analysis data containing the whole body posture information;

[0077] The second-layer posture analysis data is input into the third layer of the hybrid graph neural network. The third-layer hybrid graph neural network is used to analyze the overall coordination of the postures of various parts of the user's body, and adjust the data of the first and second layers through a dynamic feedback loop to finally generate posture risk assessment data.

[0078] The system performs topological data analysis on the comprehensive posture data. One of the key steps is to extract the shape features of the user's sitting posture through persistent homology calculation. Persistent homology is a topological data analysis technique that captures the shape features of the data by calculating the homology groups of different dimensions in the data set. Specifically, the homology group represents the holes or loop structures formed by the data in different dimensions.

[0079] After generating posture topological feature data, the system inputs these data into a hybrid graph neural network for further analysis. Hybrid graph neural networks (GNNs) are able to process non-Euclidean data with complex connections and extract different levels of features in the data through a multi-layer structure. In the first layer of the neural network, the system mainly analyzes the local features of the user's sitting posture, such as the angle changes of individual joints or the pressure distribution of specific parts. By analyzing these local features, the system is able to identify local posture abnormalities. For example, if certain local features show abnormally high pressure or unusual angle changes, the system can determine that these parts may have potential health risks.

[0080] The output data of the first layer, that is, the first-layer posture analysis data, will be input into the second layer of the hybrid graph neural network. In the second layer, the network further analyzes the user's whole-body posture characteristics. The core of this step is to integrate the posture data of various parts of the body and identify the posture coordination within the whole body. Through this layer of analysis, the system can detect deviations or imbalances in the overall posture, such as the pressure on one side of the body is significantly higher than the other side, or there is a significant difference between the whole-body posture and the ideal state. The generated second-layer posture analysis data provides a basis for further adjustment of the whole-body posture.

[0081] The system inputs the second-layer posture analysis data into the third layer of the hybrid graph neural network, where the network conducts an in-depth analysis of the overall coordination of the postures of various parts of the user's body. The third layer is not just a simple posture feature extraction, but adjusts the output data of the first and second layers through a dynamic feedback loop to ensure that the final posture analysis results are highly accurate and consistent. Through this level of analysis, the system is able to generate the final posture risk assessment data, which comprehensively considers the local characteristics of the user's posture, the coordination of the whole body, and the overall stability, thereby providing a scientific basis for the posture adjustment strategy.

[0082] Example: Assume that a user gradually leans forward in an office environment for a long time, resulting in increased pressure on the cervical spine and shoulders. The system first collects the user's comprehensive posture data through pressure sensors and posture angle sensors. Next, persistent coherence calculation analyzes the long life cycle characteristics of the user in the neck area, prompting the user to maintain a bad posture for a long time. Subsequently, the first layer of the hybrid graph neural network identifies the local cervical spondylosis and excessive shoulder flexion posture abnormalities, and the second layer further discovers the imbalance of the whole body posture, especially the increased burden on the spine caused by the forward leaning of the upper body. The third layer corrects the analysis data of the first two layers through dynamic feedback, and finally generates accurate posture risk assessment data, prompting the user that this posture may cause long-term cervical and shoulder problems. Based on this, the system recommends that the user adjust his posture and implements appropriate posture correction measures through the intelligent material adjustment module.

[0083] Preferably, the step of generating posture risk assessment data comprises:

[0084] Compare the third-layer posture analysis data with the user's historical posture data to identify potential posture problems and long-term health risks by comparing the difference between the user's current posture and historical data;

[0085] Based on the results of the comparative analysis, posture risk assessment data including the user's posture health risk level and recommended posture adjustment measures are generated to guide the adjustment of smart materials.

[0086] The system characterizes the user's current sitting posture through the posture analysis data generated by the third layer of the hybrid graph neural network. These data contain various aspects of the user's current posture, including local features, whole-body coordination, and overall stability. Next, the system compares and analyzes these third-layer posture analysis data with the user's historical posture data. Historical posture data is accumulated by the system during long-term monitoring, and can reflect the changing trends and long-term posture characteristics of the user's daily sitting posture. By comparing the current posture with historical data, the system can identify abnormal changes in the user's posture, such as sudden deviations of certain posture features or gradual deterioration of long-term posture.

[0087] The core principle of this comparative analysis is to detect the differences between posture features and evaluate potential posture problems by quantifying these differences. For example, if the system finds that the user's current posture deviates significantly from historical data at multiple key points, such as increased back curvature and increased cervical forward tilt, these may indicate that the user is developing a bad sitting habit. Furthermore, the system can predict long-term health risks through the cumulative effects of these deviations. For example, long-term poor posture may lead to spinal problems, shoulder and neck pain, etc. The system will assess the severity of the risk by calculating the magnitude and frequency of deviations.

