Intelligent furniture self-adaptive adjusting system based on user behavior perception

Abstract

The invention relates to the technical field of intelligent furniture, in particular to an intelligent furniture self-adaptive adjusting system based on user behavior perception, which is characterized in that pressure distribution data and action frequency data of a user are synchronously acquired through a data acquisition module; the data processing module extracts pressure distribution mode characteristics and action mode characteristics according to the pressure distribution mode characteristics and the action mode characteristics, and generates a comprehensive user state characterization quantity through correlation analysis; the strategy adjustment module generates a fatigue index based on the characterization quantity, and dynamically decides the adjustment triggering opportunity and the adjustment parameter change gradient in combination with the real-time change trend of the pressure distribution mode characteristics; according to the process, a closed loop from multi-modal behavior data fusion analysis to intelligent adjustment strategy generation is realized, the accuracy of fatigue state recognition is effectively improved, and the timeliness and the seaworthiness of adaptive adjustment are ensured.

Description

Technical Field

[0001] This invention relates to the field of smart furniture technology, specifically to a smart furniture adaptive adjustment system based on user behavior perception. Background Technology

[0002] In the field of smart furniture technology, especially in systems involving user comfort adjustment, existing solutions typically rely on pre-set programs or simple external triggers to achieve functional adjustments. While these methods provide some degree of automation, they fall short when dealing with the complex state of user fatigue. Fatigue, a key factor affecting user work efficiency and health, manifests differently in individuals and dynamically changes over time. Current technologies lack multi-dimensional and continuous monitoring of user behavior, failing to extract effective state indicators from subtle changes in movement and pressure distribution. For example, relying solely on pressure sensors may misinterpret a brief postural adjustment as a fatigue signal, while relying solely on accelerometers may ignore information contained in static pressure distribution. This limitation of a single data source makes it difficult for the system to accurately distinguish whether a user is in an active, resting, or fatigued state.

[0003] Furthermore, even when systems attempt to combine multiple sensors, data processing often remains at the level of simple threshold comparisons, failing to achieve deep feature fusion and resulting in insufficient accuracy in state recognition. Consequently, adjustment strategies lack specificity and may not intervene at the optimal time and with the most appropriate intensity, thus reducing the system's practicality and user satisfaction. Therefore, developing a system capable of intelligently identifying fatigue and adaptively adjusting based on multi-source behavioral data has become an urgent problem to be solved. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent furniture adaptive adjustment system based on user behavior perception to solve the problems mentioned in the background art. Specific technical problems include how to accurately generate a fatigue index based on multimodal behavior data to solve the problems of accuracy in fatigue state identification and dynamic optimization of adjustment strategies in intelligent furniture adaptive adjustment.

[0005] To achieve the above objectives, the present invention provides the following technical solution: This user behavior-aware intelligent furniture adaptive adjustment system includes a data acquisition module, a data processing module, and a strategy adjustment module, wherein: The data acquisition module includes a pressure sensor array and a triaxial accelerometer. The pressure sensor array is fixedly installed on the surface of the smart furniture cushion to collect pressure distribution data in real time. The pressure distribution data records the pressure values ​​of different sensing points in the form of a two-dimensional matrix. The triaxial accelerometer is used to monitor the user's body movement changes in real time and to collect movement frequency data by counting the number of movements and intervals based on a preset time window.

[0006] This data acquisition module integrates a pressure sensor array and a triaxial accelerometer, providing a complementary multimodal behavioral data foundation for accurate fatigue identification. Pressure distribution data characterizes the stability of the user's body pressure from a static perspective, while movement frequency data reflects the user's activity level from a dynamic perspective. This simultaneous acquisition of static and dynamic data overcomes the potential for misjudgment from a single data source (such as misjudging a brief rest as inactivity), laying a comprehensive and objective data foundation for subsequent integrated analysis of the user's fatigue signs from the inside out, and is the primary step in achieving accurate state perception.

