A method and system for generating a sleep pillow parameter

Abstract

This invention belongs to the field of pillow design technology and provides a method and system for generating pillow parameters. The method includes: acquiring a multi-pose 3D model of a target individual and generating an initial set of physiological target parameters through simulation; constructing a coupled design model including a set of physiological target parameters, a set of geometric parameters, and a set of material distribution parameters; inputting the initial set of physiological target parameters into the coupled design model as the variable to be optimized in the outer loop, and assigning initial values ​​to the geometric parameter set and material distribution parameter set in the inner loop; performing nested optimization of the inner and outer loops; terminating the optimization and outputting the optimal parameter triplet when the nested optimization reaches the convergence condition, wherein the optimal parameter triplet includes the optimal set of physiological target parameters, the corresponding optimal set of geometric parameters, and the optimal set of material distribution parameters. This invention effectively solves the technical problems of fragmented design, limited adaptability, and difficulty in simultaneously achieving multi-objective collaborative optimization in existing technologies.

Description

Technical Field

[0001] This invention relates to the field of pillow design technology, and in particular to a method and system for generating pillow parameters. Background Technology

[0002] The core function of a pillow is to provide adequate support for the head and neck during sleep to maintain the physiological curvature of the cervical spine and relieve muscle fatigue. Traditional pillow designs are mostly based on average body size data, and their shape, height, and firmness are fixed, making it difficult to adapt to the physiological curves, sleeping postures, and individual needs of different users. This may result in insufficient or excessive support, affecting sleep quality and spinal health.

[0003] To improve adaptability, two main directions of improvement have emerged in existing technologies. One is the "zoned sleeping pillow," which has pre-defined physical divisions for supine and side-lying positions on the pillow body, using different heights and firmnesses in each zone to accommodate different sleeping postures. However, this method has significant limitations: First, its zoning parameters are mostly based on experience and lack precise data support for individual users; second, fixed physical zoning may create gaps in support or abrupt transitions when the user turns over, affecting comfort; finally, its design focuses primarily on macroscopic geometry and fails to coordinate with the non-uniform material distribution inside the pillow core, limiting further optimization of support performance.

[0004] Secondly, there is personalized customization technology based on 3D scanning, which uses user body data to customize the pillow shape. However, this type of technology is usually just a "one-to-one" shape replication or simple parameter adjustment. Its design process often treats shape design, material distribution, and biomechanical objectives separately. For example, a target shape is first determined based on a model, and then a matching material is selected. This process fails to treat material properties as an active design variable and integrate them with the target shape for optimization. More importantly, existing methods usually only optimize for a single standard sleeping position (such as supine), or simply design and stitch together multiple sleeping position targets independently. They lack a mathematical model and solution framework that can systematically and collaboratively optimize multiple sleeping position requirements, macroscopic geometry, and microscopic material gradient distribution. This results in the final solution possibly being only a compromise of multiple independent sub-objectives, rather than a solution with optimal overall performance.

[0005] Therefore, there is an urgent need in this field for an intelligent generation method that can deeply integrate human biomechanics, parametric geometric design and material distribution optimization, and automatically generate pillow parameters that can provide personalized, continuous and adaptive support for specific users in various sleeping postures. Summary of the Invention

[0006] To address the aforementioned technical problems, this invention provides a method and system for generating pillow parameters. By constructing a multi-level coupled model and a nested optimization framework, it provides an intelligent method that can automatically and accurately generate pillow parameters that are highly personalized, adaptable to multiple postures, and have good manufacturability. This effectively solves the technical problems of fragmented design, limited adaptability, and difficulty in achieving multi-objective collaborative optimization in the prior art.

[0007] This invention provides a method for generating pillow parameters, the method comprising the following steps:

[0008] Obtain multi-pose 3D models of the target individual and generate an initial set of physiological target parameters through simulation;

[0009] A coupled design model is constructed, comprising a set of physiological target parameters, a set of geometric parameters, and a set of material distribution parameters. The set of geometric parameters is used to define the continuous curved surface morphology of the pillow core, and the set of material distribution parameters is used to define the material gradient distribution inside the pillow core.

[0010] The initial set of physiological target parameters is input into the coupled design model as the variables to be optimized in the outer loop, and initial values ​​are assigned to the geometric parameter set and material distribution parameter set of the inner loop.

[0011] Perform nested optimization, wherein the nested optimization includes iteratively adjusting the set of physiological target parameters in the outer loop and co-optimizing the set of geometric parameters and the set of material distribution parameters based on the set of physiological target parameters updated by the outer loop in the inner loop, and evaluating the comprehensive performance index of the three in real time;

[0012] When the nested optimization reaches the convergence condition, the optimization is terminated and the optimal parameter triplet is output, wherein the optimal parameter triplet includes the optimal physiological target parameter set, the corresponding optimal geometric parameter set, and the optimal material distribution parameter set.

[0013] Preferably, the step of acquiring a multi-pose 3D model of the target individual and generating an initial set of physiological target parameters through simulation includes:

[0014] Obtain three-dimensional models of the head, neck, and shoulders of the target individual in supine and lateral positions;

[0015] Based on the three-dimensional model of the head, neck and shoulder region, the target support surface parameters corresponding to each posture were determined by biomechanical simulation.

[0016] Based on the target support surface parameters corresponding to the supine and lateral positions, the initial physiological target parameter set is generated through parametric fusion.

[0017] Preferably, the construction of the coupled design model includes:

[0018] A collaborative optimization model is established with a set of physiological target parameters as input and a set of geometric parameters and a set of material distribution parameters as optimization variables. The set of physiological target parameters is used to define the control point coordinates and curvature constraints of the target support surface.

