Plantar pressure detection method based on hybrid hysteresis model and detection sensor

Abstract

The disclosure provides a plantar pressure detection method and a detection sensor based on a hybrid hysteresis model, and relates to the technical field of plantar pressure detection, and comprises the following steps: acquiring a plantar pressure monitoring sensor capacitance signal; considering the hysteresis effect caused by materials during pressure loading and unloading, compensating the capacitance value of the plantar pressure monitoring sensor based on a hybrid model hysteresis model, which comprises the following steps: segment fitting the pressure loading section and the unloading section, first fitting and compensating the hysteresis main structure by using the hysteresis model, and then learning the asymmetric, dynamic and high-order residual parts that cannot be fitted by the hysteresis model by using a support vector regression SVR submodel, so as to realize the capacitance value compensation of the hysteresis effect. The disclosure realizes high-precision real-time detection of the sensor capacitance signal.

Description

Technical Field

[0001] This disclosure relates to the field of plantar pressure detection technology, specifically to a plantar pressure detection method and its detection sensor based on a hybrid hysteresis model. Background Technology

[0002] The statements in this section are merely background information relating to this disclosure and do not necessarily constitute prior art.

[0003] Plantar pressure measurement has significant applications in medical and health monitoring, sports rehabilitation, and gait analysis. Traditional rigid plantar pressure measurement systems typically use metal-based or silicon-based sensors, which are difficult to adapt to complex foot movements and flexibility requirements, and are also inconvenient to carry. In recent years, capacitive flexible pressure sensors have gradually become the preferred technology for flexible plantar pressure measurement due to their high sensitivity, wide operating range, and good flexibility.

[0004] However, existing capacitive flexible pressure detection systems still have some problems in practical applications, such as unreasonable sensor array design, complex data acquisition system, and large signal interference, which limit their widespread application in plantar pressure detection.

[0005] Specifically, in existing solutions, the electrode array design mostly uses resistive sensors, which do not need to consider the parasitic capacitance caused by the direct alignment of upper and lower circuits. However, if a capacitive sensor is used, the parasitic capacitance introduced by the wires in its design will significantly affect the sensor's readings. In traditional data acquisition systems, the sensor's range and acquisition frequency directly affect its application scope, and existing capacitive sensors often do not pay attention to the sensor's inherent hysteresis, or only reduce errors by optimizing the material properties themselves, resulting in low accuracy of the detection results. Summary of the Invention

[0006] To address the aforementioned issues, this disclosure proposes a plantar pressure detection method and sensor based on a hybrid hysteresis model. By designing a stepped microstructure, the electrode contact area is gradually expanded using multi-level stepped deformation to alleviate stress concentration effects. Furthermore, by optimizing the dielectric material, a flexible dielectric sensor with high sensitivity, wide measurement range, long lifespan, and low cost is designed. Based on the hybrid hysteresis model, the sensor hysteresis process is calculated and compensated to achieve real-time transmission and detection of plantar pressure data.

[0007] According to some embodiments, the present disclosure adopts the following technical solutions: A plantar pressure detection sensor includes an upper shielding layer, an upper electrode layer, a stepped elastic modulus dielectric layer, a lower electrode layer, and a lower shielding layer. The stepped elastic modulus dielectric layer is disposed between the upper electrode layer and the lower electrode layer. The stepped elastic modulus dielectric layer is composed of multiple stepped microstructures. Each stepped microstructure is separated by an auxiliary layer that fills the internal stepped gaps. From the uppermost to the lowermost stepped microstructure, the radius of each stepped microstructure increases sequentially from top to bottom. The auxiliary layer is a PDMS-MWCNTs thin film with holes. The upper electrode layer and the lower electrode layer are bonded and fixed to the stepped elastic modulus dielectric layer to form an integrated capacitive flexible pressure sensor array.

[0008] Furthermore, the upper and lower shielding layers are made of conductive fabric, and the upper and lower electrode layers use polyethylene terephthalate as the electrode substrate and conductive carbon adhesive as the conductive material. The conductive carbon adhesive is printed on the electrode substrate by screen printing process to form a patterned electrode array.

[0009] Furthermore, the stepped microstructure is polydimethylsiloxane doped with multi-walled carbon nanotubes.

[0010] According to some embodiments, the present disclosure adopts the following technical solutions: Plantar pressure detection methods based on hybrid hysteresis models include: Acquire the capacitance signal from the plantar pressure sensor and transmit the capacitance signal to the host computer in real time using Bluetooth serial communication; Considering the hysteresis effect caused by materials during pressure loading and unloading, capacitance compensation is performed on the plantar pressure detection sensor based on the hybrid hysteresis model. The compensated capacitance signal is processed by bilinear interpolation to calculate the pressure value at the interpolation point and obtain the magnitude of the plantar pressure distribution. Among them, the capacitance compensation of the plantar pressure detection sensor based on the hybrid model hysteresis model includes: segmented fitting of the pressure loading segment and the unloading segment; firstly, the hysteresis model is used to quickly fit and compensate the hysteresis backbone structure; then, the support vector regression (SVR) sub-model is used to learn the asymmetric, dynamic and higher-order residual parts that the hysteresis model cannot fit, so as to achieve capacitance compensation for the hysteresis effect.

