Intelligent table and chair sitting posture adjusting method based on pressure center track

Abstract

The invention discloses an intelligent table and chair sitting posture adjusting method based on a pressure center track, and the method comprises the following steps: S1, obtaining pressure distribution data of a user in real time through a pressure sensor array arranged on a seat surface and / or a backrest, and calculating and generating a pressure center track changing along with time; s2, preprocessing the acquired pressure center trajectory data, and extracting kinematics and statistics characteristics of the pressure center trajectory data; s3, based on the extracted features, performing real-time identification and classification on the sitting posture state of the current user by using a pre-trained machine learning classification model, and judging whether the sitting posture belongs to a standard sitting posture or one or more pre-defined bad sitting postures; and S4, if the identification result is that the sitting posture is bad, generating a corresponding adjustment instruction, and controlling a driving mechanism of the table and chair to execute a preset adjustment action so as to promote or assist the user to recover to the standard sitting posture. According to the invention, the user can be physically guided to recover to the standard sitting posture unconsciously, and the comfort and health level are improved.

Description

Technical Field

[0001] This invention relates to a method for adjusting the posture of intelligent tables and chairs based on the pressure center trajectory, belonging to the fields of ergonomics and smart home technology. Background Technology

[0002] With the development of information technology, computer-based office work has become the main mode of work. Statistics show that over 30% of adults spend more than 6 hours a day sitting. Prolonged sitting makes it difficult for users to maintain correct posture, easily leading to a series of health problems such as lower back pain, cervical spondylosis, and musculoskeletal disorders. This not only affects individual health and quality of life but also brings a huge social medical burden and loss of productivity.

[0003] Currently, posture recognition technology has received widespread attention. Researchers use devices such as pressure sensors, inertial measurement units, or visual sensors installed in chairs to collect user posture data and then classify and identify postures using machine learning algorithms (such as support vector machines, random forests, and convolutional neural networks). However, existing technologies have significant shortcomings in correcting poor posture. The mainstream approach is to remind users to adjust their posture themselves via voice, light, or mobile app notifications after poor posture is detected. This approach has significant drawbacks: first, the reminders are delayed; by the time the user receives the reminder, the poor posture has already put a certain strain on the body; second, frequent reminders can interfere with the user's concentration during work or study, affecting efficiency, and may even cause the user to turn off the reminder function due to annoyance, rendering the correction system ineffective. Summary of the Invention

[0004] The purpose of this invention is to provide an intelligent table and chair posture adjustment method based on pressure center trajectory. This invention analyzes the pressure center trajectory characteristics derived from changes in seat pressure distribution to identify poor posture in real time and imperceptibly. It can then drive the table and chair mechanism to adaptively adjust, physically guiding the user to unconsciously return to a standard sitting posture, thereby improving comfort and health.

[0005] The technical solution of this invention: A method for adjusting the posture of a smart table and chair based on the pressure center trajectory, comprising the following steps:

[0006] S1: Data Acquisition: Real-time acquisition of user pressure distribution data through an array of pressure sensors arranged on the seat and / or backrest, and calculation of the pressure center trajectory that changes over time;

[0007] S2: Data processing and feature extraction: Preprocess the collected pressure center trajectory data and extract its kinematic and statistical features;

[0008] S3: Posture Recognition and Classification: Based on the extracted features, a pre-trained machine learning classification model is used to identify and classify the current user's posture in real time, determining whether it belongs to a standard sitting posture or one or more predefined poor sitting postures.

[0009] S4: Active intervention and adjustment: If the identification result is poor sitting posture, a corresponding adjustment command is generated to control the drive mechanism of the table and chair to perform preset adjustment actions to prompt or assist the user to return to the standard sitting posture.

[0010] In the above-mentioned intelligent table and chair posture adjustment method, in step S1, the pressure sensor array is distributed in a grid pattern, covering the main pressure-bearing areas of the seat surface and backrest.