[0088] After identifying potential posture problems, the system generates posture risk assessment data based on the results of comparative analysis. This data includes not only the health risk level of the user's posture, but also specific posture adjustment suggestions for these risks. The posture health risk level is usually represented by a quantitative indicator. The system classifies the risks based on the results of comparative analysis, for example, into three levels: low, medium, and high, each representing a different degree of health risk. For high-risk situations, the system will specifically point out and remind the user that they need to make immediate posture adjustments.

[0089] At the same time, the posture adjustment suggestions generated by the system are based on a comprehensive analysis of the user's historical data and current posture data to ensure the personalization and accuracy of the suggestions. For example, if the system detects that the user's cervical spine is tilted forward at a large angle and lasts for a long time, it may suggest that the user lean back slightly and achieve this posture change through the adjustment of smart materials. This adjustment suggestion can not only alleviate the adverse effects of the current posture, but also prevent possible health problems in the future.

[0090] Example: A user works at a desk for a long time. His historical posture data shows that he usually maintains a good sitting posture, with his back straight and shoulders relaxed. However, the latest third-layer posture analysis data shows that the user's shoulders are gradually bending forward and the back is more curved. Through comparative analysis, the system finds that the deviation between these changes and historical data is gradually increasing. Through risk assessment, the system marks this situation as moderate risk, generates posture risk assessment data, and recommends that the user tilt his back slightly backward and adjust the back support strength of the seat to help the user return to a healthy sitting posture.

[0091] Through this process, the system effectively identifies and warns users of potential posture problems and provides specific measures based on intelligent material adjustments to avoid the negative impact of long-term poor posture on health. This approach not only takes preventive measures before problems occur, but also helps users develop good sitting habits, thereby reducing health problems caused by long-term poor posture.

[0092] Based on the posture risk assessment data, the intelligent material collaborative adjustment strategy is generated through the nonlinear dynamics and adaptive shape coupling model; according to the intelligent material collaborative adjustment strategy, the shape of the liquid metal and programmable material in the smart seat is adjusted, the posture correction is implemented, and the adjustment feedback data is generated;

[0093] Liquid metal in the present invention is a material that has fluidity and can change its shape under external control. Generally, this type of material refers to metals based on gallium or gallium alloys (such as gallium-indium-tin alloy, Galinstan), which are liquid at room temperature and have good electrical and thermal conductivity. In addition, liquid metal has extremely strong plasticity and can change its shape and position under the action of external electric field, magnetic field or mechanical force. This unique property makes liquid metal have broad application prospects in flexible electronics, soft robots and smart seats.

[0094] In the present invention, liquid metal is applied to the structure of the smart seat, mainly used to dynamically adjust the shape of the seat to adapt to the user's posture requirements. For example, when the user's posture data indicates that the back needs more support, the system will adjust the position and shape of the liquid metal to form a support structure on the back of the seat, thereby dispersing pressure and increasing comfort. The deformability of liquid metal enables the seat to provide the best support effect under different user postures, which can not only maintain the user's healthy posture but also improve comfort.

[0095] Application example: Suppose a user works leaning forward for a long time, and the system detects that his back posture needs to be adjusted. Liquid metal will form an arched support structure in the back area of ​​the chair. As the user moves, the liquid metal can adaptively adjust its shape to ensure that the user's back is always properly supported, thereby reducing back fatigue and discomfort. This dynamic support capability makes liquid metal an indispensable part of smart chairs.

[0096] Programmable materials are smart materials that can change their physical properties (such as hardness, elasticity, shape, etc.) under external stimuli (such as electric fields, magnetic fields, temperature changes, etc.). Common programmable materials include shape memory alloys (Shape Memory Alloys, SMA), shape memory polymers (Shape Memory Polymers, SMP), electrorheological fluids (Electrorheological Fluids, ERF) and magnetorheological fluids (Magnetorheological Fluids, MRF).

[0097] In the present invention, programmable materials are mainly used in the structure of smart chairs, and the hardness and shape of the materials are adjusted to respond to the posture adjustment strategy of the system. For example, when the system detects that the user's posture requires more lumbar support, the programmable material can increase the support for the lumbar area by changing its hardness and shape. These materials can respond quickly to system instructions, allowing the smart chair to provide dynamic adjustments based on the user's real-time posture changes.

[0098] Application example: In an office environment, users may experience waist discomfort due to maintaining a fixed posture for a long time. The system can instruct the programmable material to increase the hardness in the waist area to provide additional support for the user. Specifically, the shape memory alloy can change its shape after being stimulated by electric current, thereby forming a harder and more supportive area on the user's waist. The programmability of this material allows it to be adjusted at any time according to user needs, ensuring that users can get the best support and comfort in different postures.

[0099] The application of liquid metal and programmable materials in the present invention enables the smart seat to be adjusted in real time and dynamically according to the user's posture requirements. The deformability and response characteristics of these materials enable the seat to provide customized support and reduce health problems caused by long-term poor sitting posture. By combining posture risk assessment data, these smart materials can actively adjust to optimize the user's sitting posture and improve overall comfort and health level. This dynamic adjustment solution based on smart materials not only innovatively solves the problem that traditional seats cannot dynamically respond to user needs, but also provides an effective technical path for personalized health management in the future.