[0011] According to some embodiments, the present disclosure adopts the following technical solutions: A computer program product includes a computer program that, when executed by a processor, implements the plantar pressure detection method based on a hybrid hysteresis model.

[0012] According to some embodiments, the present disclosure adopts the following technical solutions: A non-transitory computer-readable storage medium is provided for storing computer instructions, which, when executed by a processor, implement the plantar pressure detection method based on a hybrid hysteresis model.

[0013] According to some embodiments, the present disclosure adopts the following technical solutions: An electronic device includes a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to perform the foot pressure detection method based on a hybrid hysteresis model.

[0014] Compared with the prior art, the beneficial effects of this disclosure are as follows: This disclosure discloses a plantar pressure detection sensor, comprising an upper shielding layer, an upper electrode layer, a stepped elastic modulus dielectric layer, a lower electrode layer, and a lower shielding layer. The stepped elastic modulus dielectric layer, through a stepped microstructure design, utilizes multi-level stepped deformation to gradually increase the electrode contact area, alleviating stress concentration effects. PDMS-MWCNTs composite dielectric material optimization (using the high conductivity of multi-walled carbon nanotubes (MWCNTs) to improve the dielectric constant) overcomes the technical bottlenecks of traditional sensors, such as the difficulty in balancing sensitivity and measurement range, insufficient dynamic stability, and complex fabrication processes. Ultimately, this achieves a flexible dielectric sensing material with high sensitivity, wide measurement range, long lifespan, and low cost. Furthermore, the stepped structure also exhibits obvious abrupt change points and segmented sensitivity, which can help to better determine the stress conditions on the plantar surface.

[0015] This disclosure discloses a plantar pressure detection sensor that, through optimized material composition and structural design, achieves a capacitive flexible pressure sensor array with high sensitivity, wide operating range, and fast response, suitable for complex plantar pressure detection. Using conductive fabric as a shielding layer effectively isolates external interference, increasing the stability and accuracy of the sensing system.

[0016] This disclosed plantar pressure detection method based on a hybrid hysteresis model employs a Pcap01 capacitor-to-digital converter chip and wire parasitic capacitance compensation technology to reduce the impact of internal system errors, achieving high-precision real-time acquisition of sensor capacitance signals and improving detection accuracy. A hybrid hysteresis model is used to calculate and compensate for the sensor hysteresis process. First, a hysteresis model (PI model) is used to quickly fit and compensate for the hysteresis backbone structure (symmetric part). Then, support vector regression (SVR sub-model) is used to learn the asymmetric, dynamic, and higher-order residual parts that the PI model cannot fit. This reduces dependence on large-scale training data, improves generalization ability and real-time performance, and provides better interpretability. Real-time processing and transmission of plantar pressure data are achieved through interpolation algorithms and Bluetooth serial communication technology.

[0017] The plantar pressure detection method based on the hybrid hysteresis model disclosed herein takes into account practical application requirements (such as the layout of key points on the plantar surface and wearable integration), and verifies its effectiveness and reliability under various postures such as standing, internal and external rotation, walking, and jumping through examples. Attached Figure Description

[0018] The accompanying drawings, which form part of this disclosure, are used to provide a further understanding of this disclosure. The illustrative embodiments of this disclosure and their descriptions are used to explain this disclosure and do not constitute an undue limitation of this disclosure.

[0019] Figure 1 This is a schematic diagram of the stepped microstructure of the dielectric layer in an embodiment of this disclosure; Figure 2 This is a schematic diagram of the auxiliary layer structure according to an embodiment of the present disclosure; Figure 3 This is a schematic diagram of the assembly of the plantar pressure detection sensor according to an embodiment of the present disclosure; Figure 4 Capacitance-voltage response curves of dielectric films with different step numbers in embodiments of this disclosure; Figure 5 Sensitivity curves of dielectric thin films with different step numbers in embodiments of this disclosure; Figure 6 This is a node diagram of the plantar pressure detection sensor array according to an embodiment of the present disclosure; Figure 7 This is a schematic diagram of the overall structure of the plantar pressure detection sensor according to an embodiment of the present disclosure; Figure 8 This is a schematic diagram of the drift mode connection method for capacitance measurement according to an embodiment of the present disclosure; Figure 9 This is a schematic diagram illustrating the correspondence between plantar pressure and capacitance in an embodiment of this disclosure; Figure 10 This is an overall data flow diagram of the sensing system according to an embodiment of the present disclosure; Figure 11 This is a schematic diagram of standing foot pressure detection according to an embodiment of the present disclosure; in, Figure 11 (a) is a diagram of the standing posture and a pressure comparison table; (b) is the relative capacitance value of the sensor and the pressure image of test subject 1 when standing; (c) is the relative capacitance value of each sensor and the pressure image of test subject 2 when standing; and (d) is the relative capacitance value of each sensor and the pressure image of test subject 3 when standing. Figure 12 This is a schematic diagram of foot internal and external rotation pressure detection according to an embodiment of the present disclosure; in, Figure 12(a) is a pressure comparison table, (b) is the relative capacitance values ​​of each sensor and pressure image under the outward octagonal posture, and (c) is the relative capacitance values ​​of each sensor and pressure image under the inward octagonal posture. Figure 13 This is a schematic diagram of foot pressure detection for walking posture according to an embodiment of the present disclosure; in, Figure 13 (a) is the capacitance change curve of each sensor under walking posture, (b) is the foot pressure image during the heel-to-ground phase, (c) is the foot pressure image during the heel-to-ground phase, (d) is the foot pressure image during the heel-off phase, and (e) is the foot pressure image during the toe-off phase. Figure 14 This is a schematic diagram of foot pressure detection in jumping posture according to an embodiment of the present disclosure; in, Figure 14 (a) is the capacitance change curve of each sensor in the jumping posture, (b) is the foot pressure image during the standing phase, (c) is the foot pressure image during the squatting pre-swing phase, and (d) is the foot pressure image during the take-off phase. Detailed Implementation