[0007] The pressure distribution pattern feature extraction unit in the data processing module generates pressure distribution pattern features that characterize the user's sitting posture stability based on pressure distribution data, specifically including: Calculate the pressure center coordinates at each time point in the pressure distribution; within a preset time window, calculate the total length of the pressure center movement trajectory and the standard deviation of the pressure center point, and output the total length of the pressure center movement trajectory and the standard deviation of the pressure center point as the pressure distribution pattern characteristic representing the user's sitting posture stability; where: The calculation process for the total length of the pressure center's movement trajectory specifically includes: Within a preset time window, the straight-line distance between two adjacent pressure center coordinate points is calculated sequentially according to time order, and the straight-line distances between all adjacent points are accumulated to obtain the total length of the pressure center movement trajectory. The calculation process for the standard deviation of the pressure center point specifically includes: Calculate the arithmetic mean of all pressure center coordinates within a preset time window as the average position; calculate the Euclidean distance from each pressure center coordinate to the average position; and calculate the standard deviation of this Euclidean distance as the standard deviation of the pressure center point.

[0008] This pressure distribution pattern feature extraction unit transforms raw pressure distribution data into two quantitative features: the total length of the pressure center movement trajectory and the standard deviation of the pressure center point. This allows for a precise mathematical representation of the user's unconscious and subtle changes in sitting posture. The total trajectory length reflects the overall range of the user's posture adjustment, while the standard deviation quantifies the degree of posture dispersion (i.e., the stability of sitting posture). This processing objectively quantifies the vague subjective feeling of unstable sitting posture into a calculable indicator, providing key static behavioral characteristics for identifying states such as muscle stiffness and frequent fine-tuning caused by fatigue.

[0009] The action pattern feature extraction unit in the data processing module also generates action pattern features that characterize the user's activity level based on action frequency data, specifically including: Based on the action frequency data, the number of valid actions is counted within a preset time window. By monitoring the synthetic acceleration vector of the triaxial accelerometer in real time, when the amplitude of the synthetic acceleration vector exceeds the preset action judgment threshold, it is recorded as a valid action, and the number of times is counted is accumulated within the preset time window. Calculate the integral of the acceleration signal amplitude corresponding to the effective action. For each effective action, start from the moment when the amplitude of the synthetic acceleration vector exceeds the action judgment threshold and end when it falls back below the action judgment threshold. Perform time integration on the amplitude of the synthetic acceleration vector within this time period to obtain the integral of the acceleration amplitude of a single effective action. The sum of the integrals of the acceleration amplitude of all single effective actions is taken as the integral of the acceleration amplitude within the preset time window. The number of effective actions and the integral of the acceleration amplitude are combined to output the action pattern feature that characterizes the user's activity level.

[0010] This motion pattern feature extraction unit extracts dual features representing the user's activity level from dynamic behavior by statistically analyzing the number of effective movements and calculating the integral of the acceleration signal amplitude. This not only counts the frequency of large-amplitude movements but also captures the overall intensity or energy consumption of the movements through integral calculations. This design can effectively distinguish between different types of activities (such as occasional hand raising versus continuous body swaying), thereby more precisely characterizing the user's dynamic behavior patterns and providing quantitative support for determining whether the user is still due to focus or reduced activity due to fatigue.

[0011] The correlation analysis unit in the data processing module performs correlation analysis between pressure distribution pattern features and action pattern features, and generates user state representation quantities, specifically including: The total length of the pressure center movement trajectory and the standard deviation of the pressure center point in the pressure distribution pattern feature are integrated with the number of effective actions and the acceleration amplitude in the action pattern feature, and then linearly weighted and fused according to the preset weight coefficients to calculate the user state representation quantity.

[0012] The core function of this correlation analysis unit is to linearly weight and fuse static and dynamic features through preset weights to generate a comprehensive user state representation. This step achieves deep fusion of multimodal behavioral information. It no longer views sitting stability or activity level in isolation, but correlates the two to form a higher-level single indicator that can more comprehensively reflect the user's overall state. This significantly improves the dimensionality and reliability of state assessment, providing key inputs that have undergone data fusion and dimensionality reduction for the accurate generation of fatigue index.

[0013] The strategy adjustment module identifies user fatigue states based on user state representation metrics and generates a fatigue index, specifically including: The user status representation is compared with a preset fatigue threshold range. If the user status representation is higher than the upper limit of the fatigue threshold range, an increasing fatigue index is generated; if the user status representation is consistently lower than the lower limit of the fatigue threshold range, a decreasing fatigue index is generated.

