A target pressure area mattress support hardness self-adaptive control method and system based on digital twinning

Abstract

This invention discloses a method and system for adaptive control of mattress support firmness in a target pressure area based on digital twins. The method collects pressure and temperature distribution data from the mattress surface to identify the human lying posture and main support points. Responding to a target coordinate area command input from an external terminal, it determines the target pressure area and its corresponding set of adjustable mattress zones. A digital twin model of the mattress-human contact state is established based on a discrete mattress grid, human contact area, zone firmness state, pressure observations, and temperature observations. Through calibration mapping of firmness state with equivalent stiffness / damping, pressure prediction error, temperature prediction error, and parameter smoothing constraints, the aim is to reduce the error between the predicted field and the measured matrix and suppress abrupt changes in model parameters. The equivalent stiffness, equivalent damping, and thermal conductivity parameters are iteratively corrected. Based on the corrected model, normalized pressure deviation, normalized temperature deviation, and normalized sustained pressure value are calculated to generate a support risk value. When the support risk value meets a clear threshold, the firmness adjustment is solved through a constrained optimization problem including a target area pressure deviation term, an adjacent area pressure transfer penalty term, and an adjustment amplitude penalty term. The firmness adjustment is then converted into a pneumatic, hydraulic, or mechanical stroke drive signal. This invention combines multimodal sensing, discrete contact modeling, model parameter inversion, adjacent partition pressure transfer constraints, and actuator calibration mapping. It can be distinguished from schemes that only perform digital twin synchronization of multidimensional monitoring data and output intervention decisions, thereby improving the accuracy of mattress support adjustment for the target pressure area.

Description

Technical Field

[0001] This invention relates to the fields of smart mattresses, sensor data processing, and support firmness control technology, and particularly to a method and system for adaptive control of mattress support firmness in a target pressure area based on digital twins. Background Technology

[0002] Current smart mattresses typically use an array of pressure sensors to detect the pressure distribution of the human body on the mattress and adjust the firmness of different areas of the mattress through air bladders, liquid bladders, or mechanical support components. Some mattresses also incorporate temperature sensors to obtain the temperature status of the mattress contact interface, thereby adjusting the comfort level.

[0003] However, existing solutions typically adjust firmness directly based on sensor thresholds or preset rules, or simply synchronize multi-dimensional monitoring data to a digital twin and output intervention decisions from a pressure distribution analysis model. These solutions lack a calculable, parameterized description of the relationship between mattress zone firmness, human contact areas, pressure transmission, and temperature diffusion. When a target pressure area needs to have its pressure reduced, simply lowering the firmness of that area may lead to increased pressure in adjacent areas, creating new pressure concentrations. Furthermore, if only instantaneous values ​​of the pressure or temperature matrix are used for control, the system struggles to distinguish the impact of sensor noise, posture changes, differences in mattress material response, and differences in actuator response on the control results.

[0004] Therefore, there is a need for a mattress support firmness adaptive control method that can establish a digital twin model of mattress-human contact state based on multimodal sensing data, and generate zoned firmness adjustment amounts using explicit equivalent stiffness / damping calibration mapping, pressure-temperature joint prediction error, and adjacent zone pressure transfer constraints. Summary of the Invention

[0005] The technical problem solved by this invention is: how to transform mattress pressure distribution data, temperature distribution data, human lying posture, target pressure area and zone hardness state into a digital twin model of mattress-human contact state that can be iteratively corrected, and form a reproducible model correction target based on pressure prediction error, temperature prediction error and parameter smoothing constraints, and then generate a zone hardness adjustment amount based on the model that both reduces pressure deviation in the target pressure area and inhibits pressure transfer in adjacent areas.

[0006] To address the aforementioned technical problems, this invention provides a method for adaptive control of mattress support firmness in a target pressure area based on digital twins, comprising: acquiring a pressure distribution matrix. and temperature distribution matrix ; Identify human lying posture and relative pose between the human body and the mattress coordinate system; In response to target coordinate region commands input from external terminals, determine the target pressure region A and the target adjustable partition set. and adjacent support partition set Establish a digital twin model of the mattress-human contact state; iteratively correct the model parameters based on the calibration mapping of firmness state and equivalent stiffness / damping, pressure prediction error, temperature prediction error, and parameter smoothing constraints; calculate the support risk value based on the corrected model. A stiffness adjustment amount is generated when the support risk value R is greater than or equal to the support risk threshold Rth. The mattress zone actuator adjusts the firmness of the corresponding zone support and continues to correct the model in the next sampling cycle.