[0100] Preferably, the step of generating a smart material collaborative adjustment strategy through a nonlinear dynamics and adaptive shape coupling model based on posture risk assessment data includes:

[0101] Based on the posture risk assessment data, a nonlinear dynamics model is used to calculate the potential change trend of the user's posture, and dynamically adjusted prediction data is generated. The dynamically adjusted prediction data is used to predict the possible posture changes of the user in the next period of time. The calculation formula of the dynamically adjusted prediction data is as follows:

[0102]

[0103] in, Indicates time The state vector at the moment; represents the initial state vector; Kinetic equations representing changes of state; Indicates at time The state vector at the moment; Indicates at time The control input vector at time t; represents the integral variable, which means from the initial time To current time at any moment; through this formula, the system can predict the user's posture changes in the future, and the prediction result is the dynamically adjusted prediction data.

[0104] The dynamic adjustment prediction data is input into the adaptive shape coupling model, and the adaptive shape coupling model generates an intelligent material collaborative adjustment strategy according to the dynamic adjustment prediction data, wherein the strategy includes adjustment instructions for the liquid metal morphology and adjustment instructions for the hardness and morphology of the programmable material.

[0105] Based on the previously generated posture risk assessment data, the system calculates the potential change trend of the user's posture through a nonlinear dynamics model. The role of the nonlinear dynamics model is to describe and predict the changes in states in a complex system, including the changes in the user's sitting posture at different time points.

[0106] In this prediction process, the kinetic equation Describes how the user's posture state is affected by the current posture and control input The control input can include adjustment instructions for the smart seat, such as adjusting the seat back tilt angle, changing the pressure distribution of the seat cushion, etc. This prediction can identify the possible development trend of the user's posture, such as the formation or deterioration of a certain bad posture, thus providing a basis for subsequent adjustments.

[0107] After the dynamic adjustment prediction data is generated, the system inputs it into the adaptive shape coupling model. The core function of the adaptive shape coupling model is to convert the predicted posture changes into specific material adjustment strategies. By analyzing the dynamic adjustment prediction data, the adaptive shape coupling model can calculate the optimal adjustment method, so that smart materials (such as liquid metal and programmable materials) can adapt to the user's posture changes in real time.

[0108] For example, if the dynamic adjustment prediction data indicates that the user is about to lean forward, resulting in increased pressure on the back and neck, the adaptive shape coupling model will generate a corresponding intelligent material collaborative adjustment strategy. This strategy may include the following: first, by adjusting the shape of the liquid metal, the back of the seat can better support the user's back and prevent excessive forward leaning; second, adjust the hardness and shape of the programmable material to increase the support for the waist to disperse the pressure. These adjustment instructions are based on the prediction results of the dynamic model, aiming to prevent the formation of bad postures and maintain the user's healthy posture.

[0109] Example: When a user is working at a desk, the system detects that the user's posture is gradually leaning forward, which may cause health problems for the cervical and lumbar spine over a long period of time. Through the nonlinear dynamic model, the system predicts that this forward leaning trend may intensify, so it generates dynamic adjustment prediction data. Based on this data, the adaptive shape coupling model generates a series of adjustment strategies: the liquid metal will form a dynamic support structure on the back of the chair, gradually increasing the support force for the user's back to prevent it from continuing to lean forward; at the same time, the hardness of the programmable material will increase to support the user's waist and evenly distribute his weight. These adjustments effectively correct the user's sitting posture, thereby avoiding possible health problems.

[0110] By combining the nonlinear dynamic model and the adaptive shape coupling model, the sitting posture monitoring system of the present invention can predict the user's posture change trend in real time, and dynamically adjust the shape of the smart material accordingly. Such a design can not only correct the user's bad posture in time, but also avoid the damage to health caused by long-term bad posture through early prevention. The advantage of this method lies in its predictive and adaptive nature, which enables the system to take measures before the posture problem has a serious impact on health, thereby improving the user's comfort and health level. This intelligent adjustment strategy is particularly effective in long-term office environments and can significantly improve the user's posture health management experience.

[0111] like Figure 2 As shown, preferably, the smart material collaborative adjustment strategy includes:

[0112] According to the smart material collaborative adjustment strategy, the liquid metal in the smart seat is adjusted to make its shape meet the user's current and predicted posture requirements, forming a support structure that adapts to the user's posture;

[0113] At the same time, the programmable material in the smart seat is adjusted, including its hardness and shape, to enhance the coordination with the change of liquid metal shape, thereby optimizing the overall support effect;

[0114] Generate adjustment feedback data, which is used to evaluate the actual effect of the smart material adjustment and provide a reference for subsequent adjustments.