[0011] In the aforementioned intelligent table and chair posture adjustment method, the coordinates of the pressure center trajectory are calculated using the following formula:

[0012] ;

[0013] ;

[0014] In the formula: , These are the x-axis and y-axis coordinates of the pressure center in the planar coordinate system, respectively. For the first The pressure value detected by each pressure sensor. , For the first The coordinates of each pressure sensor This represents the total number of pressure sensors.

[0015] In the aforementioned intelligent table and chair posture adjustment method, step S2 includes preprocessing such as filtering and noise reduction, coordinate normalization, and time window segmentation; the extracted features include at least the position mean of the pressure center trajectory in the plane coordinate system, the trajectory standard deviation, displacement velocity, total length of the moving trajectory, component features of the trajectory in the front-back and left-right directions, and the spectral features of the trajectory.

[0016] In the aforementioned intelligent table and chair posture adjustment method, in step S3, the pre-trained machine learning classification model is a classifier built based on random forest, support vector machine or neural network, which is trained through a pressure center trajectory dataset labeled with different posture categories.

[0017] The aforementioned intelligent table and chair posture adjustment method uses a feature-weighted fusion strategy in its machine learning classification model. The weights are optimized using a particle swarm optimization algorithm, and the objective function is as follows:

[0018] ;

[0019] In the formula: This is the sum of classification errors. For the first Weight coefficients of class features For the first The classification error corresponding to the class feature. This represents the total number of feature categories.

[0020] In the aforementioned intelligent table and chair posture adjustment method, in step S4, the preset adjustment action is to change the height, tilt angle or front and back position of the seat and / or backrest through the drive mechanism, and the adjustment parameters include at least one of the adjustment amplitude, adjustment rate and adjustment duration.

[0021] The aforementioned intelligent table and chair posture adjustment method calculates the adjustment range using the following formula:

[0022] ;

[0023] In the formula, To adjust the amplitude, This is the proportionality coefficient. The coordinates of the current pressure center are: The coordinates of the pressure center correspond to the standard sitting posture.

[0024] The aforementioned intelligent table and chair posture adjustment method calculates the adjustment rate using the following formula:

[0025] ;

[0026] In the formula, To adjust the rate, This is the rate adjustment coefficient. Based on the adjustment rate, The confidence level for poor sitting posture.

[0027] The aforementioned intelligent table and chair posture adjustment method, after performing the adjustment action in step S4, also includes a feedback step: continuously monitoring the change in the user's pressure center trajectory after adjustment and calculating the posture recovery index.

[0028] ;

[0029] in, For the degree of recovery from sitting posture, , The coordinates of the pressure center in the plane coordinate system after adjustment, along the x and y axes. , The reference coordinates of the pressure center on the x and y axes are the preset standard sitting posture. , The minimum and maximum values ​​of the pressure center coordinates in the preset x-axis direction are given. , These are the preset minimum and maximum values ​​of the pressure center coordinates in the y-axis direction.

[0030] like If the value is less than the set threshold and the duration exceeds the set time, a secondary adjustment will be triggered or a warning signal will be issued.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] 1. This invention collects pressure distribution data through a grid-like pressure sensor array on the seat and backrest, combines it with multi-dimensional feature extraction of the pressure center trajectory, and uses a machine learning classification model with particle swarm optimization weights to achieve real-time and accurate identification of standard sitting posture and various poor sitting postures. This effectively improves the accuracy and robustness of sitting posture classification and avoids the lag and misjudgment problems of traditional identification methods.

[0033] 2. After identifying poor sitting posture, the present invention actively adjusts the height of the desktop, the height of the seat and backrest, the tilt angle or the front and back position through the drive mechanism. The adjustment range and rate are dynamically calculated based on the pressure center coordinate deviation and the confidence level of poor sitting posture. The invention uses physical guidance to encourage the user to unconsciously restore the standard sitting posture, which ensures timely correction without interfering with the user's work and study concentration.

[0034] 3. This invention can establish a personalized correction parameter system by quantifying the definition standards of poor sitting posture and the pressure center trajectory threshold of different groups of people, so as to adapt to the needs of users with different heights and body types. At the same time, the parameter optimization and feedback mechanism of the adjustment movement (based on the sitting posture recovery index) ensures that the correction process is smooth and comfortable, reduces user resistance, and improves long-term acceptance.