[0020] The present disclosure will be further described below with reference to the accompanying drawings and embodiments.

[0021] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of this disclosure. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains.

[0022] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this disclosure. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms “comprising” and / or “including” are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0023] Example 1 One embodiment of this disclosure provides a plantar pressure detection sensor, including an upper shielding layer, an upper electrode layer, a stepped elastic modulus dielectric layer, a lower electrode layer, and a lower shielding layer. The stepped elastic modulus dielectric layer is disposed between the upper electrode layer and the lower electrode layer. The stepped elastic modulus dielectric layer is composed of multiple stepped microstructures. An auxiliary layer fills the internal stepped gaps between each stepped microstructure. From the uppermost stepped microstructure to the lowermost stepped microstructure, the radius of each stepped microstructure increases sequentially from top to bottom. A PDMS-MWCNTs thin film with pores in the auxiliary layer is used to bond and fix the upper electrode layer and the lower electrode layer to the stepped elastic modulus dielectric layer, forming an integrated capacitive flexible pressure sensor array.

[0024] As one example, such as Figure 7 As shown, a foot pressure detection sensor comprises five layers: an upper shielding layer, an upper electrode layer, a stepped elastic modulus dielectric layer, a lower electrode layer, and a lower shielding layer. The upper and lower shielding layers are made of conductive fabric, exhibiting excellent conductivity and effectively isolating external electric field interference. The upper and lower electrode layers use polyethylene terephthalate (PET) as the electrode substrate and conductive carbon adhesive as the conductive material. The conductive carbon adhesive is printed onto the electrode substrate using a screen printing process to form a patterned electrode array. The dielectric layer is polydimethylsiloxane (PDMS) doped with multi-walled carbon nanotubes (MWCNTs), designed with a stepped microstructure to create a flexible detection sensor with a multi-level stepped structure.

[0025] like Figure 7 , Figure 1 As shown, a stepped elastic modulus dielectric layer is disposed between the upper electrode layer and the lower electrode layer. The stepped elastic modulus dielectric layer is composed of multiple stepped microstructures. An auxiliary layer is used to fill the internal step gaps between each stepped microstructure. From the uppermost to the lowermost stepped microstructure, the radius of each stepped microstructure increases sequentially from top to bottom. Taking a three-layer stepped structure as an example, the radius of the uppermost step can be 3 mm, and the radius of the lowermost step can be 5.5 mm.

[0026] As one embodiment, multiple sets of stepped elastic modulus dielectric layers are provided, with the distance between the centers of each stepped microstructure being approximately 15 mm.

[0027] As one embodiment, the fabrication process of the stepped structure elastic modulus dielectric layer is as follows: First, a stepped mold is prepared using polytetrafluoroethylene (PTFE). The number of steps can be 2-6, with a total thickness of 0.8-1.5 mm and a total height of 0.4 mm equal to the total film thickness minus 0.4 mm. Then, a mixture of PDMS main agent and MWCNTs with isopropanol (C3H8O) is prepared (the mass of MWCNTs is controlled to be approximately 2.5% of the mass of PDMS), followed by heating and stirring until homogeneous. After cooling, a curing agent is added at a ratio of PDMS main agent: curing agent = 10:1. After stirring for 5 minutes to ensure uniform dispersion, the mixture is poured into the stepped mold, allowed to stand for vacuum degassing, and then cured in an 80°C oven for 2 hours. Upon demolding, a stepped dielectric layer is obtained, with the uppermost step radius being 3 mm and the lowermost step radius being 5.5 mm, resulting in the final stepped structure.