[0115] The intelligent material collaborative adjustment strategy is based on the system's assessment of the user's current posture and its potential changes. These assessment data are provided by the posture risk assessment module, and the system uses these data to predict the user's posture requirements in the future. Based on this prediction, the system will instruct the liquid metal to make corresponding shape adjustments. Due to its plasticity and fluidity, liquid metal can quickly change its shape under the control of an electric or magnetic field, thereby forming a support structure that meets the user's current or predicted posture requirements. For example, when the user tends to lean forward, the liquid metal can form a raised support point on the back of the seat to prevent the user from continuing to lean forward and reduce pressure on the cervical and lumbar vertebrae.

[0116] At the same time, the system will adjust the programmable materials in the smart seat. The core characteristics of programmable materials are their adjustable hardness and shape, which enables them to effectively cooperate with the morphological changes of liquid metal. For example, when liquid metal forms support on the back, the programmable material can increase the hardness in this area, providing stronger support force to prevent the user's posture from deviating from the preset healthy state. In this way, the collaborative work of liquid metal and programmable materials can continuously optimize the user's sitting posture in a dynamic environment, ensuring that the support effect under different posture requirements is always in the best state.

[0117] Example: Assume that the user habitually leans forward at work, and the system predicts that this posture may cause long-term shoulder and neck pain. The system first uses a liquid metal adjustment strategy to form a moderate support bulge on the back of the seat to limit the extent of the user's forward leaning. In order to enhance the effect of this adjustment, the hardness of the programmable material will increase on the basis of the liquid metal support point to ensure the stability and durability of the support structure. In addition, the system may also appropriately reduce the hardness in the lumbar area of ​​the seat to encourage the user to adjust his sitting posture and avoid excessive forward leaning. This collaborative adjustment strategy of intelligent materials can not only respond to the user's current posture needs in a timely manner, but also prevent potential health risks by guiding and correcting the posture.

[0118] After completing the above adjustments, the system will generate adjustment feedback data in real time. The generation of adjustment feedback data is achieved through a series of sensors and monitoring modules, which continuously monitor whether the adjusted posture meets the expected health standards. If the user's posture still deviates after adjustment, the system will readjust the shape and hardness of the smart material based on the feedback data to further optimize the support structure. Adjustment feedback data can also be used for long-term analysis to help the system gradually improve the adjustment strategy to adapt to the user's personalized needs.

[0119] Generate dynamic virtual posture data using adjustment feedback data through generative adversarial networks, and generate simulated posture feedback data in the adaptive posture simulation module; optimize the adjustment strategy using simulated posture feedback data and adjustment feedback data through the collaborative reinforcement learning module, generate optimized adjustment strategy data, and implement further intelligent material adjustment;

[0120] like Figure 3 As shown, preferably, the step of generating dynamic virtual posture data by adjusting feedback data through a generative adversarial network includes:

[0121] Using the adjustment feedback data and the user's historical posture data to perform adversarial training through a generative adversarial network to generate dynamic virtual posture data that is highly matched with the actual situation, and the dynamic virtual posture data is used to simulate the user's possible reactions under different posture adjustments;

[0122] The dynamic virtual posture data is input into the adaptive posture simulation module to generate simulated posture feedback data, which is used to correct and optimize the further adjustment strategy of the smart material.

[0123] The system processes the adjustment feedback data and the user's historical posture data through a generative adversarial network (GAN). A generative adversarial network is a deep learning model that consists of two adversarial neural networks, a generator and a discriminator. The generator is responsible for generating new data samples based on the input data, while the discriminator is responsible for distinguishing the generated data samples from the real data samples. In the present invention, the generator uses the user's historical posture data and adjustment feedback data to generate possible dynamic virtual posture data, while the discriminator evaluates the degree of match between these generated data and the real adjustment feedback data. Through this adversarial training, the generator can gradually generate dynamic virtual posture data that is highly matched with the user's actual situation.

[0124] These dynamic virtual posture data are of great significance. They not only reflect the posture state of the user under the current adjustment, but also predict the possible posture changes of the user under different adjustment strategies. The core of this process is that through the adversarial training of GAN, the system can simulate the user's response under various posture adjustment strategies, thereby providing reliable data support for optimizing the adjustment strategy. For example, if the system detects that a certain posture adjustment strategy causes the user's posture to deviate further in the simulation, the strategy will be judged as unsatisfactory, and the system will try other strategies to achieve better adjustment results.

[0125] Next, the system inputs the generated dynamic virtual posture data into the adaptive posture simulation module. The function of the adaptive posture simulation module is to perform further posture simulation based on the virtual posture data and generate simulated posture feedback data. These feedback data reflect the actual changes in the user's posture under different adjustment strategies and are used to evaluate the adjustment effect of the smart material. By analyzing the simulated posture feedback data, the system can correct the existing adjustment strategy and optimize the morphological adjustment instructions of the smart material in a targeted manner to ensure that the final posture adjustment can achieve the best effect.

[0126] Example: Assume that when a user is using a smart chair, the system detects that there is a potential risk in his current sitting posture, such as long-term forward shoulder leaning, which may cause shoulder and neck pain. The system generates a series of possible dynamic virtual posture data by using the user's historical posture data and real-time adjustment feedback data through a generative adversarial network. These virtual data simulate the user's posture response under different adjustment strategies. The generative adversarial network continuously optimizes these virtual posture data through adversarial training to make them highly match the user's actual situation.