[0035] 4. This invention reduces the duration of poor sitting postures such as forward head and neck tilting and hunching back through real-time monitoring and active intervention, thereby reducing the risk of musculoskeletal diseases such as lower back pain and cervical spondylosis caused by prolonged sitting, alleviating the social medical burden, and improving the comfort of office and life for people who sit for long periods of time, taking into account both health value and user experience. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of the steps of the present invention;

[0037] Figure 2 This is a schematic diagram of the arrangement of pressure sensor arrays for tables and chairs;

[0038] Figure 3 This is a schematic diagram of the trajectory of the pressure center. Detailed Implementation

[0039] The present invention will be further described below with reference to the accompanying drawings and embodiments, but this should not be construed as limiting the present invention.

[0040] Example: A method for adjusting the posture of a smart table and chair based on the pressure center trajectory, such as... Figure 1 As shown, it includes the following steps:

[0041] S1: Data Acquisition: Real-time acquisition of user pressure distribution data through an array of pressure sensors arranged on the seat and / or backrest, and calculation of the pressure center trajectory that changes over time;

[0042] In this step, a flexible piezoresistive pressure sensor is selected, with a measurement range of 0-200 kPa, a resolution of 0.1 kPa, a response time of ≤10 ms, and an operating temperature of -10℃ to 60℃, meeting the needs of daily office environments. The sensor size is designed to be 2cm × 2cm, facilitating dense placement without affecting seat comfort. Figure 2 As shown, a grid-like distribution strategy is adopted. The seat area has 80 sensors arranged in a 10×8 array, covering the main pressure-bearing areas of the buttocks and legs (30cm from the front to the back edge of the seat, and 15cm on each side). The backrest area has 48 sensors arranged in an 8×6 array, covering the corresponding areas from the thoracic to the lumbar spine (40cm in height and 30cm in width), ensuring no blind spots in pressure distribution data acquisition. Each sensor is connected to the signal acquisition board via an I2C bus. The acquisition board integrates a 16-channel ADC converter (12-bit resolution) to convert the analog pressure signal into a digital signal, which is then transmitted to the main control unit via an RS485 bus at a baud rate of 115200bps to ensure low data transmission latency (≤50ms) and interference resistance.

[0043] During initialization, sensor calibration is performed by pressing each sensor with standard weights (5kg, 10kg, and 15kg), recording the correspondence between the output value and the actual pressure, and establishing a calibration equation: ,(in, It is a calibration value. For measured values, This is the sensitivity coefficient. (This is the offset), ensuring that the pressure detection error is ≤ ±2%.

[0044] The coordinates of the pressure center trajectory are calculated using the following formula:

[0045] ;

[0046] ;

[0047] In the formula: , These are the x-axis and y-axis coordinates of the pressure center in the planar coordinate system, respectively. For the first The pressure value detected by each pressure sensor. , For the first The coordinates of each pressure sensor The total number of pressure sensors is 80 for the seat surface and 48 for the backrest. Figure 3 A schematic diagram showing the trajectory of the pressure center over time is presented.

[0048] S2: Data processing and feature extraction: Preprocess the collected pressure center trajectory data and extract its kinematic and statistical features;

[0049] In this step, a Butterworth low-pass filter is used to process the raw pressure data, with a cutoff frequency of 5Hz to filter out interference from slight human body movements and circuit noise; at the same time, a median filter (window size 3×3) is used to remove isolated outliers to ensure data stability.

[0050] A two-dimensional coordinate system is established on the plane where the sensor array is located. The upper left corner of the seat is the origin (0,0) and the lower right corner is (30,45) (unit: cm). The actual coordinates of each sensor are normalized to the [0,1] interval by the maximum and minimum normalization method to eliminate the coordinate differences caused by different table and chair sizes.