[0007] Compared with the prior art, the present invention has at least the following beneficial effects:

[0008] 1. This invention extends mattress firmness control from simple threshold control or rule-based intervention after digital twin synchronization to predictive control based on a discrete contact state digital twin model.

[0009] 2. This invention discloses a pressure prediction field, a temperature prediction field, and a model parameter correction objective function, making the mapping process from a two-dimensional sensing matrix to a contact state model feasible.

[0010] 3. When generating the hardness adjustment amount, the present invention considers both the target adjustable zone and the adjacent support zone, and introduces a target area pressure deviation term, an adjacent area pressure transfer penalty term, and an adjustment range penalty term into the same optimization target, which can reduce the pressure deviation of the target pressure area and suppress new pressure concentration.

[0011] 4. This invention uses a back-back trigger electrical signal input from a hardware interface or external terminal to trigger back-back control, avoiding reliance on human subjective intent as the boundary of the claims.

[0012] 5. The present invention aims to control the equipment by supporting comfort and controlling the pressure state. The method steps all revolve around sensor data processing, model parameter correction and mattress actuator control. Attached Figure Description

[0013] Figure 1 This is a flowchart illustrating a method for adaptive control of mattress support firmness in a target pressure area according to an embodiment of the present invention.

[0014] Figure 2 This is a schematic diagram of the structure of a mattress support firmness adaptive control system for the target pressure area provided in an embodiment of the present invention;

[0015] Figure 3 A mattress discrete grid, target pressure region A, and target adjustable partition set are provided in one embodiment of the present invention. and adjacent support partition set A schematic diagram of the mapping;

[0016] Figure 4 This is a schematic diagram of a digital twin model parameter calibration process provided in an embodiment of the present invention;

[0017] Figure 5 This is a schematic diagram of hardness adjustment constraint optimization and actuator drive conversion provided in an embodiment of the present invention. Detailed Implementation

[0018] The embodiments of the present invention will now be described with reference to the accompanying drawings. These embodiments are used to illustrate the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention.

[0019] Example 1: Multimodal data acquisition and determination of target pressure area

[0020] Pressure and temperature sensor arrays are installed on the mattress surface or in the inner area near the surface. The pressure sensor array is formed according to the mattress coordinate system. OK The temperature and pressure sensor arrays are arranged in a grid pattern, either co-located or adjacent to each other. The controller operates according to the sampling period. Read the pressure distribution matrix and temperature distribution matrix .

[0021] right and Outlier removal, Kalman filtering, and moving average smoothing are performed. For missing grid data, adjacent grid interpolation is used for compensation. After processing, the human supine posture is identified based on the area of ​​the pressure connectivity region, the direction of the major axis, the pressure center of gravity, and the relative position between the main support points, and the human contact area is determined. .

[0022] An external terminal sends a target coordinate region command to the system. This command can be a rectangular region, a polygonal region, or a set of grid numbers in the mattress coordinate system. The region mapping module determines the target pressure region A based on the target coordinate region command, and determines the target adjustable zone set based on the overlap relationship between A and the mattress zone coordinates. and adjacent support partition set .

[0023] like Figure 3 As shown, the mattress coordinate system is discretized into multiple grid cells, and the target coordinate region command is projected onto the discrete mattress grid. The target pressure area A is then formed. The system determines the adjustable partitions whose overlapping area with the target pressure area A reaches a preset ratio as the target adjustable partition set. and will be combined with the target adjustable partition set Partitions with adjacent boundaries or adjacent grid relationships, and which may result in pressure transfer, are defined as the set of adjacent support partitions. . The hardness adjustment is used to reduce the pressure deviation in the target pressure area. Used to bear pressure transfer constraints and provide necessary support compensation.

[0024] Example 2: Digital Twin Model of Mattress-Human Contact State

[0025] The digital twin model of mattress-human contact states includes discrete mattress meshes. Human contact area Zoned hardness status Equivalent stiffness parameters Equivalent damping parameters Thermal conductivity parameters Pressure prediction field and temperature prediction field .