[0127] After inputting these dynamic virtual posture data into the adaptive posture simulation module, the system further simulates the user's posture changes under different adjustment strategies. For example, a strategy may involve increasing the support force of the seat back to reduce the forward shoulder tilt. The simulation results show that this adjustment can effectively improve the user's posture and reduce shoulder pressure. Based on these simulated posture feedback data, the system finally determines the optimal adjustment strategy and implements this strategy through real-time adjustment of smart materials.

[0128] Preferably, the step of optimizing the adjustment strategy by using the simulated posture feedback data and the adjustment feedback data through the collaborative reinforcement learning module includes:

[0129] The collaborative reinforcement learning module realizes the collaborative optimization of multiple agents in the adjustment strategy of each part of the smart seat through joint training of the simulated posture feedback data and the adjustment feedback data. The agents include the liquid metal adjustment module, the programmable material adjustment module and the posture sensing module in the seat. The agents of each part coordinate and cooperate with each other during the execution process to generate optimized adjustment strategy data.

[0130] The optimized adjustment strategy data is applied to further adjust the smart materials, so that the smart seat can achieve optimal support and posture correction for the user under complex posture conditions.

[0131] The core of the collaborative reinforcement learning module lies in the application of reinforcement learning (RL). Reinforcement learning is a machine learning method that enables an agent to learn the optimal behavior strategy through continuous interaction with the environment. Specifically, in the present invention, multiple modules (agents) of the smart seat continuously optimize their adjustment strategies through collaborative reinforcement learning to adapt to the user's real-time posture requirements.

[0132] The agents in the system include:

[0133] Liquid metal adjustment module: responsible for adjusting the shape of liquid metal to provide dynamic support.

[0134] Programmable material adjustment module: responsible for adjusting the hardness and shape of programmable materials to enhance the support effect.

[0135] Posture sensing module: responsible for real-time monitoring of user's posture data and providing feedback information.

[0136] These agents are optimized through collaborative reinforcement learning. The uniqueness of collaborative reinforcement learning is that multiple agents do not operate independently, but through joint training, they coordinate with each other and jointly optimize the adjustment strategy. For example, when the liquid metal adjustment module and the programmable material adjustment module perform adjustments, they need to consider each other's state and action at the same time to ensure that the final support effect is optimal. Through the joint training of simulation posture feedback data and adjustment feedback data, the system can continuously optimize the collaborative working mode between modules and form a more effective adjustment strategy.

[0137] In this process, simulated posture feedback data and adjustment feedback data are important inputs for reinforcement learning. These data come from the system's simulation of the user's response to different posture adjustments and the posture monitoring results after actual adjustments. Through repeated training of these data, the collaborative reinforcement learning module can identify the effects of different adjustment strategies and gradually optimize them, so that the smart seat can provide the best support and correction effects under various complex posture conditions.

[0138] Example: Assume that a user gradually leans to one side during a long period of office work. The system detects this trend through the posture sensing module and makes preliminary adjustments through the liquid metal adjustment module to try to correct this posture. However, the adjustment of a single module may not be enough to achieve the desired effect, especially when the user's tilted posture is more stubborn. At this time, the collaborative reinforcement learning module will be started. Through the joint training of simulated posture feedback data and adjustment feedback data, the system finds that the adjustment of liquid metal alone cannot completely correct the user's tilted posture. Therefore, the collaborative reinforcement learning module instructs the programmable material adjustment module to increase the hardness, and at the same time optimizes the support form of the liquid metal adjustment module. Under the joint action of the two, the user's posture is effectively corrected.

[0139] This collaborative optimization mechanism can significantly improve the system's response speed and adjustment effect. Compared with traditional single module adjustment, the multi-agent collaborative optimization of the present invention can provide more accurate support and correction under complex and dynamic posture conditions, thereby effectively preventing and alleviating health problems caused by poor sitting posture. This optimization mechanism is particularly suitable for long-term, high-intensity working environments, and can provide users with personalized health posture management services.

[0140] Generate personalized posture evaluation data based on the optimization adjustment strategy data and historical adjustment data; formulate a personalized posture maintenance plan based on the personalized posture evaluation data, and perform dynamic adjustments and regularly generate posture reports based on the personalized posture maintenance plan.

[0141] Preferably, the step of generating personalized posture assessment data according to the optimization adjustment strategy data and the historical adjustment data includes:

[0142] Summarize optimization adjustment strategy data and historical adjustment data, and identify users' long-term posture changes and potential health problems by analyzing users' long-term posture change patterns and health trends;

[0143] Based on the analysis results, personalized posture assessment data is generated, where the personalized posture assessment data includes the user's current posture health status, possible health risks, and personalized posture improvement suggestions.