[0028] like Figure 2 As shown, to fill the stepped gaps and retain the degree of freedom of deformation, an auxiliary layer was prepared using a mold with uniform holes (hole diameter 11 mm). The auxiliary layer is a PDMS-MWCNTs thin film with holes. The upper electrode layer and the lower electrode layer are respectively bonded and fixed to the stepped structure elastic modulus dielectric layer to form an integrated capacitive flexible pressure sensor array.

[0029] As one embodiment, a PDMS-MWCNTs solution is injected into a cavity mold, and after curing, an auxiliary dielectric layer is formed. This auxiliary layer is precisely matched with the stepped structure by the cavity positions; that is, each cavity encapsulates a stepped structure, and the thickness is equal to the height of the stepped structure. This ensures that the stepped gaps are evenly distributed after filling, while avoiding pressure imbalance, ultimately resulting in the auxiliary layer structure.

[0030] As one embodiment, the assembly and packaging process of the plantar pressure detection sensor includes: Polyethylene terephthalate (PET) is used as the electrode substrate, and conductive carbon adhesive is used as the electrode conductive material, with dimensions matching the dielectric layer. The upper and lower electrode layers are aligned with light to ensure accurate electrode positioning. A stepped-structured dielectric layer with elastic modulus is placed between the upper and lower electrode layers, with an auxiliary layer filling the gap. Silicone adhesive sealant is used for fixation, and air is expelled to form an integrated capacitive flexible pressure sensor. The unprinted portions of the electrodes are clamped or glued to secure the upper and lower electrode positions. Furthermore, the silicone adhesive sealant, after curing, does not affect the sensor's compressibility and flexibility.

[0031] Furthermore, polyethylene terephthalate (PET) is used as the electrode substrate, conductive carbon adhesive is used as the electrode conductive material, and polydimethylsiloxane (PDMS) and multi-walled carbon nanotubes (MWCNTs) are used as dielectric materials.

[0032] As one embodiment, the dielectric layer is selected to have 6 steps and is equipped with an auxiliary layer.

[0033] As one embodiment, the distribution of foot pressure sensor array points is set, such as... Figure 6 As shown, it specifically includes: Eleven sensing points are arranged at the first and third metatarsal bones, the first, third, and fifth metatarsals, the forefoot arch, the posterior foot arch, and the heel. Conductive carbon adhesive is screen-printed onto a PET substrate to form patterned electrodes. Multi-wavelength carbon nanotubes (MWCNTs) are dispersed in a PDMS matrix and cured to form a dielectric layer with high dielectric constant and good elasticity. The upper and lower electrodes are bonded to the middle dielectric layer with a soft adhesive to form an integrated capacitive flexible pressure sensor array.

[0034] Simulation Experiment A structural dielectric layer was fabricated by implanting PDMS / MWCNTs into a silicon stepped mold, with 6 steps and a total thickness of 1.0 mm. The radius of the top step was 3 mm, and the radius of the bottom step was 5.5 mm. To better compare the effects of stepped structures, several structures with different numbers of stepped layers are listed as comparative examples. A six-step structure was ultimately selected as the dielectric layer for the sensor. Specific comparative implementation settings are shown in Table 1.

[0035] Table 1. Settings of Comparative Examples

[0036] Example 2 One embodiment of this disclosure provides a method for detecting plantar pressure based on a hybrid hysteresis model, including: Acquire the capacitance signal from the plantar pressure sensor and transmit the capacitance signal to the host computer in real time using Bluetooth serial communication; Considering the hysteresis effect caused by materials during pressure loading and unloading, capacitance compensation is performed on the plantar pressure detection sensor based on the hybrid hysteresis model. The compensated capacitance signal is processed by bilinear interpolation to calculate the pressure value at the interpolation point and obtain the magnitude of the plantar pressure distribution. Among them, the capacitance compensation of the plantar pressure detection sensor based on the hybrid model hysteresis model includes: segmented fitting of the pressure loading segment and the unloading segment; firstly, the hysteresis model is used to quickly fit and compensate the hysteresis backbone structure; then, the support vector regression (SVR) sub-model is used to learn the asymmetric, dynamic and higher-order residual parts that the hysteresis model cannot fit, so as to achieve capacitance compensation for the hysteresis effect.

[0037] As one embodiment, this disclosure discloses a plantar pressure detection method based on a hybrid hysteresis model, which uses a hybrid hysteresis model (PI + SVR) to calculate and compensate for the sensor hysteresis process. Real-time processing and transmission of plantar pressure data are achieved through interpolation algorithms and Bluetooth serial communication technology. The specific implementation process is as follows: Step 1: Acquire the capacitance signal from the plantar pressure sensor and transmit the capacitance signal to the host computer in real time using Bluetooth serial communication; Specifically, a PCB board is constructed by setting chip parameters and measurement circuit connection methods. The PCB board is used to collect the capacitance signal of the foot pressure detection sensor. The chip is a capacitance-to-digital converter chip. The drift mode is selected for connection. Each capacitor does not interfere with the others. In drift mode, two wires are grounded and connected to the capacitor under test. The capacitance of the wires is measured and stored in advance.