[0026] For the Each grid cell is used by the system based on the current firmness of the mattress zone containing that grid cell. Determine the equivalent stiffness and equivalent damping . This represents the normalized hardness state, with a value ranging from 0 to 1. and Determine the calibration mapping as follows:

[0027]

[0028]

[0029] in, , , , and These are the calibration coefficients for the corresponding mattress zones. During calibration, each mattress zone is positioned at least three firmness levels, for example... , and Apply a standard load at each hardness state. The standard load can be a 20kg, 40kg, or 60kg weight or an equivalent loading fixture. Collect the steady-state compression and unloading response time of the zone. Obtain the equivalent stiffness sample value by dividing the load by the steady-state compression. Obtain the equivalent damping sample value by fitting the unloading response curve. Then, use least squares fitting to obtain the equivalent stiffness sample value. , , , and Equivalent compression The pressure prediction field can be estimated based on pressure observations, initial preload pressure, and current equivalent stiffness, and is limited to a preset maximum compression. The pressure prediction field is calculated using the following formula:

[0030]

[0031] The temperature prediction field is calculated using the heat dissipation diffusion relationship:

[0032]

[0033] in, Indicates the first Summation of the eight-neighbor grids of each grid cell; The neighborhood thermal diffusivity, For the contact heat source coefficient, For pressure distribution matrix Specific contact heat source items, ; For environmental heat dissipation coefficient, The ambient temperature. The contact heat source proportionality coefficient can be determined by recording the temperature rise of the contact area per unit time under a known load.

[0034] The model calibration module constructs the objective function:

[0035]

[0036] in, Including equivalent stiffness parameters Equivalent damping parameters and thermal conductivity parameters , This represents the L2 norm. To ensure parameter reproducibility, The value range is from 0.01 to 0.25. The value range is from 0.001 to 0.20. The value range is from 0.001 to 0.10; , , For normalized weights, satisfying Preferred The value ranges from 0.45 to 0.70. The value ranges from 0.20 to 0.45. The value ranges from 0.05 to 0.20. , and To support the normalized weights in risk calculation, and to meet the requirements... Preferred The value ranges from 0.45 to 0.75. The value ranges from 0.10 to 0.35. The values ​​range from 0.10 to 0.35. These weights can be determined through no-load calibration, standard weight loading calibration, or short-term calibration during the initial user phase.

[0037] In one specific implementation, the equivalent compression amount Determine using the following formula:

[0038]

[0039] in, For the first Initial preload pressure of each grid cell, This is the minimum effective stiffness, used to avoid the denominator being too small; For maximum compression, Take 5% to 35% of the mattress thickness, preferably 10% to 25%. Indicates will Limit to the lower limit and upper limit between.

[0040] In one specific implementation, model calibration employs recursive least squares. Let the vector of parameters to be estimated be... Including the equivalent stiffness parameter, equivalent damping parameter, and thermal conductivity parameter of at least one partition, let the error vector be... Depend on and The recursive least squares method is performed as follows:

[0041] The first step is to use the previous sampling period. As initial values, and set the covariance matrix ;

[0042] The second step is to construct a regression matrix based on the current pressure distribution matrix, temperature distribution matrix, and zone hardness status. ;

[0043] The third step is to calculate the gain matrix. ;

[0044] Step 4, Update ;

[0045] Step 5, Update ;

[0046] Step 6, Project it onto the preset parameter boundary.

[0047] in, The forgetting factor is set between 0.90 and 0.995. Let be the regularization coefficient, and take . to The maximum number of iterations is 5 to 50; when the objective function of two adjacent iterations... The decrease rate is less than 0.5% or the objective function The iteration stops when the error falls below a preset error threshold. The preset error threshold can be set to a pressure mean square error of less than 2% to 10% of the sensor range within the target area, and a temperature mean square error of less than 0.2 degrees Celsius to 1.0 degrees Celsius.