[0144] The system comprehensively analyzes the user's long-term posture change pattern by summarizing the optimization adjustment strategy data and historical adjustment data. The optimization adjustment strategy data is generated by the system after each posture adjustment. These data reflect the system's adjustment strategy and its effect based on the user's current posture. The historical adjustment data is the posture information accumulated by the system during the long-term monitoring process, including the user's posture status and adjustment records under various conditions. The comprehensive analysis of these data can reveal the user's posture change trajectory and health trends in different periods, thereby identifying potential long-term health problems. For example, if the system detects that the user's posture has gradually leaned forward in recent months, this may indicate that the user is forming a bad posture habit, which may lead to cervical problems.

[0145] Through in-depth analysis of these data, the system generates personalized posture assessment data. This assessment data includes the user's current posture health status, possible health risks, and personalized posture improvement suggestions for these problems. For example, if the system finds that the user has maintained a certain bad posture for a long time, the assessment data may point out the health risks that this posture may cause, such as shoulder and neck pain, increased lumbar burden, etc. At the same time, the system will provide specific posture improvement suggestions, such as suggesting that users adjust the seat height, increase back support, etc. These suggestions are not only based on the current posture analysis, but also combined with the user's historical posture data to ensure the effectiveness and pertinence of the suggestions.

[0146] After generating personalized posture assessment data, the system will further develop a personalized posture maintenance plan. This plan is based on the assessment data and includes a series of dynamic adjustment measures and regular inspection strategies to help users maintain a healthy sitting posture in daily life. For example, when the system detects that the user's posture deviates from the healthy standard, it may automatically adjust the support structure of the smart seat, or remind the user to change the posture when the user maintains a certain posture for a long time. In addition, the system will regularly generate posture reports to summarize the user's posture changes, adjustment records, and health status. These reports can help users better understand their posture health and make corresponding adjustments based on the recommendations in the report.

[0147] Example: Assume that a user is working in an office for a long time. The system finds through historical adjustment data that the user's sitting posture is gradually leaning forward, and has shown a trend of increasing neck pressure in recent months. Based on this trend, the personalized posture assessment data generated by the system points out the cervical health risks that the user may face, and recommends that the user increase back support and avoid leaning forward for a long time. Subsequently, the system develops a personalized posture maintenance plan based on this assessment data, automatically adjusts the liquid metal and programmable material of the seat to increase back support, and issues reminders when the user leans forward for a long time, suggesting that he adjust his posture. One month later, the system generated a posture report showing that the user's forward posture had improved and the neck pressure had been reduced, verifying the effectiveness of the personalized maintenance plan.

[0148] like Figure 4 As shown, a sitting posture monitoring system is used to implement the sitting posture monitoring method, comprising:

[0149] The multimodal sensing module includes a pressure sensor, an inertial measurement unit sensor, an electromyographic sensor, and a temperature sensor. The multimodal sensing module is used to collect the user's pressure distribution data, posture angle data, electromyographic signal data, and temperature distribution data, and perform time synchronization correction on the collected data to generate comprehensive posture data; the multimodal sensing module performs time synchronization correction on these data and generates comprehensive posture data through data fusion. This data provides basic information for subsequent analysis and adjustment, ensuring that the system's perception of the user's posture is comprehensive and accurate.

[0150] The data processing module includes a topological data analysis unit and a hybrid graph neural network unit. The topological data analysis unit is used to extract topological features from the comprehensive posture data and generate posture topological feature data. The hybrid graph neural network unit is used to perform multi-level posture analysis based on the posture topological feature data and generate posture risk assessment data. In this way, the system can identify potential problems in the user's posture and quantify risks, providing a basis for intelligent adjustment.

[0151] A nonlinear dynamics and adaptive shape coupling module is used to generate dynamic adjustment prediction data through a nonlinear dynamics model according to posture risk assessment data, and generate an intelligent material collaborative adjustment strategy based on the dynamic adjustment prediction data. The intelligent material collaborative adjustment strategy includes adjustment instructions for the liquid metal shape and adjustment instructions for the hardness and shape of the programmable material. The system generates an intelligent material collaborative adjustment strategy based on these prediction data, and instructs the liquid metal and programmable material to perform adaptive adjustments in shape and hardness to correct the user's bad posture.

[0152] The intelligent material adjustment module includes a liquid metal adjustment unit and a programmable material adjustment unit. The liquid metal adjustment unit is used to adjust the shape of the liquid metal, and the programmable material adjustment unit is used to adjust the hardness and shape of the programmable material to implement posture correction and generate adjustment feedback data; the generated adjustment feedback data is used to evaluate the adjustment effect and provide data support for subsequent optimization.

[0153] Generate adversarial network and adaptive posture simulation module, the module is used to generate dynamic virtual posture data using adjustment feedback data through a generative adversarial network, and generate simulated posture feedback data through adaptive posture simulation; these feedback data further optimize the system's adjustment strategy.