[0051] The pressure center trajectory data is segmented using a sliding time window with a window length of 2 seconds and a sliding step of 1 second. Each window contains 200 sets of pressure center coordinate data (sampling frequency of 100Hz) to ensure both the capture of dynamic trajectory features and real-time performance.

[0052] The extracted features include at least the mean position of the pressure center trajectory in the planar coordinate system, the standard deviation of the trajectory, the displacement velocity, the total length of the trajectory, the component features of the trajectory in the front-back and left-right directions, and the spectral features of the trajectory.

[0053] The location mean is calculated within a single time window. and The average value and the standard deviation of the trajectory are within a single time calculation window. and The standard deviation of the displacement velocity is calculated by using the pressure center coordinates of two adjacent moments to determine the instantaneous velocity. The total length of the trajectory is the Euclidean distance between the pressure centers at adjacent moments within the accumulation window. The front-to-back / left-to-right component features are the total displacement of the trajectory along the x-axis (left-to-right direction) and y-axis (front-to-back direction), i.e., the difference between the maximum and minimum values ​​of the pressure center x-axis or y-axis coordinates within a single time window. The spectral features are obtained by performing a Fast Fourier Transform (FFT) on the pressure center x-axis and y-axis coordinate sequences to extract the peak frequency and spectral energy proportion in the 0-2Hz frequency band, reflecting the periodicity of the trajectory.

[0054] S3: Posture Recognition and Classification: Based on the extracted features, a pre-trained machine learning classification model is used to identify and classify the current user's posture in real time, determining whether it belongs to a standard sitting posture or one or more predefined poor sitting postures.

[0055] In this step, the model uses a random forest classifier. In other embodiments, it can also be a classifier built with a support vector machine or a neural network, which is trained on a stress center trajectory dataset labeled with different sitting posture categories.

[0056] The dataset was constructed by recruiting 30 healthy volunteers (half male and half female, height 155-190cm, weight 45-90kg) and collecting pressure center trajectory data for five typical sitting postures: standard sitting posture, head and neck forward tilt, hunched back, body tilt to the left, and body tilt to the right. Ten sets of data were collected for each sitting posture, with each set lasting 30 seconds, resulting in a labeled dataset containing 1500 samples (each sample contains the above 12 features).

[0057] Model training: The number of decision trees is set to 100, the maximum tree depth is 15, and the minimum number of sample splits is 5; a feature-weighted fusion strategy is adopted, and the feature weights are optimized through particle swarm optimization. The objective function is:

[0058] ;

[0059] In the formula: This is the sum of classification errors. For the first The weight coefficient for the class feature is set to 0.7. For the first Classification error corresponding to class features This represents the total number of feature categories. The number of optimization iterations is set to 100. Five-fold cross-validation is used to evaluate model performance, requiring an overall accuracy ≥ 95%, macro-average precision ≥ 94%, macro-average recall ≥ 94%, and macro-average F1 score ≥ 94%, meeting practical application requirements.

[0060] The trained model is exported in ONNX format and embedded into the ARM Cortex-A7 processor of the main control unit. Model inference is implemented in C++ language, with an inference time of ≤100ms, ensuring real-time recognition of sitting posture.

[0061] S4: Active intervention and adjustment: If the identification result is poor sitting posture, a corresponding adjustment command is generated to control the drive mechanism of the table and chair to perform preset adjustment actions to prompt or assist the user to return to the standard sitting posture.

[0062] In this step, the drive control board uses an STM32F407 microcontroller as the drive core, integrating the L298N motor drive chip, supporting PWM speed regulation and position closed-loop control. The table and chair drive control includes:

[0063] Seat height adjustment: A DC geared motor (rated voltage 12V, rated power 20W, reduction ratio 1:100) is used, paired with a ball screw (lead 5mm) to achieve stepless adjustment of seat height within the range of 50-80cm, with an adjustment accuracy of ±0.1cm.

[0064] Seat front and rear adjustment: Using linear guide rails and stepper motors (step angle 1.8°, subdivision 8 times), the front and rear of the seat can be adjusted from 0-15cm with a positioning accuracy of ±0.2cm.