[0019] An explicit mathematical relationship is constructed between the set of geometric parameters and the set of material distribution parameters. The explicit mathematical relationship is a coupled optimization objective function that includes the following constraints: the macroscopic surface morphology defined by the set of geometric parameters and the material gradient distribution defined by the set of material distribution parameters must satisfy the biomechanical simulation boundary conditions derived from the current set of physiological objective parameters.

[0020] Based on the coupled optimization objective function and the collaborative optimization model, a coupled design model is formed among the physiological target parameter set, geometric parameter set, and material distribution parameter set.

[0021] Preferably, the step of inputting the initial physiological target parameter set into the coupled design model as the variable to be optimized in the outer loop, and assigning initial values ​​to the geometric parameter set and material distribution parameter set of the inner loop, includes:

[0022] Substitute the initial set of physiological target parameters into a preset parameter response surface model to calculate the initial set of macroscopic geometric parameters corresponding to the initial set of physiological target parameters;

[0023] Based on the initial set of macroscopic geometric parameters, the initial values ​​of the material distribution parameter set in the inner loop are generated using a finite element mesh mapping algorithm.

[0024] Preferably, the nested optimization includes:

[0025] In each iteration of the outer loop, an updated set of physiological target parameters is generated;

[0026] The updated set of physiological target parameters generated by the outer loop in the current iteration is used as the optimization target input of the inner loop;

[0027] In the inner loop, a collaborative optimization subproblem is constructed using the updated set of physiological target parameters as boundary conditions, and the set of geometric parameters and the set of material distribution parameters are updated by solving the collaborative optimization subproblem.

[0028] Based on the updated set of geometric parameters, the set of material distribution parameters, and the updated set of physiological target parameters, biomechanical performance indicators and manufacturing feasibility indicators are calculated using the finite element method. Based on a predefined weighting relationship, the comprehensive performance index value for the current iteration is calculated.

[0029] Preferably, when the nested optimization reaches the convergence condition, terminating the optimization and outputting the optimal parameter triplet includes:

[0030] After the outer loop completes one iteration, the updated set of physiological target parameters and their corresponding comprehensive performance index values ​​generated by that iteration are obtained.

[0031] Determine whether the convergence condition is met, wherein the convergence condition is: in the most recent K consecutive outer loop iterations, the standard deviation of the comprehensive performance index value is less than a first threshold, and the norm of the parameter gradient vector calculated from the updated physiological target parameter set is less than a second threshold.

[0032] When the convergence condition is met, the iteration with the best comprehensive performance index value is selected from the candidate parameter set of the most recent K consecutive iterations.

[0033] The set of physiological target parameters generated by the outer loop in the optimal iteration, and the set of geometric parameters and the set of material distribution parameters obtained by co-optimization in the corresponding inner loop are used as the final optimal parameter triplet.

[0034] Preferably, the generation of the first threshold includes:

[0035] During the nested optimization process, the sequence of comprehensive performance index values ​​generated by the most recent M outer loop iterations is obtained in real time.

[0036] Based on the comprehensive performance index value sequence, calculate its moving average and moving standard deviation;

[0037] The first threshold to be satisfied in the current iteration cycle is dynamically determined based on the product of the preset convergence sensitivity coefficient and the moving standard deviation.

[0038] Preferably, the generation of the second threshold includes:

[0039] During the nested optimization process, the update magnitude of the physiological target parameter set in each iteration of the outer loop is recorded;

[0040] Based on the historical update magnitude sequence, the average rate of change of the physiological target parameter set in the most recent N iterations is calculated;

[0041] Set the second threshold to the output value of a function that is negatively correlated with the average rate of change.

[0042] The present invention also provides a pillow parameter generation system for performing the pillow parameter generation method described above, the generation system comprising:

[0043] The initial parameter generation module is used to obtain multi-pose 3D models of the target individual and generate an initial set of physiological target parameters through simulation.

[0044] The parameter model construction module is used to construct a coupled design model including a set of physiological target parameters, a set of geometric parameters, and a set of material distribution parameters. The set of geometric parameters is used to define the continuous curved surface shape of the pillow core, and the set of material distribution parameters is used to define the material gradient distribution inside the pillow core.

[0045] The loop initialization module is used to input the initial physiological target parameter set into the coupled design model as the variable to be optimized in the outer loop, and to assign initial values ​​to the geometric parameter set and material distribution parameter set of the inner loop.

[0046] The performance evaluation module is used to perform nested optimization, wherein the nested optimization includes iteratively adjusting the set of physiological target parameters in the outer loop and co-optimizing the set of geometric parameters and the set of material distribution parameters based on the set of physiological target parameters updated by the outer loop in the inner loop, and evaluating the comprehensive performance index of the three in real time.

[0047] The optimal parameter output module is used to terminate the optimization and output the optimal parameter triplet when the nested optimization reaches the convergence condition. The optimal parameter triplet includes the optimal physiological target parameter set, the corresponding optimal geometric parameter set, and the optimal material distribution parameter set.

[0048] Compared with related technologies, the method and system for generating pillow parameters provided by the present invention have the following beneficial effects:

[0049] This invention establishes a three-layer coupled model encompassing physiological objectives, geometric shape, and material distribution, and introduces a nested optimization algorithm. This allows for the simultaneous optimization of a unified design that satisfies the biomechanical requirements of both supine and lateral sleeping positions. The output of this design is a seamless, continuous surface and a matched material gradient field, rather than physically pieced-together partitions. This provides a smooth and continuous support transition when the user changes sleeping positions, significantly improving comfort and adaptability.

