Passenger body type classification system and method based on dynamic pressure sensing

Abstract

The invention discloses a passenger body type classification system and method based on dynamic pressure sensing, and relates to the technical field of intelligent cabins. The invention provides a system which comprises a pressure sensor matrix, a data acquisition module and a processor. The core of the system is that the processor is configured to receive multi-frame pressure distribution data to form time sequence data in a preset time window after a passenger takes a seat; determining, based on the time series data, a stable value of a first physiological parameter (e.g., ischial spacing) representative of the body structure of the passenger; simultaneously calculating at least one second physiological parameter (such as a backrest inclination angle and a pressure distribution entropy); and finally, inputting the parameters into a classification model to generate a body type classification result. Through dynamic perception and analysis, the most stable physiological features of the passenger can be captured, interference in the sitting posture adjusting process is eliminated, and therefore the accuracy and reliability of body type classification are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent cockpit technology, and in particular to a passenger body shape classification system and method based on seat pressure sensing. Background Technology

[0002] With the rapid development of intelligent connected vehicles, personalized in-vehicle services have become crucial for enhancing user experience. Accurately recognizing passenger body shape is fundamental to enabling adaptive adjustments to systems such as seat position, seatbelts, and airbags. Currently, deploying pressure sensors within the seat to detect passenger body shape is a mainstream technological solution.

[0003] In the prior art, for example, Chinese patent CN222372784U discloses a mounting structure for a pressure sensor, which mainly focuses on the physical fixation of the sensor, with signal transmission relying on traditional physical wiring harness connections. This connection method is prone to poor contact or wire fatigue breakage under long-term use and vibration environments, leading to unstable signal acquisition or even failure.

[0004] Another Chinese patent, CN117725690A, discloses a method for predicting user body shape based on pressure sensor data. This method typically acquires a single static pressure distribution map after the passenger has settled into a seat and extracts features based on this data for body shape prediction. However, the process of a passenger sitting down is a dynamic behavior, with posture constantly adjusting. A single static measurement may only capture data from a transitional posture, rather than the actual data when the passenger is sitting stably. This is especially true for key skeletal structural features such as the ischial intervertebral distance (ITD), whose representation on the pressure map changes significantly during posture adjustments. Therefore, features extracted based on static measurements are not accurate enough, directly affecting the accuracy of subsequent body shape classification.

[0005] In summary, existing technologies for passenger body shape classification suffer from technical problems such as insufficient signal acquisition stability and inaccurate feature extraction and low classification accuracy due to the use of static measurement methods. Summary of the Invention

[0006] To address the problems in the background technology, such as unstable signal acquisition, reliance on static measurement leading to inaccurate extraction of key physiological features, and low final body shape classification accuracy, this invention aims to provide a passenger body shape classification system and method based on dynamic pressure perception, so as to improve the accuracy and reliability of body shape classification.

[0007] To address the aforementioned technical problems, in a first aspect, the present invention provides a passenger body shape classification system based on dynamic pressure sensing, comprising: a pressure sensor matrix disposed on a passenger carrying device; a data acquisition module connected to the pressure sensor matrix; and a processor connected to the data acquisition module; characterized in that the processor is configured to perform the following operations: within a preset time window, receiving multiple frames of pressure distribution data from the pressure sensor matrix through the data acquisition module to form time-series pressure data; determining a stable value of a first physiological parameter representing the passenger's body structure based on the time-series pressure data; calculating at least one second physiological parameter based on the pressure distribution data; and inputting the stable value of the first physiological parameter and the at least one second physiological parameter into a preset classification model to generate a passenger body shape classification result.

[0008] By employing a technique of acquiring multiple frames of data within a time window and determining stable parameter values ​​from them, this invention can capture the dynamic process of passengers taking their seats and extract stable physiological parameters that best reflect their true body structure, thus avoiding posture errors that may result from a single static measurement and significantly improving the accuracy of body type classification.