[0154] A collaborative reinforcement learning module is used to achieve multi-agent collaborative optimization through joint training of simulated posture feedback data and adjustment feedback data. The intelligent agents include a liquid metal adjustment module, a programmable material adjustment module and a posture sensing module, which generate optimized adjustment strategy data and apply the optimized adjustment strategy data to further adjustment of the intelligent material; each intelligent agent (such as the liquid metal adjustment module, the programmable material adjustment module and the posture sensing module) cooperates with each other to generate optimized adjustment strategy data and guide the intelligent material to make further adjustments.

[0155] Personalized posture assessment and maintenance module, which is used to summarize the optimization adjustment strategy data and historical adjustment data, generate personalized posture assessment data, and formulate a personalized posture maintenance plan based on the personalized posture assessment data, while performing dynamic adjustment and generating regular posture reports. Based on these data, the system formulates a personalized posture maintenance plan, and through dynamic adjustment and regular generation of posture reports, helps users maintain a healthy sitting posture for a long time.

[0156] A smart chair is provided with the sitting posture monitoring system.

[0157] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware.

[0158] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A sitting posture monitoring method, characterized in that: The following steps are involved: Collect the user's pressure distribution data, posture angle data, electromyographic signal data and temperature distribution data; perform time synchronization correction on the collected data, and perform multi-modal fusion to generate comprehensive posture data; Based on the hybrid graph neural network and topological analysis, the comprehensive posture data is analyzed at multiple levels, the posture topological features are extracted, and the posture topological data is generated; the posture topological data is input into the multi-layer structure of the hybrid graph neural network, and the posture risk assessment is performed through a dynamic feedback loop to generate posture risk assessment data; The steps of multi-level analysis of comprehensive posture data based on hybrid graph neural network and topological analysis include: Performing topological data analysis on the comprehensive posture data, extracting shape features in the user's sitting posture through persistent homology calculation, and generating posture topological feature data; the persistent homology calculation includes calculating the homology group of a single sitting posture and calculating its life cycle to identify features with a long life cycle; The posture topology feature data is used as input and sent to the first layer of the hybrid graph neural network. The first layer of the hybrid graph neural network is used to analyze the local posture features and generate the first layer of posture analysis data containing local posture information; The first layer of posture analysis data is input into the second layer of the hybrid graph neural network, and the second layer of the hybrid graph neural network is used to analyze the user's whole body posture characteristics and generate the second layer of posture analysis data containing the whole body posture information; The posture analysis data of the second layer is input into the third layer of the hybrid graph neural network. The third layer of the hybrid graph neural network is used to analyze the overall coordination of the postures of various parts of the user's body, and adjust the data of the first and second layers through a dynamic feedback loop to finally generate posture risk assessment data; Based on the posture risk assessment data, the intelligent material collaborative adjustment strategy is generated through the nonlinear dynamics and adaptive shape coupling model; according to the intelligent material collaborative adjustment strategy, the shape of the liquid metal and programmable material in the smart seat is adjusted, the posture correction is implemented, and the adjustment feedback data is generated; Generate dynamic virtual posture data using adjustment feedback data through generative adversarial networks, and generate simulated posture feedback data in the adaptive posture simulation module; optimize the adjustment strategy using simulated posture feedback data and adjustment feedback data through the collaborative reinforcement learning module, generate optimized adjustment strategy data, and implement further intelligent material adjustment; Generate personalized posture evaluation data based on the optimization adjustment strategy data and historical adjustment data; formulate a personalized posture maintenance plan based on the personalized posture evaluation data, and perform dynamic adjustments and regularly generate posture reports based on the personalized posture maintenance plan.

2. The sitting posture monitoring method according to claim 1, characterized in that: The persistent coherence calculation formula is as follows: in, Indicates dimensional homology group; Indicates Boundary operators of layers; represents the kernel space of the boundary operator; Represents the image space of the previous layer boundary operator.

3. The sitting posture monitoring method according to claim 1, characterized in that: The steps to generate posture risk assessment data include: Compare the third-layer posture analysis data with the user's historical posture data to identify potential posture problems and long-term health risks by comparing the difference between the user's current posture and historical data; Based on the results of the comparative analysis, posture risk assessment data including the user's posture health risk level and recommended posture adjustment measures are generated to guide the adjustment of smart materials.

4. The sitting posture monitoring method according to claim 1, characterized in that: Based on the posture risk assessment data, the steps of generating the intelligent material collaborative adjustment strategy through the nonlinear dynamics and adaptive shape coupling model include: Based on the posture risk assessment data, a nonlinear dynamics model is used to calculate the potential change trend of the user's posture, and dynamically adjusted prediction data is generated. The dynamically adjusted prediction data is used to predict the possible posture changes of the user in the next period of time. The calculation formula of the dynamically adjusted prediction data is as follows: in, Indicates time The state vector at the moment; represents the initial state vector; Kinetic equations representing changes of state; Indicates at time The state vector at the moment; Indicates at time The control input vector at time t; represents the integral variable, which means from the initial time To current time any time; The dynamic adjustment prediction data is input into the adaptive shape coupling model, and the adaptive shape coupling model generates an intelligent material collaborative adjustment strategy according to the dynamic adjustment prediction data, wherein the strategy includes adjustment instructions for the liquid metal morphology and adjustment instructions for the hardness and morphology of the programmable material.