[0065] Backrest tilt angle adjustment: A torque motor (rated torque 5 N·m) is used, combined with a gear transmission mechanism, to achieve a backrest tilt angle adjustment of 90°-110° (90° for vertical), with an adjustment accuracy of ±1°.

[0066] Desktop height adjustment: A dual-motor synchronous drive scheme is adopted, using two DC geared motors (the same model as the seat height adjustment motor: rated voltage 12V, rated power 20W, reduction ratio 1:100), which are deployed inside the legs on both sides of the desktop; paired with trapezoidal lead screws (8mm lead, load capacity ≥50kg) to ensure smooth desktop lifting without tilting; the desktop height adjustment range is set to 60-90cm (suitable for seated and light standing office needs), with an adjustment accuracy of ±0.2cm; the motors integrate a synchronous control module, which synchronizes the speed of the two motors through pulse signals, with a synchronization error ≤50ms, to prevent desktop tilting.

[0067] In this step, the adjustment parameters include at least one of the following: adjustment amplitude, adjustment rate, and adjustment duration, wherein:

[0068] The adjustment range is calculated using the following formula:

[0069] ;

[0070] In the formula, To adjust the amplitude, This is a proportionality coefficient, with values ​​of 0.8 (seat height), 0.5 (seat front-to-back), and 0.3 (backrest tilt angle). The coordinates of the current pressure center are: The coordinates of the pressure center corresponding to the standard sitting posture were determined by statistically analyzing the average pressure center values ​​of 30 volunteers in the standard sitting posture (seat surface: x=0.5, y=0.6; backrest: x=0.5, y=0.4, all normalized coordinates).

[0071] The adjustment rate is calculated using the following formula:

[0072] ;

[0073] In the formula, To adjust the speed, the value should be between 0.5 and 1.5 (dynamically adjusted according to the sitting posture). The basic adjustment speed is set as follows: seat height 2cm / min, seat front-to-back 1cm / min, backrest tilt 5° / min. This is the rate adjustment coefficient, with a value of 0.5. Based on the adjustment rate, The confidence level for poor posture is output by a machine learning classification model (between 0 and 1). The higher the confidence level, the faster the adjustment rate.

[0074] The adjustment duration is adaptively calculated based on the adjustment amplitude and rate. The formula is: adjustment duration = adjustment amplitude / adjustment rate. The shortest duration is 30 seconds and the longest duration is 3 minutes to avoid over-adjustment.

[0075] The main control unit sends a PWM control signal to the drive control board based on the identification results and the calculated adjustment parameters, which drives the motor to perform adjustment actions. The motor's built-in encoder provides real-time feedback of position information, forming a closed-loop control to ensure that the adjustment parameter error is ≤±5%.

[0076] Furthermore, after performing the adjustment action, a feedback step is also included: continuously monitoring the changes in the user's center of pressure trajectory after adjustment and calculating the sitting posture recovery index.

[0077] ;

[0078] in, For the degree of recovery from sitting posture, , The coordinates of the pressure center in the plane coordinate system after adjustment, along the x and y axes. , The reference coordinates of the pressure center on the x and y axes are the preset standard sitting posture. , The minimum and maximum values ​​of the pressure center coordinates in the preset x-axis direction are given. , These are the preset minimum and maximum values ​​of the pressure center coordinates in the y-axis direction.

[0079] Posture recovery The value ranges from 0 to 1, with values ​​closer to 1 indicating better recovery. In this embodiment, the sitting posture recovery threshold is set to 0.85. If R < 0.85 after adjustment and the duration exceeds 60 seconds, a second adjustment is triggered. The amplitude of the second adjustment is 1.2 times that of the first adjustment, while the adjustment rate remains unchanged. If after the second adjustment... If the value is still <0.85, a warning signal will be issued through the buzzer built into the seat (frequency 1kHz, volume 60dB), and a prompt "Please adjust your posture" will be displayed on the desktop display screen.