[0050] This invention integrates the traditionally fragmented design steps (setting goals, designing shape, and selecting materials) into a closed-loop optimization framework. The set of physiological target parameters acts as a bridge connecting human needs and product realization, driving the co-evolution of the macroscopic geometric parameter set and the microscopic material distribution parameter set. This method allows the final shape and internal structure of the pillow core to be automatically generated by the algorithm based on explicit biomechanical goals (such as minimizing intervertebral disc pressure and maintaining cervical curvature) and manufacturing constraints. This achieves an intelligent and precise mapping from "functional definition" to "manufacturable structure," overcoming the limitations of relying on designer experience.

[0051] This invention utilizes parametric models and real-time evaluation based on finite element simulation. The design process is grounded in quantifiable biomechanical and physical simulations, resulting in more scientific and reliable outcomes. The nested optimization mechanism, particularly the dynamic convergence threshold method, intelligently balances optimization accuracy and computational efficiency, accelerating the convergence process. Simultaneously, the optimization process incorporates manufacturing feasibility indicators (such as material utilization), ensuring that the output optimal parameter triples can be directly used to guide digital manufacturing processes such as 3D printing. This avoids the problem of idealized designs being difficult to manufacture, achieving a high-fidelity conversion from digital models to physical products. Attached Figure Description

[0052] Figure 1 A flowchart illustrating a method for generating pillow parameters provided by the present invention;

[0053] Figure 2 The present invention provides a module structure diagram of a pillow parameter generation system. Detailed Implementation

[0054] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the drawings, not all structures. Moreover, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0055] It should also be noted that, for ease of description, the accompanying drawings show only the parts relevant to the invention and not all of them. Before discussing exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations (or steps) as sequential processes, many of the operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but it may also have additional steps not included in the drawings. The process may correspond to a method, function, procedure, subroutine, subroutine, etc.

[0056] Example 1

[0057] This invention provides a method for generating pillow parameters, referencing... Figure 1 As shown, the generation method includes the following steps:

[0058] S1: Obtain multi-pose 3D models of the target individual and generate an initial set of physiological target parameters through simulation.

[0059] Specifically, step S1 includes the following steps:

[0060] S11: Obtain a three-dimensional model of the head, neck, and shoulder area of ​​the target individual in supine and lateral positions.

[0061] In this embodiment, the anatomical data of the target individual must first be acquired. A high-precision 3D human scanner is used to scan the target individual in a controlled environment. To obtain an accurate supine posture model, the individual lies flat on a calibration plate, ensuring their head is in a naturally relaxed state with the Frankfurt plane perpendicular to the horizontal plane. A suspended or gantry scanner is used to scan from above to obtain complete 3D point cloud data including the head, neck, and upper shoulder edges. For the lateral posture, the individual lies in their habitual lateral position with their head resting on a temporary support to simulate a sleeping state. Similarly, a scanner is used to acquire 3D point clouds of the head, neck, and shoulder contours from the side. Subsequently, the acquired raw point cloud data is preprocessed, including noise removal, smoothing, and meshing to generate a 3D surface model (typically in STL or OBJ format) that can be used for subsequent analysis. To ensure model accuracy, the scanning resolution should be better than 1 mm.

[0062] S12: Based on the three-dimensional model of the head, neck and shoulder region, the target support surface parameters corresponding to each posture are determined by biomechanical simulation.

[0063] In this embodiment, after obtaining an accurate three-dimensional model, the curved shape of the pillow core under ideal support needs to be determined through biomechanical simulation.

[0064] The specific process is as follows: First, import the obtained three-dimensional model into a professional finite element analysis software (such as Abaqus or ANSYS) and assign it biomechanical properties. For example, model the cervical vertebrae as a rigid body, the intervertebral discs as a hyperelastic material, and the surrounding muscles and ligaments as tension elements, thereby constructing a simplified head-neck-shoulder biomechanical model.

[0065] Then, boundary conditions are defined: for the supine position, the shoulder position is fixed, and a gravitational load is applied to the head; for the lateral position, the corresponding positions of the shoulder and the side of the head are fixed. The core of the simulation is inverse solving: by iteratively adjusting the shape parameters (such as curvature and key point height) of a "virtual support surface" in contact with the head and neck, static simulation is run until the cervical spine curve in the biomechanical model reaches the predefined "physiological neutral position" (i.e., the ideal state where the pressure distribution of each intervertebral disc is uniform and the joint stress is minimal). At this point, the shape parameters of the "virtual support surface" (for example, represented by the coordinates of a set of control points of a NURBS surface) are the target support surface parameters obtained under the current posture.

[0066] S13: Based on the target support surface parameters corresponding to the supine and lateral positions, the initial physiological target parameter set is generated through parametric fusion.

[0067] In this embodiment, since the final requirement is a pillow that can adapt to two sleeping positions, it is necessary to merge the two target support surface parameter sets obtained independently in step S12 to generate a unified initial parameter set as the starting point for subsequent optimization.

[0068] The parametric fusion here is not a simple averaging, but a weighted and topological merging based on biomechanical principles.

[0069] Specifically, firstly, two NURBS surfaces representing ideal support in supine and lateral positions are aligned in three-dimensional space. Then, the overlapping regions (e.g., the neck support area) and unique regions (e.g., the head depression area in lateral position) of the two surfaces are analyzed. For overlapping regions, a weighted average algorithm based on support area or mechanical importance is used to calculate the fused control point coordinates; for unique regions, their characteristics (e.g., height compensation required in lateral position) are integrated into the fused surface. Finally, the control point coordinates, weights, and overall curvature constraints embodied by this fused, unified NURBS surface together constitute the initial physiological target parameter set used to initiate subsequent optimization processes. This set defines a preliminary ideal support surface target that takes into account both attitude requirements.