[0009] In a preferred embodiment of the present invention, the processor is configured to determine the stable value of the first physiological parameter by: calculating the instantaneous value of the first physiological parameter corresponding to each frame of pressure distribution data in the time-series pressure data; and selecting an extreme value from all instantaneous values ​​as the stable value of the first physiological parameter. By selecting the extreme value in the time series, the moment when the passenger's sitting posture is most stable and skeletal features are clearest can be effectively located, and the data at this moment best represents the passenger's true vital signs.

[0010] In a preferred embodiment of the present invention, the first physiological parameter is the passenger's ischial intercostal distance (ITD), and the extreme value is the minimum value among all instantaneous values. When a passenger sits down, as the body conforms to the seat, the distance between the peak pressure points of the two ischial bones changes first and then tends to stabilize. Its minimum value usually corresponds to the moment when the ischial bone is in the most firm contact with the seat surface. Therefore, using the minimum value of ITD as the stable value allows for accurate measurement of this key skeletal dimension.

[0011] As a preferred embodiment of the present invention, the second physiological parameter includes at least one of the following: a backrest tilt angle (θ) calculated based on the coordinates of the shoulder center point (SCP) and the ischium center point (ICP); and a pressure distribution entropy (H) calculated based on the pressure distribution data. By introducing multi-dimensional parameters such as the backrest tilt angle and pressure distribution entropy, the passenger's sitting posture and body pressure distribution characteristics can be more comprehensively described, providing richer information for the classification model and further improving the precision and accuracy of the classification.

[0012] As a preferred embodiment of the present invention, the backrest tilt angle (θ) is calculated using the following formula:

[0013]

[0014] Among them, (x) a y a (x) represents the coordinates of the center point of the shoulder. b y b () represents the coordinates of the center point of the ischium.

[0015] As a preferred embodiment of the present invention, the pressure distribution entropy (H) is calculated using the following formula:

[0016]

[0017] Among them, P i Let P be the pressure value at the i-th sensing point in the pressure sensor matrix. sum This is the sum of the pressure values ​​at all sensing points.

[0018] Pressure distribution entropy can quantify the uniformity of pressure distribution. Passengers with different body types (such as obese and thin) have significantly different pressure distribution uniformity. Thus, it is possible to effectively and accurately distinguish different body type characteristics.

[0019] In a preferred embodiment of the present invention, the processor is further configured to perform at least one preprocessing operation on the pressure distribution data before determining the physiological parameters. The preprocessing operation includes median filtering or temperature compensation. Through data preprocessing, random noise in the raw data can be effectively filtered out, and sensor performance drift caused by changes in ambient temperature can be corrected, ensuring the purity and reliability of the data input to the algorithm model.

[0020] As a preferred embodiment of the present invention, the classification model is a decision tree model, a neural network model, or a support vector machine model. By providing a variety of optional classification models, the present invention can flexibly select the optimal algorithm according to different application scenarios and accuracy requirements, exhibiting good adaptability and scalability.

[0021] As a preferred embodiment of the present invention, the pressure sensor matrix is ​​a piezoelectric sensor matrix, a piezoresistive sensor matrix, or a capacitive sensor matrix.

[0022] To address the aforementioned technical problems, in a second aspect, the present invention provides a passenger body shape classification method based on dynamic pressure sensing, characterized by comprising the following steps: within a preset time window, collecting multi-frame pressure distribution data output by a pressure sensor matrix on a passenger carrying device to form time-series pressure data; based on the time-series pressure data, determining a stable value of a first physiological parameter representing the passenger's body structure; based on the pressure distribution data, calculating at least one second physiological parameter; and inputting the stable value of the first physiological parameter and the at least one second physiological parameter into a preset classification model to generate a passenger body shape classification result.

[0023] This method, through dynamic data acquisition and analysis, fundamentally solves the problem of insufficient accuracy in existing static measurement methods, enabling accurate and reliable classification of passenger body shapes.

[0024] In a preferred embodiment of the present invention, the step of determining the stable value of the first physiological parameter includes: calculating the instantaneous value of the first physiological parameter corresponding to each frame of pressure distribution data in the time-series pressure data; and selecting an extreme value from all instantaneous values ​​as the stable value of the first physiological parameter, wherein the first physiological parameter is the ischial intervertebral distance (ITD), and the extreme value is the minimum value. This method can accurately locate and extract the stable ITD value that best represents the skeletal characteristics of the passenger.