[0080] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for adjusting the posture of a smart table and chair based on a pressure center trajectory, characterized in that, Includes the following steps: S1: Data Acquisition: Real-time acquisition of user pressure distribution data through an array of pressure sensors arranged on the seat and / or backrest, and calculation of the pressure center trajectory that changes over time; S2: Data processing and feature extraction: Preprocess the collected pressure center trajectory data and extract its kinematic and statistical features; S3: Posture Recognition and Classification: Based on the extracted features, a pre-trained machine learning classification model is used to identify and classify the current user's posture in real time, determining whether it belongs to a standard sitting posture or one or more predefined poor sitting postures. S4: Active intervention and adjustment: If the identification result is poor sitting posture, a corresponding adjustment command is generated to control the drive mechanism of the table and chair to perform preset adjustment actions to prompt or assist the user to return to the standard sitting posture.

2. The intelligent table and chair posture adjustment method according to claim 1, characterized in that, In step S1, the pressure sensor array is distributed in a grid pattern, covering the main pressure-bearing areas of the seat surface and backrest.

3. The intelligent table and chair posture adjustment method according to claim 1, characterized in that, The coordinates of the pressure center trajectory are calculated using the following formula: ； ； In the formula: , These are the x-axis and y-axis coordinates of the pressure center in the planar coordinate system, respectively. For the first The pressure value detected by each pressure sensor. , For the first The coordinates of each pressure sensor This represents the total number of pressure sensors.

4. The intelligent table and chair posture adjustment method according to claim 1, characterized in that, In step S2, the preprocessing includes filtering and denoising, coordinate normalization, and time window segmentation; the extracted features include at least the mean position of the pressure center trajectory in the plane coordinate system, the trajectory standard deviation, displacement velocity, total length of the moving trajectory, component features of the trajectory in the front-back and left-right directions, and the spectral features of the trajectory.

5. The intelligent table and chair posture adjustment method according to claim 1, characterized in that, In step S3, the pre-trained machine learning classification model is a classifier built based on random forest, support vector machine or neural network, which is trained on a stress center trajectory dataset labeled with different sitting posture categories.

6. The intelligent table and chair posture adjustment method according to claim 5, characterized in that, The machine learning classification model employs a feature-weighted fusion strategy, with the weights optimized using a particle swarm optimization algorithm. The objective function is as follows: ； In the formula: This is the sum of classification errors. For the first Weight coefficients of class features For the first The classification error corresponding to the class feature, This represents the total number of feature categories.

7. The intelligent table and chair posture adjustment method according to claim 1, characterized in that, In step S4, the preset adjustment action is to change the height of the desktop, the height of the seat and / or the backrest, the tilt angle or the front and back position through the drive mechanism. The adjustment parameters include at least one of the adjustment amplitude, adjustment rate and adjustment duration.

8. The intelligent table and chair posture adjustment method according to claim 7, characterized in that, The adjustment range is calculated using the following formula: ； In the formula, To adjust the amplitude, This is the proportionality coefficient. The coordinates of the current pressure center are: The coordinates of the pressure center correspond to the standard sitting posture.

9. The intelligent table and chair posture adjustment method according to claim 7, characterized in that, The adjustment rate is calculated using the following formula: ； In the formula, To adjust the rate, This is the rate adjustment coefficient. Based on the adjustment rate, The confidence level for poor sitting posture.

10. The intelligent table and chair posture adjustment method according to claim 1, characterized in that, After the adjustment action is performed in step S4, a feedback step is also included: continuously monitoring the change in the user's center of pressure trajectory after adjustment and calculating the sitting posture recovery index. ； in, For the degree of recovery from sitting posture, , The coordinates of the pressure center in the plane coordinate system after adjustment, along the x and y axes. , The reference coordinates of the pressure center on the x and y axes are the preset standard sitting posture. , The minimum and maximum values ​​of the pressure center coordinates in the preset x-axis direction are given. , These are the preset minimum and maximum values ​​of the pressure center coordinates along the y-axis. If... If the value is less than the set threshold and the duration exceeds the set time, a secondary adjustment will be triggered or a warning signal will be issued.

Images