[0070] S2: Construct a coupled design model including a set of physiological target parameters, a set of geometric parameters, and a set of material distribution parameters, wherein the set of geometric parameters is used to define the continuous curved surface shape of the pillow core, and the set of material distribution parameters is used to define the material gradient distribution inside the pillow core.

[0071] Specifically, step S2 includes the following steps:

[0072] S21: Establish a collaborative optimization model with a set of physiological target parameters as input and a set of geometric parameters and a set of material distribution parameters as optimization variables, wherein the set of physiological target parameters is used to define the coordinates of the control points and the curvature constraints of the target support surface.

[0073] In this embodiment, the first step in implementation is to construct a collaborative optimization mathematical framework. The core of this model is to use the initial physiological target parameter set obtained in step S1 (… Using the macroscopic geometry of the pillow core (described by the geometric parameter set P) and the internal material distribution (described by the material distribution parameter set P) as input conditions, the system takes these parameters as input conditions. (Description) Together, they serve as design variables to be optimized. The geometric parameter set P can be parameterized using NURBS surfaces, specifically including control point coordinate matrices, node vectors, and weighting factors; its dimensions determine the degrees of freedom of the surface shape. Material distribution parameter set This approach parameterizes the non-uniform material property field within the pillow core by discretizing its three-dimensional space into a voxel grid, assigning each voxel a scalar value representing the material density or modulus. The objective function of this collaborative optimization model aims to minimize the set of physiological target parameters. The biomechanical performance difference between the defined ideal support surface and the actual designed surface is considered, while also satisfying manufacturing process constraints. The mathematical expression of the optimization problem can be formalized as: given... Under the given conditions, find the optimal P and , so that the objective function Obtain the minimum value.

[0074] S22: Construct an explicit mathematical association between the set of geometric parameters and the set of material distribution parameters, wherein the explicit mathematical association is a coupled optimization objective function including the following constraints, wherein the constraints are that the macroscopic surface morphology defined by the set of geometric parameters and the material gradient distribution defined by the set of material distribution parameters must satisfy the biomechanical simulation boundary conditions derived from the current physiological objective parameter set.

[0075] In this embodiment, this step is crucial for achieving co-design of geometry and materials, requiring the establishment of P and The explicit mathematical relationship between them, rather than treating them as independent variables.

[0076] Specifically, this is achieved by constructing a coupled optimization objective function. This function not only includes the main objective term that measures the design performance (such as the deviation of cervical intervertebral disc pressure from the target value), but more importantly, it introduces constraints that force the coupling between geometry and material distribution.

[0077] These constraints stem from the current set of physiological target parameters ( Boundary conditions derived through biomechanical simulation. For example, the constraint condition can be expressed as: for a point on the pillow core surface determined by the geometric parameter set P, its normal displacement under head pressure must be equal to that determined by the material distribution parameter set P. The amount of compression deformation of the internal material of the pillow core below the determined point, calculated using linear elastic or hyperelastic constitutive relations under the same pressure.

[0078] This essentially links macroscopic geometric deformation with microscopic material mechanical responses through physical equations (such as an extended form of Hooke's Law). The specific formulation of this coupling objective function is as follows:

[0079]

[0080]

[0081] in It is a set of geometric parameters, a vector, that defines the macroscopic continuous surface shape of the pillow core. For example... It can represent the three-dimensional coordinates of n control points that define the NURBS surface of the pillow core surface;

[0082] It is a set of material distribution parameters, a vector that defines the material gradient distribution inside the pillow core. For example... It can represent the relative material density or elastic modulus of each voxel after discretizing the interior of the pillow core into m voxels (or finite elements);

[0083] It is a set of physiological target parameters given by the outer loop, which serves as the target and boundary conditions for the inner optimization. It defines the control point coordinates and curvature constraints of the target supporting surface;

[0084] This indicates a biomechanical performance indicator. This indicator quantifies the performance characteristics in the current design. Below, the biomechanical response (such as cervical intervertebral disc pressure) obtained through finite element simulation calculations. Muscle stress ) and by Defined ideal physiological target value ( The deviation between the two values ​​is a common form of the weighted sum of squares.

[0085] This is a manufacturing feasibility indicator. It penalizes material distributions that might lead to manufacturing difficulties or excessive costs; for example, it encourages smooth transitions in material distribution and avoids overly drastic gradient changes. A gradient-based regularization term can be used.

[0086] These are the core coupling constraint equations. This system of equations enforces the geometry. With material distribution Satisfy by The derived biomechanical simulation boundary conditions. Specifically, it requires boundary conditions derived from geometric surfaces. The displacement (or reaction force) at the defined contact boundary must be related to the material distribution. and constitutive models (such as linear elasticity) The physical field results obtained by solving the problem using the finite element method are consistent.

[0087] The coupling objective function explicitly includes geometric parameters. and material distribution parameters Through physical equations By linking them together, any change in one aspect during the optimization process will directly affect the optimization direction of the other aspect through constraints, thereby achieving integrated and collaborative design of macroscopic morphology and microscopic material distribution.

[0088] S23: Based on the coupled optimization objective function and the collaborative optimization model, a coupled design model is formed between the physiological target parameter set, the geometric parameter set, and the material distribution parameter set.

[0089] In this embodiment, based on completing S21 and S22, this step aims to integrate the aforementioned components to form a complete and computable coupled design model. Specific implementation includes the computational realization of the model: integrating the collaborative optimization framework defined in S21 with the coupled objective function and its constraints constructed in S22 on a numerical computation platform (e.g., MATLAB's optimization toolbox combined with COMSOL Multiphysics' physics simulation capabilities, or using Abaqus scripts in conjunction with Python's optimization library).