[0025] In a preferred embodiment of the present invention, the second physiological parameter includes the backrest tilt angle (θ) calculated based on the coordinates of the shoulder center point (SCP) and the ischium center point (ICP), and the pressure distribution entropy (H) calculated based on the pressure distribution data. This method, by fusing multi-dimensional physiological parameters, constructs a more comprehensive user profile, providing a solid foundation for accurate classification.

[0026] As a preferred embodiment of the present invention, the method further includes a step of preprocessing the pressure distribution data before determining the physiological parameters. The preprocessing includes: applying a mean-mode filtering algorithm to eliminate noise in the pressure distribution data; and performing temperature compensation on the pressure distribution data based on the seat temperature to correct sensor drift. This method, by adding a preprocessing step, ensures the accuracy and robustness of subsequent feature extraction and classification.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] 1. High classification accuracy: This invention captures the most stable sitting state of passengers by dynamically collecting and analyzing data within a time window after they are seated, and uses this data to determine key physiological parameters such as ischial intercostal distance. This effectively avoids posture errors caused by static single measurements and fundamentally improves the accuracy of body type classification.

[0029] 2. Strong system robustness: The method of this invention includes data preprocessing steps such as median filtering and temperature compensation, which can effectively filter out noise, overcome sensor temperature drift, and ensure data quality, so that the entire classification system can still work stably and reliably in complex vehicle environments.

[0030] 3. Rich feature dimensions: This invention not only focuses on the ischial intercostal distance, but also integrates physiological parameters of multiple dimensions such as backrest tilt angle and pressure distribution entropy to construct a more comprehensive feature set, enabling the classification model to better distinguish the subtle differences between different body types and achieve more refined classification.

[0031] 4. Good technical adaptability: This invention does not impose a unique limitation on the type of sensor at the bottom layer and the classification model at the top layer. It can be applied to piezoelectric and piezoresistive sensors, as well as decision trees and neural network models, making the technical solution of this invention have good portability and scalability. Attached Figure Description

[0032] To more clearly illustrate the technical solutions of the embodiments disclosed in this invention, the accompanying drawings of the embodiments will be briefly described below. These drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention.

[0033] Figure 1 This is a schematic diagram of a system architecture provided in an embodiment of the present invention.

[0034] Figure 2 This is a schematic diagram of a data processing flow provided in an embodiment of the present invention.

[0035] Figure 3 This is a front view schematic diagram of the passenger carrying device and sensor arrangement in an embodiment of the present invention.

[0036] Figure 4 This is a side view schematic diagram of the passenger carrying device and sensor arrangement in an embodiment of the present invention.

[0037] Figure 5 This is a system timing diagram provided in an embodiment of the present invention. Detailed Implementation

[0038] The technical solutions (including preferred technical solutions) of the present invention will be further described in detail below with reference to the accompanying drawings and by way of listing some optional embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0039] Example 1

[0040] This embodiment provides a passenger body shape classification system based on dynamic pressure sensing, such as... Figure 1 As shown, this system can be integrated into a vehicle's smart cockpit system. The system includes a pressure sensor matrix, a data acquisition module, and a processor CPU mounted on a passenger-carrying device 100 (specifically, a car seat). The passenger-carrying device 100 includes a backrest area 10, a seat cushion area 20, and a leg rest area 30.

[0041] like Figure 3 and Figure 4 As shown, in this embodiment, the pressure sensor matrix consists of three parts: a backrest pressure sensor matrix 1, a seat cushion pressure sensor matrix 2, and a leg rest pressure sensor matrix 3, which together form a 12x12 sensor array covering the main contact areas. Specifically, the pressure sensor matrix is ​​a 12x12 piezoelectric ceramic sensor array containing 144 sensing points. These sensors are arranged in the key stress areas of the seat, including the lower backrest (corresponding to the lumbar region), the seat cushion (corresponding to the thighs and buttocks), and the leg rest (corresponding to the calves). The spatial resolution, i.e., the physical pixel size, of each sensor is set to 10 mm. This arrangement can comprehensively capture the pressure distribution exerted by the passenger's body on the seat with high resolution. Of course, in other embodiments, the pressure sensor matrix can also employ piezoresistive or capacitive sensor arrays, which can also achieve pressure distribution measurement, all falling within the scope of protection of this invention.