[0048] like Figure 4 As shown, the parameter correction process uses the pressure distribution matrix of the current sampling period. Temperature distribution matrix Human contact area Relative pose and zoned hardness state and the parameters of the previous sampling period As input, the pressure prediction field is first calculated based on the digital twin model of the mattress-human body contact state. and temperature prediction field Then, based on the error between the predicted field and the measured matrix, as well as the parameter smoothing constraints, an objective function is constructed. Then, update the equivalent stiffness parameters within the parameter boundaries. Equivalent damping parameters and thermal conductivity parameters If the objective function does not converge but does not exceed the real-time iteration constraints, prediction and updates continue; if the convergence condition is met or the real-time constraints are met, the corrected function is output. It is used to support risk calculation and optimize hardness adjustment.

[0049] Example 3: Supporting Risk Calculation and Hardness Adjustment Generation

[0050] The risk calculation module supports the calculation of normalized pressure deviation within the target pressure zone A. Normalized temperature deviation and normalized sustained pressure value Pressure deviation value It is obtained by weighting the average pressure exceeding the threshold, the maximum pressure value, and the proportion of the area exceeding the threshold; temperature deviation value. The value is obtained by weighting the average temperature exceeding the threshold, the maximum temperature, and the rate of temperature rise; continuous pressure value. The number of sampling periods in which the pressure deviation value continuously meets the threshold condition is determined. Specifically, it can be calculated according to the formula in claim 1, so that... , , All are limited to the range of 0 to 1.

[0051] Support risk value Determine using the following formula:

[0052]

[0053] in, , and For normalized weights. If Greater than or equal to the support risk threshold Then the hardness adjustment generation module generates a target adjustable partition set. and adjacent support partition set Hardness adjustment amount . The value is 0.45 to 0.85, preferably 0.60 to 0.75.

[0054] In one implementation, the target adjustable partition set The decrease in hardness and Positive correlation; Adjacent support partition set The hardness compensation amount is determined by the predicted pressure transfer constraint. Adjustment amount This is determined by solving the following constrained optimization problem:

[0055]

[0056] in, The adjusted predicted pressure deviation value, The allowable pressure transfer threshold for adjacent support zones. , and For normalized weights, satisfying Preferred The value ranges from 0.45 to 0.75. The value ranges from 0.15 to 0.40. The value ranges from 0.05 to 0.20. The pressure range can be 3% to 20%, or 5% to 30% of the average pressure of the target pressurized area before adjustment. The constraints are:

[0057]

[0058]

[0059]

[0060] and These are the lower and upper limits of the zoned hardness, which can be determined by the range of stiffness output by the actuator. The maximum adjustment range in a single operation can be 5% to 25% of the adjustable hardness range. To ensure real-time performance, this implementation uses a finite-level enumeration search to solve the above optimization problem: First, the partition with the largest overlap area with the target pressure area A is selected as the target adjustable partition, and at most five partitions adjacent to the target adjustable partition with the highest pressure increment are selected as adjacent support partitions, so that the number of partitions participating in the optimization does not exceed six; then, the candidate adjustment amount for each partition participating in the optimization is limited to {-3dU, -2dU, -dU, 0, dU, 2dU, 3dU}, where dU is the single-level hardness adjustment step size; then, candidate combinations are enumerated one by one, and those that do not meet the requirements are eliminated. , , and Combining constraints; finally, selecting the objective function. The smallest candidate combination as When the number of partitions involved in optimization is six, the maximum number of candidate combinations is [number missing]. The controller can complete the solution within one sampling period using a pre-calculated prediction matrix and a parallel partitioned evaluation method. The sampling period dt can be 1 to 10 seconds; if the solution is not completed within 0.5dt, the effective model parameters and minimum feasible adjustment level of the previous sampling period are used to ensure that the control link is not interrupted.

[0061] The system will Before being sent to the mattress zoning actuator, it will also This is converted into an actuator drive signal. For airbag actuators, DeltaU is converted into the target air pressure increment. ,in Determined through factory calibration or no-load calibration; for mechanically supported actuators, Converted to support stroke increment For liquid bladder actuators, Convert to target hydraulic increment During calibration, the actuator is subjected to multiple known drive signals, including at least three pneumatic pressure values, hydraulic pressure values, or support stroke values. A standard load ranging from 20 kg to 80 kg is applied under each drive signal, and the steady-state compression of the corresponding zone under the standard load is measured. The equivalent support stiffness is calculated based on the load and steady-state compression. Then, the mapping coefficient or lookup table relationship between the drive signal and the support stiffness is obtained through linear fitting, polynomial fitting, or piecewise table lookup. The data is reacquired in the next sampling period. and And continue to perform model correction.