[0090] The parameters of the optimization algorithm need to be precisely set, for example, for the outer loop (optimization). A global optimization algorithm (such as a genetic algorithm) can be selected to optimize the inner loop. and Gradient-based algorithms (such as sequential quadratic programming) can be selected. The final coupled design model is a model that includes the input (initial...) or updated Design variables The model is a complete computational system consisting of a coupled objective function and constraints, as well as an optimization solver. It can receive a set of physiological target parameters and automatically output the optimal set of geometric parameters and material distribution parameters that match them, providing the core computational engine for subsequent nested optimization loops.

[0091] S3: Input the initial physiological target parameter set into the coupled design model as the variable to be optimized in the outer loop, and assign initial values ​​to the geometric parameter set and material distribution parameter set of the inner loop.

[0092] Specifically, step S3 includes the following steps:

[0093] S31: Substitute the initial physiological target parameter set into a preset parameter response surface model to calculate the initial macroscopic geometric parameter set corresponding to the initial physiological target parameter set.

[0094] In this embodiment, the first step is to construct or obtain a pre-trained parametric response surface model. This model is a surrogate model that establishes a nonlinear mapping relationship from the set of physiological target parameters to the set of macroscopic geometric parameters.

[0095] The construction process is as follows: During the system development phase, a large number of sample points are generated within a reasonable physiological target parameter space using sampling techniques (such as Latin hypercube sampling). For each sample point (i.e., a set of virtual physiological target parameters), a corresponding, biomechanically reasonable preliminary macroscopic geometry is calculated through simplified biomechanical simulation or based on expert experience rules. This generates a large-scale training dataset containing the correspondence between physiological target parameters and macroscopic geometric parameters. Using this dataset, a regression model (e.g., Gaussian process regression, radial basis function neural network, or support vector regression) is trained as a parametric response surface model.

[0096] In the actual execution of step S31, the initial physiological target parameter set generated in step S1 for a specific target individual is used. As input, these parameters are substituted into this trained response surface model. The model will quickly predict and output a response surface that is consistent with... Correspondingly, a more reasonable initial macroscopic geometric parameter set from a biomechanical perspective .this This provides a high-quality starting point for subsequent optimizations, far superior to random initialization, and can significantly accelerate the convergence speed of the inner loop. For example, if the geometric parameter set P is defined by the coordinates of the control points of the NURBS surface, then the response surface model outputs the initial coordinate matrix of these control points.

[0097] S32: Based on the initial macroscopic geometric parameter set, generate the initial values ​​of the material distribution parameter set in the inner loop using the finite element mesh mapping algorithm.

[0098] In this embodiment, an initial macroscopic geometric parameter set is obtained. Then, initial values ​​for a matching set of material distribution parameters need to be generated. .

[0099] The specific implementation process is as follows: First, based on... The defined external geometric boundary of the pillow core is used to discretize the three-dimensional space inside the pillow core using automatic mesh generation techniques (such as Dela triangulation or advanced front method), generating a finite element mesh. This mesh contains a large number of elements (such as tetrahedral or hexahedral elements), each element corresponding to a set of material distribution parameters. One of the parameters.

[0100] Then, the finite element mesh mapping algorithm is executed. The core idea of ​​this algorithm is to heuristically assign initial material properties based on geometric features such as curvature and thickness variations. Specifically, it calculates the coordinates of the center of each element in the mesh, as well as the geometric features at that point (such as the distance to the surface and the local radius of curvature).

[0101] Subsequently, according to preset mapping rules, each cell is assigned an initial material density or modulus value. For example, a simple rule could be: cells closer to the pillow core surface, or those with greater local curvature, have a higher initial material density to provide stronger support; conversely, cells in the core area of ​​the pillow core have a lower material density. By traversing all mesh cells, a mapping with the initial geometry can be generated. Matching, non-uniform initial material distribution field .this As the starting point for optimizing the distribution of inner layer materials, it reflects the basic constraints of macroscopic geometry on material distribution, laying a reasonable foundation for subsequent collaborative optimization.

[0102] S4: Perform nested optimization, wherein the nested optimization includes iteratively adjusting the set of physiological target parameters in the outer loop and co-optimizing the set of geometric parameters and the set of material distribution parameters based on the set of physiological target parameters updated by the outer loop in the inner loop, and evaluating the comprehensive performance index of the three in real time.

[0103] Specifically, step S4 includes the following steps:

[0104] S41: In each iteration of the outer loop, an updated set of physiological target parameters is generated.

[0105] In this embodiment, the outer loop employs a global optimization algorithm (such as a genetic algorithm, particle swarm optimization, or simulated annealing) to explore the set of physiological target parameters. The optimization space is defined. At the beginning of each iteration, the algorithm generates new candidate solutions based on the state of the current population (or the current solution) through specific operational rules. For example, if a genetic algorithm is used, selection, crossover, and mutation operators are used to optimize the current generation. The parameter set is processed to generate a new generation. Parameter set population. Each newly generated individual represents an updated set of physiological target parameters. This process aims to globally find physiological target parameters that improve the overall objective function. Key parameters such as population size, crossover rate, and mutation rate need to be pre-set to ensure a good balance between exploration and exploitation in the algorithm. After generating the updated set of physiological target parameters, it is passed to the inner loop as a new target and input condition for the inner optimization.

[0106] S42: Use the updated set of physiological target parameters generated by the outer loop in the current iteration as the optimization target input of the inner loop.

[0107] In this embodiment, the step involves connecting the data interface between the outer loop and the inner loop. In practice, an effective data transfer mechanism needs to be established. This occurs when the outer loop completes one iteration and generates one or more updated sets of physiological target parameters. Afterwards, the system encapsulates these updated physiological target parameter sets into a standardized data format (e.g., a JSON file or a specific data structure) readable by the inner loop. Then, the optimization program of the inner loop is called, loading the encapsulated updated physiological target parameter sets as its input parameters. Before the inner loop begins solving, the boundary conditions, such as the coordinates of the new target support surface control points and curvature constraints defined by the updated physiological target parameter sets, need to be updated into the constraint equations of the coupled design model constructed in S2; that is, the coupling constraint conditions are updated. In This ensures that the goal of this inner loop optimization is to approximate the new physiological target proposed by the current outer loop.