[0042] In this embodiment, the data acquisition module consists of a 12-bit analog-to-digital converter (ADC) and a direct memory access controller (DMA). The analog pressure signal generated by the pressure sensor matrix is ​​first converted into a digital signal within the range of 0-4095 by the ADC. To efficiently transmit data without consuming excessive CPU resources, the system uses an SPI interface in conjunction with the DMA controller. Figure 5 As shown, the pressure data generated by the sensor array triggers DMA requests byte by byte. The DMA controller automatically moves the converted digital signal `adc_raw` to two buffers in memory without CPU intervention. When one buffer is full, the DMA sends a transfer completion interrupt to the CPU, which then begins processing the data in that buffer. Meanwhile, the DMA switches to the other buffer to continue writing new data, achieving parallel operation of data acquisition and processing and improving the system's real-time performance.

[0043] The CPU is the core control unit of this system. Upon receiving an interrupt indicating DMA transfer completion, the CPU begins executing the body type classification algorithm. The detailed steps of this algorithm will be described in Example 2. After executing the algorithm, the CPU obtains the body type classification result (a body_code) and sends this result to the seat controller via the CAN bus or other in-vehicle network. Based on this classification result, the seat controller can adaptively adjust the seat position, seatbelt pretension, airbag deployment strategy, etc., to provide passengers with a personalized and safer riding experience.

[0044] Example 2

[0045] This embodiment provides a passenger body shape classification method based on dynamic pressure perception, which is executed by the CPU in Embodiment 1. Figure 2 As shown, this method extracts and fuses three completely different information modalities (structural modality, posture modality, and distribution modality) from a single pressure data source to achieve a comprehensive judgment of passenger body shape. The method mainly includes three stages: data preprocessing, feature extraction, and body shape classification.

[0046] S101: Dynamic Data Acquisition

[0047] When the system detects a passenger taking a seat, it initiates a 5-second preset time window. Within these 5 seconds, the system continuously collects 12x12 pressure distribution data at 10ms sampling intervals. The time point t serves as a time index, which is derived from the system's internal timer and used to mark the position of each frame of data in the time series. In this way, one frame of adc_raw

[12]

[12] data is generated every 10ms and stored in memory by DMA.

[0048] S102: Data Preprocessing

[0049] The CPU preprocesses the latest frame of raw stress data, adc_raw, after each DMA interrupt (every 10ms).

[0050] First, perform median filtering. For each point in the 12x12 matrix, take its ADC value and the values ​​of its eight surrounding points (forming a 3x3 window), sort these nine values, and take the median as the filtered value for that point. This step can effectively filter out random noise.

[0051] Next, temperature compensation is performed. The CPU obtains the current seat surface temperature tSeat from the seat's NTC temperature sensor. Based on pre-calibrated coefficients, the zero-point drift zOffset and sensitivity gain gain are calculated:

[0052] zOffset = (0.5 / 100.0 / 10.0) * (tSeat - 30.0)

[0053] gain = 1.0 + (-0.2 / 100.0) * (tSeat - 30.0)

[0054] Then, the filtered ADC value is converted into the actual pressure value p_kPa in the range of 0-100 kPa:

[0055] p_kPa[i][j] = (med_adc / 4095.0 - zOffset) * 100.0 * gain

[0056] Where med_adc is the 12-bit ADC value after median filtering.

[0057] S103: Multimodal physiological parameter extraction, mainly including dynamic and static physiological parameter extraction.

[0058] The first physiological parameter (dynamically determined): the stable value of the intertuberous diameter (ITD). That is, structural modal parameter extraction—determining the stable ITD. ITD directly reflects the passenger's skeletal frame size and is a core "hard" feature distinguishing body size. Simply put, the CPU processes each frame of data within a 5-second time window, locating the pressure peak in the data from the seat pressure sensor matrix 2, and calculates the instantaneous ITD value. By comparing all instantaneous values, the minimum value is selected as the final, stable ITD value. This value represents the skeletal size at the moment when the passenger is fully seated and in the most stable posture.