[0038] As one embodiment, the capacitor-to-digital converter chip uses the Pcap01 capacitor-to-digital converter chip, and the connection is performed in drift mode. This ensures that the capacitors do not interfere with each other, resulting in strong accuracy and anti-interference capabilities at different measurement frequencies. The connection method in drift mode is as follows: Figure 8 As shown. In drift mode, two wires are grounded and connected to the capacitor under test. The capacitance of the wires is measured and stored in advance. The parasitic capacitance of the wires is eliminated by a calculation program written into the chip.

[0039] The measurement circuit design uses a 10 pF reference capacitor and a 20 Hz detection frequency to meet the requirements of plantar pressure detection. In the compensation circuit, each capacitance detection requires three measurements. The chip supports simultaneous measurement of three capacitances, meaning the circuit's charging and discharging frequency is 180 Hz, performing 360 charging and discharging cycles per second, with a single discharge time of 2.77 ms. Therefore, the calculated discharge resistor should be approximately 270 kΩ.

[0040] Based on the chip parameters and connection methods set above, a PCB board is constructed to achieve high-precision real-time acquisition of sensor capacitance signals. Bluetooth serial communication technology is used to transmit the acquired capacitance data to a host computer in real time, enabling subsequent real-time data detection and analysis.

[0041] Step 2: Considering the hysteresis effect caused by the material during pressure loading and unloading, the capacitance value of the plantar pressure detection sensor is compensated based on the hybrid model hysteresis model. The compensated capacitance signal is processed by bilinear interpolation, and the pressure value at the interpolation point is calculated to obtain the magnitude of the plantar pressure distribution. Specifically, based on the correlation between capacitance and pressure, the pressure on each sensor can be calculated using the measured capacitance of the sensor unit, and a plantar pressure distribution map can be plotted in real time. The correlation between plantar pressure and capacitance is illustrated below. Figure 9 As shown.

[0042] In applications, capacitive / piezoresistive pressure sensors often exhibit hysteresis due to the inherent material properties: during loading (compression) and unloading (recovery), the capacitance-pressure curve displays an asymmetric hysteresis loop, with the same capacitance value corresponding to different pressure values ​​during loading and unloading. However, many existing methods largely ignore this hysteresis. Therefore, compensation is needed for the material-induced hysteresis during pressure loading and unloading. This disclosure performs piecewise fitting on the loading and unloading segments, establishes a Prandtl-Ishlinskii hysteresis model based on test data, and determines whether the sensor is in the rising or falling segment by considering the sensor's changing trend, thereby improving accuracy.

[0043] This disclosure employs a hybrid hysteresis model combining the Prandtl-Ishlinskii (PI) model and support vector regression (SVR). First, the PI model is used to quickly fit and compensate for the hysteresis backbone structure (symmetric part). Then, the SVR sub-model is used to learn the asymmetric, dynamic, and higher-order residuals that the PI model cannot fit. This achieves compensation for the hysteresis effect. Specifically: Step 21: Construction of the Prandtl-Ishlinskii (PI) hysteresis model: The PI model represents hysteresis behavior as a weighted sum of a series of play operators:

[0044] in, P ( t )for t The pressure value output at all times; C ( t )for t The capacitance value at a given time; c 0 represents a constant bias; c i For the first i The weighting coefficients of each lag operator; r i For the first i Threshold of a hysteresis operator; Ƥ ri [ C ]( t ) is the Play hysteresis operator, which corresponds to the delayed response of the signal.

[0045] Furthermore, the Play hysteresis operator is:

[0046] in, y ( t - () represents the output value at the previous time step. r This is the hysteresis threshold.

[0047] The loading and unloading phases are fitted as follows:

[0048]

[0049] The system combines the changing trend of the plantar pressure sensor to determine whether it is in the rising or falling phase, and uses an adaptive threshold to determine whether it is in the loading or unloading phase.

[0050] Use an adaptive threshold to determine whether it is loading or unloading: ΔC = C ( t )- C ( t -1) If ΔC > ε2, it is determined that the system is in the loading phase (pressure increases). If ΔC < ε1, it is determined that the unloading phase (pressure decrease) is underway. If ε1≤ ΔC ≤ ε2, maintain the previous state (debouncing); in C ( t ), C ( t -1) is the time series of sensor capacitance values. ε 1. ε 2 is the adaptive threshold, initialized to zero. Its adaptive algorithm is as follows:

[0051] Collect the most recent N ΔC values:

[0052]

[0053]

[0054] in, k 1. k 2. β ∈(0,1), and based on experience, it can be taken as 0.5~0.8.

[0055] While the above algorithm provides a better physical explanation, it doesn't consider additional factors such as temperature and external interference. Therefore, Support Vector Regression (SVR) is used to process the data. The specific algorithm for Support Vector Regression is shown below: First, the sensor is divided into three states: loading, unloading, and stationary or slightly fluctuating: S = {L, U, S}. Where L: loading stage, U: unloading stage, and S: stationary / slightly fluctuating stage.