[0014] This process dynamically generates an increasing or decreasing fatigue index by comparing user state representations with preset fatigue threshold ranges. This process maps comprehensive behavioral characteristics onto a continuous fatigue measurement scale, upgrading the system's judgment of user state from discrete, instantaneous event identification to continuous assessment of the degree of state change trends. This quantified fatigue index provides a core, dynamically changing basis for adjustment decisions and is the key to achieving adaptive adjustment rather than mechanical triggering.

[0015] The strategy adjustment module generates an adaptive adjustment strategy based on the real-time changing trend of the pressure distribution pattern characteristics and the fatigue index, including the adjustment trigger timing and the gradient of the adjustment parameter changes, wherein: The triggering process for adjusting the trigger timing specifically includes: When the fatigue index exceeds the preset action trigger threshold, and the standard deviation of the pressure center point in the pressure distribution pattern simultaneously shows a continuous increasing trend, the adjustment trigger is immediately triggered. The determination of the continuous increasing trend is achieved by performing a linear fit using the least squares method on the pressure center point standard deviation sequence calculated within the most recent N preset time windows, and calculating its slope. When the slope is positive and greater than a preset trend determination threshold, it is determined that the pressure center point standard deviation shows a continuous increasing trend. N is the number of pressure center point standard deviation values ​​used for trend determination.

[0016] The process of generating the gradient by adjusting parameter changes specifically includes: The difference between the fatigue index and the preset action trigger threshold is calculated as the excess amplitude; this excess amplitude is input into a preset linear function, and the output value of the linear function is used as the gradient of the adjustment parameter change for this adjustment.

[0017] The above process achieves dynamic optimization of the adjustment strategy by introducing two dimensions of decision-making: the timing of adjustment triggering and the gradient of adjustment parameter changes. It does not simply trigger adjustment immediately when the fatigue index exceeds the limit, but rather conducts secondary confirmation by combining the "continuously increasing trend" of the standard deviation of the pressure center point, ensuring the accuracy of the triggering timing and avoiding erroneous intervention. At the same time, the adjustment gradient is linearly calculated based on the extent of the fatigue index exceeding the limit, so that the adjustment intensity matches the degree of fatigue. This dynamic decision-making mechanism based on dual criteria (timing and intensity) ultimately solves the problem of the timeliness and appropriateness of the adjustment strategy, completing the closed loop from accurate perception to intelligent decision-making.

[0018] Compared with the prior art, the beneficial effects of the present invention are: By integrating a pressure sensor array and a triaxial accelerometer, multimodal data acquisition of static pressure distribution and dynamic motion frequency was achieved, providing a comprehensive and objective data foundation for fatigue state recognition. The raw behavioral data was quantified into fine features of sitting posture stability and activity level, and a comprehensive user state representation was generated through linear weighted fusion, significantly improving the accuracy and reliability of fatigue recognition. Furthermore, the adjustment strategy was dynamically optimized based on the dual criteria of fatigue index and pressure distribution trend, ensuring the accuracy of triggering timing and the matching of adjustment intensity, thereby effectively avoiding false intervention and realizing adaptive closed-loop adjustment from perception to decision-making. Ultimately, the problems of inaccurate state recognition and rigid strategy in smart furniture adjustment were solved. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the overall modules of the present invention; Figure 2 This is a schematic diagram of the data processing module unit of the present invention; Figure 3 This is a schematic diagram of the core process of the data processing module of the present invention.

[0020] In the diagram: 100, Data Acquisition Module; 200, Data Processing Module; 201, Pressure Distribution Pattern Feature Extraction Unit; 202, Action Pattern Feature Extraction Unit; 203, Correlation Analysis Unit; 300, Strategy Adjustment Module. Detailed Implementation

[0021] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Next, please refer to Figure 1 The present invention provides a technical solution: an intelligent furniture adaptive adjustment system based on user behavior perception, including a data acquisition module 100, a data processing module 200, and a strategy adjustment module 300.

[0023] The data acquisition module 100 continuously collects pressure distribution data from the user in real time through a pressure sensor array fixedly installed on the surface of the smart furniture cushion. This pressure distribution data records the pressure values ​​of different sensing points in the form of a two-dimensional matrix. At the same time, through a three-axis accelerometer integrated in the smart furniture, it monitors the user's body movement changes in real time and collects movement frequency data by counting the number of all movements and intervals based on a preset time window.