[0059] Specifically, the process for determining this parameter is as follows:

[0060] 1. Locating the Ischial Center Point Coordinates: For the preprocessed pressure data p_kPa collected at each time point t within the time window, the CPU executes a peak search algorithm. This algorithm identifies the two points with the highest pressure values ​​within the sensor area corresponding to the seat cushion. These two points are considered projections of the passenger's left and right ischial pressure peaks, i.e., the ischial center points (ICP). Their horizontal coordinates are denoted as x_left(t) and x_right(t), respectively. These coordinates are pixel coordinates, derived from the physical layout index of the sensor matrix.

[0061] 2. Calculate the instantaneous physical distance: The CPU calculates the absolute value of the difference between the two coordinates, |x_left(t) - x_right(t)|, to obtain the distance in pixels. This pixel distance is then multiplied by a physical size conversion factor s. The factor s is derived from the inherent hardware parameters of the sensor matrix and its function is to convert pixel units to physical units. In this embodiment, s is 10 mm / pixel.

[0062] 3. Soft Tissue Compression Compensation: Considering that the soft tissues (muscle, fat) of a passenger's buttocks deform under pressure, causing the pressure peaks sensed by the sensor to be slightly smaller than the actual bone spacing, a soft tissue compression compensation coefficient, a_soft, is introduced. This coefficient is derived from an empirical value calibrated through experimental data, for example, 0.95. Its purpose is to correct the measured values ​​to make them closer to the actual ischial distance.

[0063] 4. Generate instantaneous ITD value: Through the above steps, the instantaneous raw ITD value itd_raw(t) at time point t is calculated. The calculation process is as follows:

[0064] itd_raw(t) = a_soft * |x_left(t) - x_right(t)|*s.

[0065] 5. Determine the stable ITD value: The processor compares all itd_raw(t) values ​​calculated within a 5-second time window and selects the minimum value as the final stable ITD value. The purpose of selecting the minimum value is that it usually corresponds to the moment when the passenger is fully seated and the posture is most stable. The measurement value at this time best represents the passenger's true body structure characteristics and eliminates data interference during the seating adjustment process.

[0066] 6. Smoothing Filtering: To further improve stability, the system can also maintain an ITD value, itd_filt, which has undergone exponential smoothing filtering. Its update strategy is: itd_filt = 0.8 * itd_filt_old + (1.0 - 0.8) * itd_raw. The itd_filt value can ultimately be used for classification.

[0067] The second physiological parameter (calculated based on single-frame data) is the backrest tilt angle θ and the pressure distribution entropy H. That is, it includes extracting posture modal parameters—calculating the backrest tilt angle θ—and extracting distribution modal parameters—calculating the pressure distribution entropy H. θ reflects the passenger's overall sitting posture, providing important context for interpreting other modalities. H quantitatively describes the uniformity of pressure distribution, reflecting the soft tissue characteristics of the passenger's hips and thighs, and is a key "soft" feature for finely distinguishing body types. A low entropy value indicates highly concentrated pressure (e.g., a slender body type), while a high entropy value indicates evenly distributed pressure (e.g., an obese body type).

[0068] The system selects the frame of pressure data p_kPa corresponding to the minimum ITD value to calculate θ and H.

[0069] Backrest tilt angle θ: Assuming that, based on prior calibration, the shoulder center point (SCP) is mainly distributed in row 2 of the matrix, and the ischium center point (ICP) is mainly distributed in row 9. The CPU uses the determined center x-coordinate of the ischium pressure peak (x_left + x_right) / 2, combined with the prior row index, to calculate the instantaneous angle value angle_raw. This value is also smoothed to obtain the final angle_filt.