[0056] Set state transition conditions:

[0057] in, y The sensor output value, θ 1. θ 2 represents the state threshold.

[0058] For different states S, collect data for each state: { , Pt} Train the regression function under different conditions:

[0059] in, For sample data; K( , ) is the kernel function, such as the RBF kernel; , For Lagrange multipliers; This is a bias term.

[0060] Solve the following expression:

[0061] in, ε This is the tolerance range; C This serves as a penalty factor to control underfitting / overfitting. φ () represents kernel mapping.

[0062] As one implementation, the PI model has a clearer physical meaning and interpretability, but it exhibits significant errors when dealing with complex asymmetric factors such as aging and environmental disturbances. SVR theoretically performs better, but it requires a large amount of data and is prone to underfitting data exceeding the training set. Therefore, this disclosure combines the two algorithms. First, the PI model is used to quickly fit and compensate for the hysteretic backbone structure (symmetric part), and then the SVR sub-model is used to learn the asymmetric, dynamic, and higher-order residual parts that the PI model cannot fit. This reduces dependence on large-scale training data, improves generalization ability and real-time performance, and offers better interpretability. The specific steps are as follows: First, the sensor is also divided into three states: loaded, unloaded, and stationary or with slight fluctuations.

[0063] S = {L, U, S} Use the Savitzky-Golay (SG) filter to determine the loading / unloading / jittering state.

[0064] Input the original capacitance sequence:

[0065] In the center of the window t In relative time As variables, fit:

[0066] Solve for the polynomial coefficients a k ,in

[0067]

[0068] Perform a status check:

[0069] in, θ 1. θ 2 represents the state threshold, which needs to be adjusted manually.

[0070] Train a PI model for each state:

[0071] in, H S ( y t )for S In state t The pressure value obtained from the PI model at any given time. C ( t )for t The capacitance value at a given time; c 0 represents a constant bias; c i For the first i The weighting coefficients of each lag operator; r i For the first i Threshold of a hysteresis operator; Ƥ ri [ C ]( t ) is the Play hysteresis operator, which corresponds to the delayed response of the signal.

[0072] Then use SVR to learn the residuals. :

[0073] Therefore, the training objective should be:

[0074] Substitute:

[0075] in, For sample data; K( , ) is the kernel function, such as the RBF kernel; , For Lagrange multipliers; For bias terms Combining the two results yields the final predicted pressure:

[0076] in, H S ( y tThe value represents the pressure obtained from the PI model under the current conditions. The residuals of the SVR learning PI model are given in the current state.

[0077] Step 3: Use an interpolation algorithm to process the collected capacitance data to increase the amount of data, expand the sensing area, and improve the ability to distinguish the pressure details of the sole of the foot.

[0078] The bilinear interpolation method is adopted. Based on the pressure values ​​of four neighboring known sensor points, the pressure values ​​of the interpolation points are calculated by linear weighting to achieve a smooth pressure distribution.

[0079] Assume the coordinates of the four vertices of a certain region in the sensor array are (x1, y1), (x1, y2), (x2, y1), and (x2, y2), and the corresponding pressure value is P. 11 P 12 P 21 P 22 The pressure value at the interpolation point (x, y) is:

[0080] in,

[0081] Center of gravity calculation: By calculating the weighted average of the plantar pressure distribution, the location of the center of plantar pressure (CoP) is determined, providing data support for gait analysis and posture correction.

[0082] Assume there are N pressure sensors on the sole of the foot, and the coordinates of each sensor are ( x i , y i ), corresponding pressure value P i CoP coordinates ( X CoP , Y CoP ) is calculated using the following formula:

[0083] Finally, the data is visualized, and data such as plantar pressure distribution map and CoP location are displayed in real time through host computer software or mobile application, providing intuitive test results.

[0084] As one example, such as Figure 9 As shown, the sensor is more sensitive when the plantar pressure is low, and decreases as the pressure increases, but still shows a noticeable response. Furthermore, the figure clearly shows that due to the sensor's hysteresis effect, the same force produces different responses during compression and unloading.

[0085] like Figure 11 As shown, under the same conditions, three experimenters of different heights (170 cm, 174 cm, 180 cm) and weights (75 kg, 65 kg, 80 kg) placed their right feet on a capacitive flexible pressure sensing array in turn, and the foot pressure measurement system was used to test the force on the feet of the three experimenters.

[0086] It allows for a direct comparison of the pressure exerted on different parts of the feet of various participants, and the calculation of the center of pressure (CoP) of the foot when the participant is standing.

[0087] Based on the capacitance-pressure calibration curve, the pressure at various points on the feet of the three participants while they were standing was measured. The foot pressure distribution trends were consistent across the three participants: the pressure at point 11 on the heel was the highest (197, 163, 176 kPa), followed by points 9 and 10; the pressure at point 4 on the metatarsal bone was relatively low (148, 125, 138 kPa); and the pressure at point 8 on the arch of the foot was the lowest (0, 0, 25 kPa). Participant 3 experienced a pressure of 25 kPa on their arch, indicating that they had flat feet, suggesting that this array can be used for flat foot detection.