[0024] Please see Figure 2-3The pressure distribution pattern feature extraction unit 201 in the data processing module 200 generates pressure distribution pattern features characterizing the user's sitting posture stability based on the pressure distribution data, specifically including: First, calculate the pressure center coordinates of the pressure distribution at each time point. Specifically, for each frame of pressure distribution data recorded in the form of a two-dimensional matrix, multiply the pressure value of each sensing point by its coordinate value in the array and sum them up, then divide by the sum of the pressure values ​​of all sensing points to calculate the horizontal and vertical coordinates of the pressure center at that time point on the two-dimensional plane. Then, within a preset time window, the total length of the trajectory formed by all continuous pressure center coordinate points within the preset time window is calculated. Specifically, in chronological order, the straight-line distance between two adjacent pressure center coordinate points within the preset time window is calculated sequentially, and then the straight-line distances between all these adjacent points are accumulated. The sum is the total length of the pressure center movement trajectory. The standard deviation of the pressure center coordinate point relative to its average position is calculated synchronously and used as the standard deviation of the pressure center point. Specifically, the arithmetic mean of all pressure center coordinate points within the preset time window is first calculated as the average position. Then, the Euclidean distance from each pressure center coordinate point to the average position is calculated, and the standard deviation of these Euclidean distances is calculated. The result is the standard deviation of the pressure center point.

[0025] Finally, the total length of the pressure center movement trajectory and the standard deviation of the pressure center point are output together as the pressure distribution pattern feature characterizing the user's sitting posture stability.

[0026] The action pattern feature extraction unit 202 in the data processing module 200 simultaneously generates action pattern features representing the user's activity level based on action frequency data, specifically including: Based on the frequency data of user body movements monitored by a triaxial accelerometer, the number of movements identified as valid movements is counted within a preset time window. Specifically, the synthetic acceleration vector of the triaxial accelerometer is monitored in real time. When the amplitude of the synthetic acceleration vector exceeds a preset movement judgment threshold, it is recorded as a valid movement. Within the preset time window, the cumulative number of all such events is counted, which is the number of valid movements. The integral of the acceleration signal amplitude corresponding to the valid action is calculated. Specifically, for each event identified as a valid action, from the moment when the amplitude of the synthesized acceleration vector exceeds the action judgment threshold to the moment when it falls back below the action judgment threshold, the amplitude of the synthesized acceleration vector within this time period is integrated over time, and the resulting integral value is the acceleration amplitude integral of a single valid action; the acceleration amplitude integral within a preset time window is the sum of the acceleration amplitude integrals of all single valid actions within that window; Ultimately, the number of effective actions and the integral of the acceleration amplitude are combined to output the action pattern feature that characterizes the user's activity level.

[0027] The correlation analysis unit 203 in the data processing module 200 performs correlation analysis between pressure distribution pattern features and action pattern features, and generates user state representation quantities, specifically including: The total length of the pressure center movement trajectory and the standard deviation of the pressure center point in the pressure distribution pattern features are linearly weighted and fused with the number of effective actions and the integral of acceleration amplitude in the action pattern features, according to preset weighting coefficients. Specifically, the total length of the pressure center movement trajectory is multiplied by a first weighting coefficient to obtain a first product, the standard deviation of the pressure center point is multiplied by a second weighting coefficient to obtain a second product, the number of effective actions is multiplied by a third weighting coefficient to obtain a third product, and the integral of acceleration amplitude is multiplied by a fourth weighting coefficient to obtain a fourth product. The first weighting coefficient is a multiplier pre-allocated to the total length of the pressure center movement trajectory, the second weighting coefficient is a multiplier pre-allocated to the standard deviation of the pressure center point, the third weighting coefficient is a multiplier pre-allocated to the number of effective actions, and the fourth weighting coefficient is a multiplier pre-allocated to the integral of acceleration amplitude. Then, the first, second, third, and fourth products are added together, and the sum is the user state representation quantity that comprehensively represents the user's static sitting posture and dynamic activities.