[0070] angle_filt = 0.9 * angle_filt_old + (1.0 - 0.9) * angle_raw

[0071] In other words, in the frame of data corresponding to the minimum ITD value, the CPU determines the coordinates (x, y) of the shoulder center point SCP and the ischium center point ICP based on the data from the backrest pressure sensor matrix 1 and the seat cushion pressure sensor matrix 2, respectively. a y a ) and (x b y b Then, strictly follow the formula.

[0072] Calculate the backrest tilt angle θ.

[0073] Pressure distribution entropy H: The sum of all pressure values ​​in the current frame calculated by the processor CPU. sum Then, for each sensing point i, if its pressure P i If the value is greater than a minimum (e.g., 1e-6), then accumulate -(P) i / P sum ) * log2(P i / P sum (to the total entropy value H).

[0074] In other words, the CPU, in the same frame of data corresponding to the minimum ITD value, according to the formula...

[0075] , calculate the entropy of the pressure distribution.

[0076] a) First, iterate through all sensor points and record their pressure values ​​P. i Summing these values ​​gives the total pressure P. sum .

[0077] b) Then, for each sensing point i, if its pressure P i If the value is greater than a minimum (e.g., 1e-6 to avoid mathematical errors in calculating log(0), then calculate -(P). i / P sum ) * log2(P i / P sum ).

[0078] c) Finally, sum up the calculation results of all sensing points to obtain the final total entropy value H.

[0079] S104: Body Type Classification

[0080] The CPU extracts three modal parameters—ITD (structural modality), θ (attitude modality), and H (distribution modality)—and inputs them into a three-decision-tree classification model. This model is a concrete manifestation of the multimodal information fusion concept.

[0081] First, scene segmentation is performed using posture modality (θ): determine whether angle_filt is less than 100.0 to distinguish between two macroscopic scenes: "sitting posture" and "lying posture".

[0082] Then, a rough classification is performed using structural mode (ITD): After determining the posture scene, the passenger's skeleton size is initially classified according to the threshold of itd_filt (120.0mm or 130.0mm).

[0083] Finally, the distribution mode (H) is used for fine-tuning: within a group with similar skeletal size, soft tissue characteristics are judged based on the threshold of hp (entropy value H) (1.6 or 1.3), and the final output is a precise classification result of "child", "slender adult", "standard adult", "robust adult" or "obese adult".

[0084] Specifically, the CPU inputs the final stable ITD value (itd_filt, in mm), the smoothed angle_filt value (in degrees), and the pressure distribution entropy H value into the preset triple decision tree model and executes the following complete logical judgment:

[0085] 1. Determine if angle_filt is less than 100.0.

[0086] If (angle_filt < 100.0), then further judgment is made:

[0087] Determine if the ITD value (itd_filt) is less than 120.0 mm.

[0088] If (ITD < 120.0mm), then further judgment is needed:

[0089] Determine whether the pressure distribution entropy H is less than 1.6.

[0090] If H < 1.6, output body code 0, representing "child".

[0091] If not (H ≥ 1.6), output body type code 1, representing "slim adult".

[0092] If not (ITD ≥ 120.0mm), then further judgment is needed:

[0093] Determine whether the pressure distribution entropy H is less than 1.6.

[0094] If (H < 1.6), output body type code 1, representing "slim adult".

[0095] If not (H ≥ 1.6), output body type code 2, representing "standard adult".

[0096] If not (angle_filt ≥ 100.0), then further judgment is made:

[0097] Determine if the ITD value (itd_filt) is less than 130.0 mm.

[0098] If (ITD < 130.0mm), then further judgment is needed:

[0099] Determine whether the pressure distribution entropy H is less than 1.3.

[0100] If (H < 1.3), output body type code 2, representing "standard adult".

[0101] If not (H ≥ 1.3), output body type code 3, representing "robust adult".

[0102] If not (ITD ≥ 130.0mm), then further judgment is needed:

[0103] Determine whether the pressure distribution entropy H is less than 1.3.

[0104] If (H < 1.3), output body type code 3, representing "robust adult".

[0105] If not (H ≥ 1.3), output body type code 4, representing "obese adult".

[0106] Finally, the decision tree outputs a unique body code (body_code, 0-4), which is then sent to the relevant vehicle controller.