[0088] Calculations showed that the plantar pressure center (CoP) of the three participants was (-3.84, 16.35), (-13.31, 11.26), and (-9.42, 0.17), respectively, all located near the arch of the foot, which is ergonomic. The difference in CoP location reflects body posture and provides data support for determining standing stability and ergonomic research.

[0089] like Figure 12 As shown, to verify the sensor array's ability to detect foot pronation (inward toeing) and supination (outward toeing), experimenter #1 simulated force by standing with a slight tilt. Real-time pressure maps visually displayed changes in the center of gravity. The experiment showed that during supination (center of gravity tilting inward), pressure significantly increased on the medial side of the foot (e.g., feet 1, 3, and 9), especially at the metatarsal bones (e.g., foot 3); pressure decreased on the lateral side of the foot (e.g., feet 2, 5, and 10). The medial side of the heel (foot 9) showed minimal pressure change due to the forward shift of the center of gravity offsetting the inward tilt.

[0090] Internal rotation (outward tilt of center of gravity): Pressure on the outer side of the foot (e.g., size 2, 5, 10) increases significantly, with the greatest change at the metatarsal bone (e.g., size 5); pressure on the inner side of the foot (e.g., size 1, 3, 9) decreases. Pressure on the outer side of the heel (size 10) also changes very little.

[0091] Calculations showed that during internal / external rotation, the center of plantar pressure (CoP) shifted not only medially / laterally compared to when standing, but also significantly forward (as shown in the experimental data). This aligns with the theory that the body requires greater foot support to maintain stability. This sensor array can effectively distinguish the plantar pressure distribution characteristics under different static movements, providing data support for the early diagnosis and treatment of bowlegs.

[0092] like Figure 13 As shown, to monitor gait, a sensor is placed inside the right shoe to detect dynamic changes in plantar pressure during walking. Walking is divided into four stages: Heel strike: The pressure of the heel sensors (9, 10, 11) increases from low to high value (e.g., 317 kPa for sensor 11), while the foot sensor reading is 0.

[0093] When the foot lands on the ground: the pressure distribution is similar to standing but the value is greater (the center of gravity shifts to the right foot).

[0094] Heel off the ground: The center of gravity shifts to the ball of the foot, and the pressure of the sensors at the front of the ball of the foot (such as No. 3, 4, and 5) reaches its peak (such as No. 3, 333 kPa), while the pressure on the heel gradually returns to zero.

[0095] Toes off the ground: When the foot leaves the ground, all sensor pressures return to zero.

[0096] Experiments have shown that this sensor array can respond to dynamic pressure and plot capacitance curves in real time. Combined with machine learning to collect big data, it can be applied to fall prevention and early warning systems for the elderly, and rehabilitation treatment for patients with cerebral palsy / stroke, showing great promise in the field of smart healthcare.

[0097] like Figure 14 As shown, to verify the sensing range and dynamic detection capability, this experiment detects jumping posture (in three stages): Standing phase: The pressure distribution is the same as normal standing, the pressure at each point is basically stable, the center of pressure is near the arch of the foot, and the body is balanced.

[0098] During the pre-squat phase (approximately 700ms): the pressure on the forefoot (sizes 1-6) increases significantly (e.g., size 3 reaches 391 kPa), while the pressure on the heel (sizes 9-11) decreases; the overall center of pressure shifts forward.

[0099] Takeoff phase: Pressure at each point increases sharply within 200ms, reaching its peak at the moment of takeoff (e.g., 608 kPa for forefoot No. 3); pressure returns to zero rapidly after takeoff.

[0100] The experimental process disclosed herein verifies the array's ability to detect high dynamic and large-scale local pressure in real time.

[0101] Example 3 One embodiment of this disclosure provides a plantar pressure detection system based on a hybrid hysteresis model, including a sensor array, a data acquisition and transmission module, and a signal processing module.

[0102] The sensor array is used to acquire the capacitive signal from the plantar pressure monitoring sensor. The data acquisition and transmission module is used to transmit the capacitance signal to the host computer in real time via Bluetooth serial communication; The signal processing module is used to consider the hysteresis effect caused by the material during pressure loading and unloading. It performs capacitance compensation on the plantar pressure monitoring sensor based on the hybrid model hysteresis model, and uses bilinear interpolation to process the compensated capacitance signal, calculate the pressure value at the interpolation point, and obtain the magnitude of the plantar pressure distribution. Among them, the capacitance compensation of the plantar pressure monitoring sensor based on the hybrid model hysteresis model includes: segmented fitting of the pressure loading segment and the unloading segment; firstly, the hysteresis model is used to quickly fit and compensate the hysteresis backbone structure; then, the support vector regression (SVR) sub-model is used to learn the asymmetric, dynamic and higher-order residual parts that the hysteresis model cannot fit, so as to achieve capacitance compensation for the hysteresis effect.