[0108] S43: In the inner loop, a co-optimization subproblem is constructed using the updated physiological target parameter set as boundary conditions, and the geometric parameter set and the material distribution parameter set are updated by solving the co-optimization subproblem.

[0109] In this embodiment, within the inner loop, for a given updated set of physiological target parameters... It is necessary to construct and solve a set of geometric parameters. and material distribution parameter set This is a collaborative optimization subproblem involving joint optimization variables. The mathematical model of this subproblem is based on the coupled optimization objective function and constraints defined in S22, but at this point... It is fixed (equal to the updated set of physiological target parameters).

[0110] The specific solution process typically employs gradient-based optimization algorithms (such as sequential quadratic programming, MMA, or alternating direction multiplier method). The algorithm starts from the initial values ​​provided by S3. and Initially, in each inner iteration, simultaneously update and During the update process, the algorithm needs to calculate the objective function. Compared to and The gradient, and considering coupling constraints. For example, the adjoint variable method can be used to efficiently calculate gradients, guiding... and The update direction ensures that, while satisfying Under the defined biomechanical boundary conditions, the objective function value continuously decreases. The inner loop iterates until it meets the convergence condition (e.g., the gradient norm is sufficiently small or the change in the objective function is less than a threshold), at which point the output... and That is, the current The (local) optimal geometric and material parameters.

[0111] S44: Based on the updated set of geometric parameters, the set of material distribution parameters, and the updated set of physiological target parameters, the biomechanical performance index and manufacturing feasibility index are calculated using the finite element method, and the comprehensive performance index value for the current iteration is calculated based on the predefined weight relationship.

[0112] In this embodiment, after obtaining the inner loop update and After that, a performance evaluation is required.

[0113] In practice, the current The triplet parameter set is configured into a high-fidelity finite element analysis model. This model is more accurate than the simplified model used in optimization. Finite element simulations are run to calculate key performance outputs, primarily including two types of metrics:

[0114] Biomechanical performance indicators: such as the average pressure, maximum pressure, pressure distribution uniformity, and deviation of cervical curvature from the ideal neutral position of the intervertebral discs.

[0115] Manufacturing feasibility indicators include, for example, the magnitude of the gradient of material density distribution (reflecting manufacturing difficulty), the total volume of required materials, or the total cost.

[0116] Then, the importance weights of each indicator are determined according to a predefined weighting relationship (e.g., by using the analytic hierarchy process or expert scoring). These heterogeneous indicators are normalized, weighted, and summed to obtain a scalarized comprehensive performance index value. The calculation formula can be expressed as:

[0117]

[0118] in It is the first A normalized performance metric value. Ultimately, the feedback is sent to the optimizer of the outer loop to evaluate the current... The advantages and disadvantages of each can be used to guide the search direction of the next generation of the outer loop.

[0119] S5: When the nested optimization reaches the convergence condition, the optimization is terminated and the optimal parameter triplet is output, wherein the optimal parameter triplet includes the optimal physiological target parameter set, the corresponding optimal geometric parameter set, and the optimal material distribution parameter set.

[0120] Specifically, step S5 includes the following steps:

[0121] S51: After the outer loop completes one iteration, obtain the updated set of physiological target parameters and their corresponding comprehensive performance index values ​​generated by the iteration.

[0122] In this embodiment, a robust data recording and tracking mechanism needs to be established during the nested optimization process. Each time the outer loop completes a full iteration (i.e., completes steps S41 to S44), the system needs to immediately capture and store the key data for that iteration. This includes:

[0123] 1) Updated set of physiological target parameters ( ): This refers to the specific parameter values ​​generated by the outer optimization algorithm (such as genetic algorithm) and optimized by the inner loop in this iteration, which are usually stored in vector or matrix form;

[0124] 2) Corresponding comprehensive performance index values ​​( ): That is, the scalar result obtained by high-fidelity finite element simulation calculation and weighted summation in step S44.

[0125] To achieve effective tracking, each iteration should be assigned a unique iteration number. The iteration number, the updated set of physiological target parameters, and the corresponding comprehensive performance index value should be written as a data unit to an optimization history database or a log file in a specific format (such as CSV or HDF5). This process ensures the complete traceability of the optimization process and provides a data foundation for subsequent convergence judgment and optimal solution selection.

[0126] S52: Determine whether the convergence condition is met, wherein the convergence condition is: in the most recent K consecutive outer loop iterations, the standard deviation of the comprehensive performance index value is less than a first threshold, and the norm of the parameter gradient vector calculated from the updated physiological target parameter set is less than a second threshold.

[0127] In this embodiment, this step is the core judgment for determining whether the optimization process should be terminated.

[0128] The specific implementation includes dynamic threshold calculation and dual condition checks:

[0129] 1. Data window extraction: Set a window size K (e.g., K=10) and extract the sequence of comprehensive performance index values ​​for the most recent K consecutive iterations from the optimization history database.

[0130] 2. First Condition Check (Performance Stability): Dynamic Generation of the First Threshold: Calculate the moving average and moving standard deviation of the sequence. The first threshold is not a fixed value but is dynamically generated; it is the product of a preset convergence sensitivity coefficient α and the moving standard deviation, where α = 0.05. This means that the smaller the fluctuation of the performance index in the most recent K iterations, the stricter the allowable standard deviation threshold. Judgment Condition: Check whether the moving standard deviation is less than the first threshold. If it is satisfied, it indicates that the objective function value has stabilized.