[0028] The strategy adjustment module 300 identifies user fatigue states based on user state representation metrics and generates a fatigue index, specifically including: The received user status representation is compared with a preset fatigue threshold range. If the user status representation is higher than the upper limit of the fatigue threshold range, it is determined that the user's fatigue level has increased, and a fatigue index that increases accordingly is generated. Specifically, when the user status representation is higher than the upper limit of the fatigue threshold range, the current user status representation is subtracted from the upper limit of the fatigue threshold range, the difference is multiplied by an increasing coefficient, and this product result is added to the fatigue index of the previous moment to generate a fatigue index that increases accordingly. If the user's state representation quantity is lower than the lower limit of the fatigue threshold interval, the user's fatigue level is determined to be reduced, and a fatigue index that decreases accordingly is generated. Specifically, when the user's state representation quantity is lower than the lower limit of the fatigue threshold interval, the current user's state representation quantity is subtracted from the lower limit of the fatigue threshold interval, the difference is multiplied by a decreasing coefficient, and this product result is subtracted from the fatigue index at the previous moment, thereby generating a fatigue index that decreases accordingly.

[0029] The strategy adjustment module 300 generates an adaptive adjustment strategy based on the real-time changing trend of the pressure distribution pattern characteristics and the fatigue index, including the adjustment trigger timing and the gradient of the adjustment parameter changes. Specifically, it includes: When the fatigue index exceeds the action trigger threshold, and the standard deviation of the pressure center point in the pressure distribution pattern characteristics simultaneously shows a continuously increasing trend, adjustment is immediately triggered; that is, the adjustment triggering time is determined to be the moment when the fatigue index first exceeds the action trigger threshold and the standard deviation of the pressure center point in the pressure distribution pattern characteristics is simultaneously detected to show a continuously increasing trend; the determination of showing a continuously increasing trend is as follows: For the pressure center point standard deviation sequence calculated within the most recent N preset time windows, the least squares method is used for linear fitting, and its slope is calculated. When the slope is positive and greater than a trend judgment threshold, it is determined that the pressure center point standard deviation shows a continuous increasing trend. N is used to determine the number of pressure center point standard deviation values. The least squares method uses the sequence number of the most recent N preset time windows as the independent variable sequence and the corresponding pressure center point standard deviation as the dependent variable sequence. By solving the optimization problem that minimizes the sum of squares of the differences between the observed values ​​of the dependent variable sequence and the linear estimates based on the independent variable sequence, the slope of the best-fit line is calculated.

[0030] After the adjustment trigger is triggered, the gradient of the adjustment parameter change is determined based on the extent to which the fatigue index exceeds the action trigger threshold. Specifically, the difference between the fatigue index and the action trigger threshold is calculated as the excess range. This excess range is input into a preset linear function, and the output value of this function is the gradient of the adjustment parameter change for this adjustment. That is, the greater the excess range of the fatigue index, the greater the gradient of the adjustment parameter change set for the preset function parameters of the smart furniture. The linear function is a linear function, and its standard form is: the gradient of the adjustment parameter change is equal to the excess range multiplied by a preset proportional coefficient, plus a preset constant term.