5. The sitting posture monitoring method according to claim 1, characterized in that: Smart material coordination strategies include: According to the smart material collaborative adjustment strategy, the liquid metal in the smart seat is adjusted to make its shape meet the user's current and predicted posture requirements, forming a support structure that adapts to the user's posture; At the same time, the programmable material in the smart seat is adjusted, including its hardness and shape, to enhance the coordination with the change of liquid metal shape, thereby optimizing the overall support effect; Generate adjustment feedback data, which is used to evaluate the actual effect of the smart material adjustment and provide a reference for subsequent adjustments.

6. The sitting posture monitoring method according to claim 1, characterized in that: The steps of generating dynamic virtual posture data by adjusting feedback data through a generative adversarial network include: Using the adjustment feedback data and the user's historical posture data to perform adversarial training through a generative adversarial network to generate dynamic virtual posture data that is highly matched with the actual situation, and the dynamic virtual posture data is used to simulate the user's possible reactions under different posture adjustments; The dynamic virtual posture data is input into the adaptive posture simulation module to generate simulated posture feedback data, which is used to correct and optimize the further adjustment strategy of the smart material.

7. The sitting posture monitoring method according to claim 1, characterized in that: The steps of optimizing the adjustment strategy by using the simulated posture feedback data and the adjustment feedback data through the collaborative reinforcement learning module include: The collaborative reinforcement learning module realizes the collaborative optimization of multiple agents in the adjustment strategy of each part of the smart seat through joint training of the simulated posture feedback data and the adjustment feedback data. The agents include the liquid metal adjustment module, the programmable material adjustment module and the posture sensing module in the seat. The agents of each part coordinate and cooperate with each other during the execution process to generate optimized adjustment strategy data. The optimized adjustment strategy data is applied to further adjust the smart materials, so that the smart seat can achieve optimal support and posture correction for the user under complex posture conditions.

8. The sitting posture monitoring method according to claim 1, characterized in that: The steps of generating personalized posture evaluation data based on the optimization adjustment strategy data and the historical adjustment data include: Summarize optimization adjustment strategy data and historical adjustment data, and identify users' long-term posture changes and potential health problems by analyzing users' long-term posture change patterns and health trends; Based on the analysis results, personalized posture assessment data is generated, where the personalized posture assessment data includes the user's current posture health status, possible health risks, and personalized posture improvement suggestions.

9. A sitting posture monitoring system, used to implement the sitting posture monitoring method according to any one of claims 1 to 8, characterized in that: include: A multimodal sensing module, including a pressure sensor, an inertial measurement unit sensor, an electromyographic sensor and a temperature sensor, wherein the multimodal sensing module is used to collect pressure distribution data, posture angle data, electromyographic signal data and temperature distribution data of the user, and to perform time synchronization correction on the collected data to generate comprehensive posture data; A data processing module, comprising a topological data analysis unit and a hybrid graph neural network unit, wherein the topological data analysis unit is used to extract topological features from the comprehensive posture data to generate posture topological feature data, and the hybrid graph neural network unit is used to perform multi-level posture analysis based on the posture topological feature data to generate posture risk assessment data; A nonlinear dynamics and adaptive shape coupling module, the module is used to generate dynamic adjustment prediction data through a nonlinear dynamics model according to the posture risk assessment data, and generate an intelligent material collaborative adjustment strategy based on the dynamic adjustment prediction data, the intelligent material collaborative adjustment strategy includes adjustment instructions for the liquid metal morphology and adjustment instructions for the hardness and morphology of the programmable material; An intelligent material adjustment module, comprising a liquid metal adjustment unit and a programmable material adjustment unit, wherein the liquid metal adjustment unit is used to adjust the shape of the liquid metal, and the programmable material adjustment unit is used to adjust the hardness and shape of the programmable material to implement posture correction and generate adjustment feedback data; A generative adversarial network and adaptive posture simulation module, the module is used to generate dynamic virtual posture data using the adjustment feedback data through a generative adversarial network, and to generate simulated posture feedback data through adaptive posture simulation; A collaborative reinforcement learning module, which is used to achieve multi-agent collaborative optimization by jointly training the simulation posture feedback data and the adjustment feedback data. The intelligent agent includes a liquid metal adjustment module, a programmable material adjustment module and a posture sensing module, generates optimization adjustment strategy data, and applies the optimization adjustment strategy data to further adjustment of the intelligent material; A personalized posture evaluation and maintenance module is used to summarize the optimization adjustment strategy data and historical adjustment data, generate personalized posture evaluation data, and formulate a personalized posture maintenance plan based on the personalized posture evaluation data, while performing dynamic adjustments and generating regular posture reports.

10. A smart seat, characterized in that: The smart seat is equipped with the sitting posture monitoring system described in claim 9.

Images