[0062] like Figure 5 As shown, when the support risk value The risk threshold for support has not been reached. When, the system maintains the current partition hardness state; when When Rth is reached or exceeded, the system generates candidate hardness adjustment values. and will Substitute the equivalent stiffness and equivalent damping calibration mapping to predict the adjusted pressure field. The system calculates the candidate adjustment values ​​separately, including... Objective function of the adjacent pressure transfer penalty term and the adjustment magnitude penalty term Eliminate those that do not meet the requirements. , , and Candidate combinations of constraints, and select the objective function. Minimum feasible Selected Then, depending on the type of actuator, it is converted into the target air pressure increment, target hydraulic pressure increment, or support stroke increment, and after the adjustment is performed, it enters the next sampling cycle for feedback correction.

[0063] Example 4: Rollback Control

[0064] The system configures a hardware interface or external terminal rollback control. When the rollback control is triggered, the system receives a rollback trigger electrical signal. In response to this rollback trigger electrical signal, the rollback control module cancels the hardness adjustment amount corresponding to the current sampling period. Alternatively, it can revert the mattress's firmness to the previously stored firmness state for each zone. This reversion control is triggered by an electrical signal and is not limited to human subjective feelings as defined in the claims.

[0065] The previously stored hardness state is saved using a circular cache. The cache content includes the sampling timestamp and the partition hardness state. Hardness adjustment amount Actuator drive signal, target pressure area A, support risk value And the corresponding pressure distribution matrix summary and temperature distribution matrix summary. The buffer depth is 30 to 100 sampling periods. When rollback is triggered, the system first freezes the new hardness adjustment output, and then synchronously rolls back the target partition and adjacent support partition to the previously stored hardness state according to the maximum change rate allowed by each actuator; if the synchronous rollback would exceed the maximum adjustment range in a single operation... Then the total rollback amount will be split into multiple amounts not exceeding [amount missing]. The sub-adjustment is performed incrementally at one sampling period interval until the previously stored hardness state is reached.