[0131] 3. Second condition check (parameter change rate): Parameter gradient norm calculation: For each of the most recent K iterations, calculate its physiological target parameter set. The update magnitude (i.e., differential gradient) relative to the previous iteration is used to calculate the L2 norm of the gradient vector. Dynamic generation of the second threshold: Calculate the average rate of change of the gradient norm over these K iterations. The second threshold is set to be negative of the average rate of change. Related functions, such as ,in This is a preset baseline threshold. This makes the convergence condition's requirement for the gradient norm more stringent as the parameter update magnitude naturally decreases in the later stages of optimization. Judgment condition: Check if the gradient norm of the most recent iteration is less than the second threshold. If satisfied, it indicates that the shift of the solution in the parameter space has become negligible.

[0132] 4. Convergence Determination: The entire nested optimization process is considered to have converged if and only if both of the above conditions are met simultaneously. Otherwise, the optimization will continue to the next outer iteration.

[0133] S53: When the convergence condition is met, select the iteration with the best comprehensive performance index value from the candidate parameter set of the most recent K consecutive iterations.

[0134] In this embodiment, once convergence is determined via S52, the final optimal solution is selected from the most recent K consecutive iterations used as the criterion. Specifically, the system queries the historical data of these K iterations and compares their recorded comprehensive performance index values. Since the optimization objective is usually to minimize Therefore, the system will execute a simple optimization algorithm to find... The iteration with the smallest value. For example, by iterating through K data units, the corresponding value can be found using the argmin function. The iteration number is used. The set of physiological target parameters corresponding to this iteration is initially selected as the optimal set of physiological target parameters. This step ensures that the output is the solution with the best overall performance during the convergence phase, rather than just the result of the last iteration.

[0135] S54: The set of physiological target parameters generated by the outer loop in the optimal iteration, and the set of geometric parameters and the set of material distribution parameters obtained by co-optimization in the corresponding inner loop, are taken as the final optimal parameter triplet.

[0136] In this embodiment, after determining the optimal iteration, it is necessary to completely extract all optimal parameters corresponding to that iteration from the storage system. Specifically:

[0137] 1. Obtain the set of physiological target parameters for this iteration from the optimization history.

[0138] 2. Based on the iteration number, locate and retrieve the final set of geometric parameters and material distribution parameters obtained after the convergence of the inner loop co-optimization in that outer iteration. These parameters are usually stored together with the physiological target parameter set at the end of the inner loop in step S43.

[0139] 3. The final set of geometric parameters, the set of material distribution parameters, and the other two parameter sets are packaged together to form a structured optimal parameter triplet. This triplet represents the globally optimal design solution that best balances biomechanical performance and manufacturing feasibility under convergence conditions.

[0140] 4. Finally, the system outputs this optimal parameter triplet as the formal optimization result, which can be converted into a standard file format (such as JSON, XML, or a specific binary format) to drive subsequent digital manufacturing processes. At the same time, the optimization process officially terminates.

[0141] Example 2

[0142] This invention also provides a pillow parameter generation system for executing the aforementioned pillow parameter generation method, with reference to... Figure 2 As shown, the generation system includes:

[0143] The initial parameter generation module 100 is used to acquire multi-pose three-dimensional models of the target individual and generate an initial physiological target parameter set through simulation.

[0144] The parameter model construction module 200 is used to construct a coupled design model including a set of physiological target parameters, a set of geometric parameters, and a set of material distribution parameters. The set of geometric parameters is used to define the continuous curved surface shape of the pillow core, and the set of material distribution parameters is used to define the material gradient distribution inside the pillow core.

[0145] The loop initialization module 300 is used to input the initial physiological target parameter set into the coupled design model as the variable to be optimized in the outer loop, and to assign initial values ​​to the geometric parameter set and material distribution parameter set of the inner loop.

[0146] The performance evaluation module 400 is used to perform nested optimization, wherein the nested optimization includes iteratively adjusting the set of physiological target parameters in the outer loop and co-optimizing the set of geometric parameters and the set of material distribution parameters based on the set of physiological target parameters updated by the outer loop in the inner loop, and evaluating the comprehensive performance index of the three in real time.

[0147] The optimal parameter output module 500 is used to terminate the optimization and output the optimal parameter triplet when the nested optimization reaches the convergence condition. The optimal parameter triplet includes the optimal physiological target parameter set, the corresponding optimal geometric parameter set, and the optimal material distribution parameter set.

[0148] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. 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, generate instructions for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0149] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, including read-only memory (ROM), random access memory (RAM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), one-time programmable read-only memory (OTPROM), electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, disk storage, magnetic tape storage, or any other computer-readable medium capable of carrying or storing data.