[0103] Example 4 One embodiment of this disclosure provides a computer program product, including a computer program that, when executed by a processor, implements the plantar pressure monitoring method based on a hybrid hysteresis model.

[0104] Example 5 One embodiment of this disclosure provides a non-transitory computer-readable storage medium for storing computer instructions, which, when executed by a processor, implement the plantar pressure monitoring method based on a hybrid hysteresis model.

[0105] Example 6 One embodiment of this disclosure provides an electronic device, including a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to implement the plantar pressure monitoring method based on a hybrid hysteresis model.

[0106] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0107] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0108] While the specific embodiments of this disclosure have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of this disclosure. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of this disclosure are still within the scope of protection of this disclosure.

Claims

1. A plantar pressure detection method based on a hybrid hysteresis model, implemented using a plantar pressure detection sensor, characterized in that, The foot pressure detection sensor includes an upper shielding layer, an upper electrode layer, a stepped elastic modulus dielectric layer, a lower electrode layer, and a lower shielding layer. The stepped elastic modulus dielectric layer is disposed between the upper electrode layer and the lower electrode layer. The stepped elastic modulus dielectric layer is composed of multiple stepped microstructures. Each stepped microstructure is separated by an auxiliary layer to fill the internal stepped gaps. From the uppermost to the lowermost stepped microstructure, the radius of each stepped microstructure increases sequentially from top to bottom. The auxiliary layer is a PDMS-MWCNTs thin film with holes. The upper electrode layer and the lower electrode layer are respectively bonded and fixed to the stepped elastic modulus dielectric layer to form an integrated capacitive flexible pressure sensor array. The upper and lower shielding layers are made of conductive fabrics, and the upper and lower electrode layers are made of polyethylene terephthalate as the electrode substrate. Conductive carbon adhesive is used as the conductive material and is printed on the electrode substrate by screen printing process to form a patterned electrode array. The stepped microstructure is polydimethylsiloxane doped with multi-walled carbon nanotubes; The plantar pressure detection method based on the hybrid hysteresis model specifically includes: Acquire the capacitance signal from the plantar pressure sensor and transmit the capacitance signal to the host computer in real time using Bluetooth serial communication; Considering the hysteresis effect caused by materials during pressure loading and unloading, capacitance compensation is performed on the plantar pressure detection sensor based on the hybrid hysteresis model. The compensated capacitance signal is processed by bilinear interpolation to calculate the pressure value at the interpolation point and obtain the magnitude of the plantar pressure distribution. A PCB board is constructed by setting chip parameters and measurement circuit connection methods. The PCB board is used to collect the capacitance signal of the plantar pressure detection sensor. The chip is a capacitance-to-digital converter chip. The drift mode is selected for connection. Each capacitor does not interfere with the others. In drift mode, two wires are grounded and connected to the capacitor under test. The capacitance of the wires is measured and stored in advance. Among them, the capacitance compensation of the plantar pressure detection sensor based on the hybrid model hysteresis model includes: segmented fitting of the pressure loading segment and the unloading segment; firstly, the hysteresis model is used to quickly fit and compensate the hysteresis backbone structure; then, the support vector regression (SVR) sub-model is used to learn the asymmetric, dynamic and higher-order residual parts that the hysteresis model cannot fit, so as to achieve capacitance compensation for the hysteresis effect.

2. The plantar pressure detection method based on a hybrid hysteresis model as described in claim 1, characterized in that, The hybrid hysteresis model combines a hysteresis model with a support vector regression (SVR) algorithm. It performs piecewise fitting on the pressure loading and unloading segments, establishes a hysteresis model based on the test data, and determines whether it is in the rising or falling segment by combining the changing trend of the plantar pressure detection sensor. The hysteresis model represents the hysteresis behavior as a weighted sum of multiple hysteresis operators, and uses an adaptive threshold to determine whether it is in the loading or unloading segment.

3. The plantar pressure detection method based on a hybrid hysteresis model as described in claim 2, characterized in that, The SVR sub-model of support vector regression is used to learn the asymmetric, dynamic and higher-order residual parts that the hysteresis model cannot fit. This includes: dividing the plantar pressure sensor into three states: loading, unloading, and static or small fluctuation; setting state transition conditions; collecting data under different states; training regression functions under different states based on the data from different states; and merging the output of the hysteresis model with the output of the regression function to obtain the final predicted pressure distribution.

4. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the plantar pressure detection method based on the hybrid hysteresis model as described in any one of claims 1-3.

5. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium is used to store computer instructions, which, when executed by a processor, implement the plantar pressure detection method based on a hybrid hysteresis model as described in any one of claims 1-3.

6. An electronic device, characterized in that, include: The device includes a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to perform the plantar pressure detection method based on a hybrid hysteresis model as described in any one of claims 1-3.

Images