Claims

1. A method for adaptive control of mattress support firmness in a target pressure area based on digital twin, characterized in that, include: S1, Obtain the pressure distribution matrix output by the mattress surface sensor array during the current sampling period. and temperature distribution matrix and the pressure distribution matrix and temperature distribution matrix S2, Perform outlier removal, filtering, and smoothing; S3, Based on the processed pressure distribution matrix... S3: Determine the pressure connectivity region, pressure center of gravity, and multiple main support points; identify the human lying posture and the relative pose between the human body and the mattress coordinate system based on the pressure connectivity region, pressure center of gravity, and multiple main support points; S4: In response to the target coordinate region command input by the external terminal, determine the target pressure region A in the mattress coordinate system, and determine the target adjustable zone set based on the overlap relationship between the target pressure region A and the mattress zone coordinates. and adjacent support partition set S4, Establish a digital twin model of the mattress-human body contact state, wherein the digital twin model of the mattress-human body contact state includes a discrete mattress mesh. Human contact area Zoned hardness status Equivalent stiffness parameters Equivalent damping parameters Thermal conductivity parameters Pressure prediction field and temperature prediction field ,in Normalized hardness state and S5, based on the hardness state of the partition The relative pose and pressure distribution matrix between the human body and the mattress coordinate system. Calculate discrete mattress mesh Equivalent compression of each grid cell The pressure prediction field is generated according to the following formula. : ,in, For the first Each grid cell is in the current partition hardness state. The equivalent stiffness is as follows. For the first Each grid cell is in the current partition hardness state. The equivalent damping is below, The sampling period; and According to the calibration mapping respectively , Determined; S6, based on temperature distribution matrix Human contact area and thermal conductivity parameters A temperature prediction field is generated based on the heat dissipation and diffusion relationship. : ,in, Indicates the first Summation of the eight neighboring grids of each grid cell. The neighborhood thermal diffusivity, For the contact heat source coefficient, For pressure distribution matrix Specific contact heat source items and , The environmental heat dissipation coefficient, For ambient temperature; S7, predict the field based on pressure. With pressure distribution matrix Pressure prediction error and temperature prediction field With temperature distribution matrix Based on the temperature prediction error and model parameter smoothing constraints, construct the objective function. Iterative correction within the preset parameter boundaries. , and This makes the pressure prediction field With pressure distribution matrix Errors between them, temperature prediction field With temperature distribution matrix This reduces the error between samples while keeping the changes in model parameters between adjacent sampling periods smooth: ,in, This represents the summation within the target pressure region A and its neighborhood. include , and , Describing the L2 norm, , , For weights; S8, based on the corrected mattress-human contact state digital twin model, calculate the normalized pressure deviation value within the target pressure area A according to the following formula. Normalized temperature deviation and normalized sustained pressure value and according to , and Generate supporting risk values : , , , ,in, The pressure threshold, Temperature threshold As a pressure normalization scale, For temperature normalization scale, This represents the number of consecutive sampling periods exceeding the threshold. The threshold for the continuous pressure period. , , S9, when supporting risk value Greater than or equal to the support risk threshold At that time, a set of target adjustable partitions is generated through constrained optimization methods. and adjacent support partition set Hardness adjustment amount Make the target adjustable partition set The predicted pressure deviation is reduced, and the adjacent support partition set is improved. The predicted pressure increment is less than the pressure transfer threshold. ,in This is determined by solving the following constrained optimization problem: ,satisfy , ,and ;in, To be Substituting the data into the digital twin model of mattress-human contact status, the predicted pressure deviation value within the target pressure area A after adjustment is obtained. To adjust the anterior pressure prediction field, To adjust the pressure prediction field, , and For normalized weights; S10, adjust the hardness adjustment amount Send to the mattress zone actuator to adjust the target adjustable zone set. and adjacent support partition set The support hardness is adjusted, and the pressure distribution matrix and temperature distribution matrix are reacquired after adjustment. The parameters of the digital twin model of the mattress-human contact state are corrected for the next sampling cycle; S11, in response to the backoff trigger electrical signal input by the hardware interface or external terminal, the hardness adjustment corresponding to the current sampling cycle is canceled. Or revert to the previous stored hardness state.

2. The method according to claim 1, characterized in that, The filtering in S1 includes Kalman filtering, the smoothing process includes moving average smoothing, and adjacent grid interpolation is used to compensate for missing grid cells.

3. The method according to claim 1, characterized in that, The target coordinate region instruction includes a polygonal region, a rectangular region, a grid number set, or a combination thereof in the mattress coordinate system.

4. The method according to claim 1, characterized in that, S2 identifies human lying posture by reclining as supine, left lateral, right lateral, semi-reclining, or sitting, based on the long axis of the pressure-connected region, the position of the pressure center of gravity, and the relative distance between multiple main support points.

5. The method according to claim 1, characterized in that, and calibration coefficients in , , , and The calibration is determined by multi-level hardness calibration, which includes: applying standard loads to corresponding zones under at least three hardness states, measuring the zone compression and unloading response time, and fitting the calibration coefficients based on the load, compression, and unloading response time.

6. The method according to claim 1, characterized in that, The pressure deviation value It is obtained by weighting at least two of the following: the average grid pressure exceeding the pressure threshold, the maximum pressure value, and the percentage of the area exceeding the threshold within the target pressure zone A.

7. The method according to claim 1, characterized in that, The temperature deviation It is obtained by weighting at least two of the following: the average grid temperature exceeding the temperature threshold, the maximum temperature value, and the temperature rise rate within the target pressure zone A.

8. The method according to claim 1, characterized in that, The sustained pressure value The number of sampling periods in which the pressure deviation value within the target pressure area A continuously meets the threshold condition is determined.

9. The method according to claim 1, characterized in that, Generate hardness adjustment amount At that time, the target adjustable partition set and adjacent support partition set The number of partitions participating in the optimization is limited to no more than six. The candidate hardness adjustment levels for each partition are {-3dU, -2dU, -dU, 0, dU, 2dU, 3dU}, where dU is the single-level hardness adjustment step size. Candidate combinations are enumerated to select the objective function. Minimum hardness adjustment amount that satisfies pressure transfer constraints .