[0150] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

Claims

1. A method for generating pillow parameters, characterized in that, The generation method includes the following steps: Obtain multi-pose 3D models of the target individual and generate an initial set of physiological target parameters through simulation; Specifically, the generation of the initial physiological target parameter set includes: Obtain three-dimensional models of the head, neck, and shoulders of the target individual in supine and lateral positions; Based on the three-dimensional model of the head, neck and shoulder region, the target support surface parameters corresponding to each posture were determined by biomechanical simulation. Based on the target support surface parameters corresponding to the supine and lateral positions, the initial physiological target parameter set is generated through parametric fusion. A coupled design model is constructed, comprising a set of physiological target parameters, a set of geometric parameters, and a set of material distribution parameters. The set of geometric parameters is used to define the continuous curved surface morphology of the pillow core, and the set of material distribution parameters is used to define the material gradient distribution inside the pillow core. The construction of the coupled design model includes: A collaborative optimization model is established with a set of physiological target parameters as input and a set of geometric parameters and a set of material distribution parameters as optimization variables. The set of physiological target parameters is used to define the control point coordinates and curvature constraints of the target support surface. An explicit mathematical relationship is constructed between the set of geometric parameters and the set of material distribution parameters. The explicit mathematical relationship is a coupled optimization objective function that includes the following constraints: the macroscopic surface morphology defined by the set of geometric parameters and the material gradient distribution defined by the set of material distribution parameters must satisfy the biomechanical simulation boundary conditions derived from the current set of physiological objective parameters. Based on the coupled optimization objective function and the collaborative optimization model, a coupled design model is formed between the physiological target parameter set, the geometric parameter set, and the material distribution parameter set; The initial set of physiological target parameters is input into the coupled design model as the variables to be optimized in the outer loop, and initial values ​​are assigned to the geometric parameter set and material distribution parameter set of the inner loop. Perform nested optimization, wherein the nested optimization includes iteratively adjusting the set of physiological target parameters in the outer loop and co-optimizing the set of geometric parameters and the set of material distribution parameters based on the set of physiological target parameters updated by the outer loop in the inner loop, and evaluating the comprehensive performance index of the three in real time; When the nested optimization reaches the convergence condition, the optimization is terminated and the optimal parameter triplet is output, wherein the optimal parameter triplet includes the optimal physiological target parameter set, the corresponding optimal geometric parameter set, and the optimal material distribution parameter set.

2. The method of claim 1, wherein, The step of inputting the initial physiological target parameter set into the coupled design model as the variable to be optimized in the outer loop, and assigning initial values ​​to the geometric parameter set and material distribution parameter set of the inner loop, includes: Substitute the initial set of physiological target parameters into a preset parameter response surface model to calculate the initial set of macroscopic geometric parameters corresponding to the initial set of physiological target parameters; Based on the initial set of macroscopic geometric parameters, the initial values ​​of the material distribution parameter set in the inner loop are generated using a finite element mesh mapping algorithm.

3. The method of claim 2, wherein, The nested optimization includes: In each iteration of the outer loop, an updated set of physiological target parameters is generated; The updated set of physiological target parameters generated by the outer loop in the current iteration is used as the optimization target input of the inner loop; In the inner loop, a collaborative optimization subproblem is constructed using the updated set of physiological target parameters as boundary conditions, and the set of geometric parameters and the set of material distribution parameters are updated by solving the collaborative optimization subproblem. Based on the updated set of geometric parameters, the set of material distribution parameters, and the updated set of physiological target parameters, biomechanical performance indicators and manufacturing feasibility indicators are calculated using the finite element method. Based on a predefined weighting relationship, the comprehensive performance index value for the current iteration is calculated.

4. The method of claim 3, wherein, When the nested optimization reaches the convergence condition, the optimization is terminated and the optimal parameter triplet is output, including: After the outer loop completes one iteration, the updated set of physiological target parameters and their corresponding comprehensive performance index values ​​generated by that iteration are obtained. Determine whether the convergence condition is met, wherein the convergence condition is: in the most recent K consecutive outer loop iterations, the standard deviation of the comprehensive performance index value is less than a first threshold, and the norm of the parameter gradient vector calculated from the updated physiological target parameter set is less than a second threshold. When the convergence condition is met, the iteration with the best comprehensive performance index value is selected from the candidate parameter set of the most recent K consecutive iterations. The set of physiological target parameters generated by the outer loop in the optimal iteration, and the set of geometric parameters and the set of material distribution parameters obtained by co-optimization in the corresponding inner loop are used as the final optimal parameter triplet.

5. The method of claim 4, wherein, The generation of the first threshold includes: During the nested optimization process, the sequence of comprehensive performance index values ​​generated by the most recent M outer loop iterations is obtained in real time. Based on the comprehensive performance index value sequence, calculate its moving average and moving standard deviation; The first threshold to be satisfied in the current iteration cycle is dynamically determined based on the product of the preset convergence sensitivity coefficient and the moving standard deviation.

6. The method of claim 5, wherein, The generation of the second threshold includes: During the nested optimization process, the update magnitude of the physiological target parameter set in each iteration of the outer loop is recorded; Based on the historical update magnitude sequence, the average rate of change of the physiological target parameter set in the most recent N iterations is calculated; Set the second threshold to the output value of a function that is negatively correlated with the average rate of change.

7. A pillow parameter generation system, used to execute the pillow parameter generation method according to any one of claims 1 to 6, characterized in that, The generation system includes: The initial parameter generation module is used to obtain multi-pose 3D models of the target individual and generate an initial set of physiological target parameters through simulation. The parameter model construction module is used to construct a coupled design model including a set of physiological target parameters, a set of geometric parameters, and a set of material distribution parameters. The set of geometric parameters is used to define the continuous curved surface shape of the pillow core, and the set of material distribution parameters is used to define the material gradient distribution inside the pillow core. The loop initialization module is used to input the initial physiological target parameter set into the coupled design model as the variable to be optimized in the outer loop, and to assign initial values ​​to the geometric parameter set and material distribution parameter set of the inner loop. The performance evaluation module is used to perform nested optimization, wherein the nested optimization includes iteratively adjusting the set of physiological target parameters in the outer loop and co-optimizing the set of geometric parameters and the set of material distribution parameters based on the set of physiological target parameters updated by the outer loop in the inner loop, and evaluating the comprehensive performance index of the three in real time. The optimal parameter output module is used to terminate the optimization and output the optimal parameter triplet when the nested optimization reaches the convergence condition. The optimal parameter triplet includes the optimal physiological target parameter set, the corresponding optimal geometric parameter set, and the optimal material distribution parameter set.

Images