[0107] Through this hierarchical fusion logic, the present invention achieves high-precision and robust classification of passenger body shapes.

[0108] It will be readily understood by those skilled in the art that the above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, combinations, substitutions, improvements, etc., made under the spirit and principles of the present invention are included within the protection scope of the present invention.

Claims

1. A passenger body shape classification system based on dynamic pressure perception, comprising: A matrix of pressure sensors installed on the passenger carrying device; The data acquisition module is connected to the pressure sensor matrix; as well as The processor connected to the data acquisition module; The processor is characterized in that it is configured to perform the following operations: Within a preset time window, the data acquisition module receives multiple frames of pressure distribution data from the pressure sensor matrix to form time-series pressure data. Based on the time-series stress data, a stable value for a first physiological parameter representing the passenger's body structure is determined; Based on the pressure distribution data, at least one second physiological parameter is calculated; and The stable value of the first physiological parameter and the at least one second physiological parameter are input into a preset classification model to generate the passenger's body type classification result.

2. The system according to claim 1, characterized in that, The processor is configured to determine a stable value for the first physiological parameter in the following manner: In the time-series stress data, the instantaneous value of the first physiological parameter corresponding to each frame of stress distribution data is calculated; and Select an extreme value from all instantaneous values ​​as the stable value of the first physiological parameter.

3. The system according to claim 2, characterized in that, The first physiological parameter is the passenger's ischial intercostal distance (ITD), and the extreme value is the minimum value among all instantaneous values.

4. The system according to claim 1, characterized in that, The second physiological parameter includes at least one of the following: The backrest tilt angle θ is calculated based on the coordinates of the shoulder center point SCP and the ischium center point ICP. The pressure distribution entropy H is calculated based on the pressure distribution data.

5. The system according to claim 4, characterized in that, The backrest tilt angle θ is calculated using the following formula: Among them, (x) a y a (x) represents the coordinates of the center point of the shoulder. b y b () represents the coordinates of the center point of the ischium.

6. The system according to claim 4, characterized in that, The pressure distribution entropy H is calculated using the following formula: Among them, P i Let P be the pressure value at the i-th sensing point in the pressure sensor matrix. sum This is the sum of the pressure values ​​at all sensing points.

7. The system according to claim 1, characterized in that, The processor is also configured to perform at least one preprocessing operation on the pressure distribution data before determining physiological parameters, the preprocessing operation including median filtering or temperature compensation.

8. The system according to claim 1, characterized in that, The pressure sensor matrix can be a piezoelectric sensor matrix, a piezoresistive sensor matrix, or a capacitive sensor matrix.

9. A passenger body shape classification method based on dynamic pressure perception, characterized in that, Includes the following steps: Within a preset time window, multiple frames of pressure distribution data output by the pressure sensor matrix on the passenger carrying device are collected to form time-series pressure data; Based on the time-series stress data, a stable value for a first physiological parameter representing the passenger's body structure is determined; Based on the pressure distribution data, at least one second physiological parameter is calculated; as well as The stable value of the first physiological parameter and the at least one second physiological parameter are input into a preset classification model to generate the passenger's body type classification result.

10. The method according to claim 9, characterized in that, The step of determining the stable value of the first physiological parameter includes: In the time-series stress data, the instantaneous value of the first physiological parameter corresponding to each frame of stress distribution data is calculated; and One extreme value is selected from all instantaneous values ​​as the stable value of the first physiological parameter, wherein the first physiological parameter is the ischial intervertebral distance (ITD), and the extreme value is the minimum value.

11. The method according to claim 9, characterized in that, The second physiological parameter includes the backrest tilt angle θ calculated based on the coordinates of the shoulder center point SCP and the ischium center point ICP, and the pressure distribution entropy H calculated based on the pressure distribution data.

12. The method according to claim 9, characterized in that, The method also includes a step of preprocessing the pressure distribution data before determining the physiological parameters, the preprocessing including: The application of a medium-frequency filtering algorithm eliminates noise in the pressure distribution data; and Temperature compensation is applied to the pressure distribution data based on seat temperature to correct sensor drift.

Images