10. The method according to claim 1, characterized in that, The mattress zoning actuator includes at least one of an airbag actuator, a liquid-filled actuator, and a mechanical support actuator.

11. The method according to claim 1, characterized in that, The parameter boundaries in S7 include: The value range is from 0.01 to 0.

25. The value range is from 0.001 to 0.

20. The value range is from 0.001 to 0.

10. , and For normalized weights and , , and For normalized weights and , Take a value between 0.45 and 0.

85.

12. The method according to claim 1, characterized in that, Equivalent compression in S5 Determined according to the following formula: ,in, For the first Initial preload pressure of each grid cell, For minimum effective stiffness, For maximum compression, Use 5% to 35% of the mattress thickness.

13. The method according to claim 1, characterized in that, The iterative correction in S7 uses recursive least squares, with the forgetting factor of recursive least squares. The regularization coefficient is set to between 0.90 and 0.

995. Pick to The maximum number of iterations is 5 to 50, and the objective function of two consecutive iterations is... The iteration stops when the decrease rate is less than 0.5%.

14. The method according to claim 1, characterized in that, Calculation in S9 At that time, adjust the candidate hardness amount Substitution , The calibration mapping is used to obtain the adjusted equivalent stiffness and equivalent damping, and the pressure prediction field is recalculated based on the adjusted equivalent stiffness and equivalent damping; where , and satisfy , Take 3% to 20% of the pressure sensor's range, or 5% to 30% of the average pressure of the target pressure area before adjustment.

15. The method according to claim 1, characterized in that, S10 will adjust the hardness amount Converting the signal into an actuator drive signal includes: for airbag actuators, Converted to target pressure increment For mechanically supported actuators, Converted to support stroke increment For liquid bladder actuators, Convert to target hydraulic increment ,in , and Determined through factory calibration or no-load calibration.

16. The method according to claim 1, characterized in that, The previously stored hardness state is stored in a circular cache, and the cache content includes the sampling timestamp and the partition hardness state. Hardness adjustment amount , actuator driving signals and supporting risk values The cache depth is 30 to 100 sampling periods.

17. The method according to claim 1, characterized in that, The retraction trigger electrical signal includes at least one of the following: hardware button level signal, touch screen control trigger signal, or external terminal communication command, and the hardness change required for synchronous retraction exceeds the maximum adjustment range in a single operation. At that time, not greater than The amplitude is rolled back step by step according to the sampling period.

18. A mattress support firmness adaptive control system based on digital twin for a target pressure area, characterized in that, include: Pressure sensor array, used to output the pressure distribution matrix on the mattress surface. Temperature sensor array, used to output the temperature distribution matrix of the mattress surface or contact interface. ; The area instruction receiving module is used to receive target coordinate area instructions input from an external terminal; The pose recognition module is used to determine the pose based on the pressure distribution matrix. Determine the human lying posture and the relative position of the human body and the mattress coordinate system; The region mapping module is used to determine the target pressure region A and the target adjustable partition set based on the target coordinate region command. and adjacent support partition set ; Digital twin modeling module, used to build discrete mattress meshes Human contact area Zoned hardness status Equivalent stiffness parameters Equivalent damping parameters Thermal conductivity parameters Pressure prediction field and temperature prediction field A digital twin model of the mattress and its contact with the human body; The model calibration module is used to construct the objective function based on pressure prediction error, temperature prediction error, and model parameter smoothing constraints. Iterative correction within the preset parameter boundaries. , and ; The support risk calculation module is used to calculate the pressure deviation value within the target pressure zone A. Temperature deviation value Continuous pressure value and supporting risk value ; The hardness adjustment value generation module is used to adjust the support risk value. Greater than or equal to the support risk threshold At that time, by using a constrained optimization problem that includes a target region pressure deviation term, an adjacent region pressure transfer penalty term, and an adjustment magnitude penalty term, a set of target adjustable partitions is generated. and adjacent support partition set Hardness adjustment amount ; Mattress zone adjustment mechanism, used to adjust the amount of pressure according to the mattress. Adjust the support firmness of the corresponding mattress zones; The rollback control module is used to cancel the hardness adjustment corresponding to the current sampling period in response to a rollback trigger electrical signal input from the hardware interface or an external terminal. Or revert to the previous stored hardness state.

Images