[0031] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent furniture adaptive adjustment system based on user behavior perception, characterized in that, It comprises a data acquisition module (100), a data processing module (200) and a strategy adjustment module (300), wherein: The data acquisition module (100) is used for acquiring pressure distribution data and action frequency data of a user using smart furniture; The data processing module (200) generates a pressure distribution pattern feature representing the user's sitting posture stability based on the pressure distribution data, and generates an action pattern feature representing the user's activity level based on the action frequency data; the pressure distribution pattern feature and the action pattern feature are analyzed in association, and a user state representation is generated; The strategy adjustment module (300) identifies the user's fatigue state based on the user state representation, generates a fatigue index, and generates an adaptive adjustment strategy including an adjustment trigger time and an adjustment parameter change gradient according to the real-time change trend of the pressure distribution pattern feature and the fatigue index. 2.The smart furniture adaptive adjustment system based on user behavior perception according to claim 1, wherein, The data acquisition module (100) comprises a pressure sensor array and a three-axis acceleration sensor; the pressure sensor array is fixedly installed on the surface of the smart furniture cushion, and is used for acquiring real-time pressure distribution data; the pressure distribution data records the pressure values of different sensing points in the form of a two-dimensional matrix; the three-axis acceleration sensor is used for monitoring the motion changes of the user's body in real time, and counts the action frequency data based on a preset time window. 3.The smart furniture adaptive adjustment system based on user behavior perception of claim 1, wherein, The data processing module (200) comprises a pressure distribution pattern feature extraction unit (201), and the process of generating the pressure distribution pattern feature by the pressure distribution pattern feature extraction unit (201) specifically comprises: The pressure center coordinates of the pressure distribution at each time point are calculated; the total length of the pressure center moving track and the standard deviation of the pressure center points are calculated within a preset time window, and the total length of the pressure center moving track and the standard deviation of the pressure center points are jointly output as the pressure distribution pattern feature representing the user's sitting posture stability; wherein: The process of calculating the total length of the pressure center moving track specifically comprises: Within the preset time window, the straight line distances between adjacent two pressure center coordinate points are calculated in time sequence, and the straight line distances between all adjacent points are accumulated to obtain the total length of the pressure center moving track; The process of calculating the standard deviation of the pressure center points specifically comprises: The arithmetic mean of all pressure center coordinate points within the preset time window is calculated as the average position; the Euclidean distance of each pressure center coordinate point to the average position is calculated; and the standard deviation of the Euclidean distance is calculated as the standard deviation of the pressure center points. 4.The smart furniture adaptive adjustment system based on user behavior perception of claim 3, wherein, The data processing module (200) comprises an action pattern feature extraction unit (202), and the process of generating the action pattern feature by the action pattern feature extraction unit (202) specifically comprises: Based on the action frequency data, the number of effective actions is counted within a preset time window, and the acceleration signal amplitude integral corresponding to the effective actions is calculated, and the number of effective actions and the acceleration amplitude integral are jointly output as the action pattern feature representing the user's activity level. 5.The smart furniture adaptive adjustment system based on user behavior perception of claim 4, wherein, The process of counting the number of effective actions specifically comprises: Real-time monitoring is performed on a synthetic acceleration vector of the triaxial acceleration sensor, and when an amplitude of the synthetic acceleration vector exceeds a preset motion determination threshold, one valid motion is recorded, and a statistical number of times is accumulated within a preset time window. 6.The smart furniture adaptive adjustment system based on user behavior perception of claim 4, wherein, The calculation process of the acceleration signal amplitude integral specifically includes: For each valid motion, starting from the time when the amplitude of the synthetic acceleration vector exceeds the motion determination threshold to the time when the amplitude falls below the motion determination threshold, the amplitude of the synthetic acceleration vector in this time period is time-integrated to obtain an acceleration amplitude integral of a single valid motion, and a sum of the acceleration amplitude integrals of all single valid motions is taken as the acceleration amplitude integral within the preset time window. 7.The smart furniture adaptive adjustment system based on user behavior perception of claim 4, wherein, The data processing module (200) includes a correlation analysis unit (203), and a process of generating the user state representation quantity by the correlation analysis unit (203) specifically includes: The total length of the pressure center moving track and the standard deviation of the pressure center point in the pressure distribution pattern feature are linearly weighted and fused with the number of valid motions and the acceleration amplitude integral in the motion pattern feature according to preset weight coefficients to calculate the user state representation quantity. 8.The smart furniture adaptive adjustment system based on user behavior perception of claim 1, wherein, The generation process of the fatigue index specifically includes: The user state representation quantity is compared with a preset fatigue threshold interval, if the user state representation quantity is higher than an upper limit of the fatigue threshold interval, an increasing fatigue index is generated, and if the user state representation quantity continuously is lower than a lower limit of the fatigue threshold interval, a decreasing fatigue index is generated. 9.The smart furniture adaptive adjustment system based on user behavior perception of claim 1, wherein, The trigger process of the adjustment trigger time in the adaptive adjustment strategy specifically includes: When the fatigue index exceeds a preset motion trigger threshold and the standard deviation of the pressure center point in the pressure distribution pattern feature continuously presents an increasing trend, the adjustment trigger time is triggered immediately, wherein the determination of the continuously increasing trend is performed by linear fitting on a sequence of the standard deviation of the pressure center point calculated in the last N preset time windows by using a least square method to calculate a slope, and when the slope is positive and greater than a preset trend determination threshold, it is determined that the standard deviation of the pressure center point presents the continuously increasing trend, wherein N is a number of values of the standard deviation of the pressure center point used for trend determination. 10.The smart furniture adaptive adjustment system based on user behavior perception of claim 1, wherein, The generation process of the adjustment parameter change gradient in the adaptive adjustment strategy specifically includes: A difference between the fatigue index and the preset motion trigger threshold is calculated as an exceeding amplitude, and the exceeding amplitude is input into a preset linear function, and an output value of the linear function is taken as the adjustment parameter change gradient of this adjustment.

Images