A computer simulation and optimization method for directional rubber skin paste fit compact

Abstract

The application discloses a computer simulation and optimization method for oriented rubber and meat close fitting, and belongs to the technical field of computer aided engineering and virtual simulation. A three-dimensional model of rubber and meat is constructed and material properties are defined, a virtual contact interface is constructed and is discretized into multiple contact units; a finite element analysis engine is used to simulate the rubber deformation process under different candidate pressure distribution modes, and corresponding deformation sequences are generated; the relative displacement between the rubber nodes and the meat nodes at each contact unit is extracted, and the overall fitting degree coefficient is calculated; the candidate pressure distribution mode is iteratively optimized until the optimized pressure distribution mode meeting the convergence condition is obtained, and is output as the target pressure application scheme. The application can automatically find the optimal pressure application scheme for the closest fitting of rubber and meat, effectively replace the traditional process debugging mode relying on experience and repeated experiments, significantly improve the fitting quality and efficiency, and provide accurate computer guidance for the formulation of actual production processes.

Description

Technical Field

[0001] This invention discloses a computer simulation and optimization method for achieving tight adhesion of oriented rubber to the flesh, belonging to the field of computer-aided engineering and virtual simulation technology. Background Technology

[0002] In the production and processing of food engineering, cultured meat, and bio-tissue engineering, it is often necessary to bond rubber-like materials with specific physical properties to the surface of meat-like substrates to achieve purposes such as product packaging, structural shaping, or functional layer construction. Traditional bonding processes typically rely on physical experiments and experience-based adjustments. Operators repeatedly try different pressure intensities, locations, and sequences, observing the actual effects after bonding to gradually determine a set of usable process parameters. This method, based on physical experiments, is not only time-consuming and material-intensive, but also suffers from the complex nonlinear mechanical properties of both the rubber and meat materials. Their deformation behavior is difficult to accurately predict through simple physical deduction, often resulting in a finalized process that falls short of the theoretically optimal bonding effect. Furthermore, for irregularly shaped meat pieces, traditional methods struggle to precisely control the tightness of the rubber bonding in various local areas, easily leading to problems such as localized wrinkling or weak bonding.

[0003] With the development of computer simulation technology, the finite element analysis method has been widely used to simulate the deformation process of various materials. By constructing a three-dimensional model of the object to be processed and assigning it corresponding material properties, the deformation process of the material under external force can be simulated in a virtual environment, thereby observing the influence of different process parameters on the final shape. However, most existing simulation methods for flexible material bonding remain at the level of verifying a single scheme. That is, operators propose one or more candidate pressure application schemes based on experience, and then simulate and compare them one by one through simulation software to select the relatively better scheme. Although this verification method can reduce some physical experiments, the proposal of the scheme still depends on human experience, and it is impossible to automatically explore the huge process parameter space, nor can it guarantee that the final selected scheme is the optimal solution in a global sense. At the same time, the simulation process and the subsequent scheme optimization stage are isolated from each other, lacking a complete technical solution that can organically combine the two and automatically iterate and optimize.

[0004] To address the aforementioned issues, a method is needed that deeply integrates 3D modeling, materials mechanics simulation, and intelligent optimization algorithms. This method would automatically simulate the deformation process of the rubber sheet under different pressure distribution modes and automatically adjust the pressure application scheme based on the evaluation indicators of the bonding effect, ultimately outputting the target pressure scheme that achieves the tightest bond between the rubber sheet and the meat piece. This method should be able to handle the nonlinear deformation characteristics of the rubber sheet material, accurately assess the local bonding quality of the bonding area, and visualize the simulated bonding effect. This would provide accurate and efficient guidance for the development of actual production processes, eliminating reliance on repeated physical experiments and experience-based adjustments. Summary of the Invention

[0005] To achieve the above objectives, this application provides the following technical solution:

[0006] A computer simulation and optimization method for directional adhesive bonding between the skin and flesh includes:

[0007] Obtain the three-dimensional model of the rubber sheet and the three-dimensional model of the meat block to be processed, and define material property parameters for the three-dimensional model of the rubber sheet and the three-dimensional model of the meat block respectively. The material property parameters include at least the elastic modulus and Poisson's ratio.

[0008] According to the preset bonding area, a virtual contact interface is constructed between the rubber three-dimensional model and the meat three-dimensional model, and the virtual contact interface is discretized into multiple contact units, each contact unit being associated with a rubber node on the rubber three-dimensional model and a meat node on the meat three-dimensional model;

[0009] Using the finite element analysis engine, multiple preset candidate pressure distribution patterns are sequentially applied to the outer surface of the rubber three-dimensional model. Based on the material property parameters, the deformation process of the rubber under each candidate pressure distribution pattern is simulated and calculated to obtain the corresponding rubber deformation sequence.

[0010] From the rubber deformation sequence, extract the relative displacement between the rubber node and the meat node at each contact unit under each candidate pressure distribution mode;

[0011] Based on the relative displacement, the overall fit coefficient corresponding to each candidate pressure distribution pattern is calculated. The overall fit coefficient is used to characterize the tightness of the fit between the rubber and the meat on the virtual contact interface.

[0012] By optimizing the engine to maximize the overall fit coefficient, the multiple candidate pressure distribution patterns are iteratively optimized until an optimized pressure distribution pattern that meets the preset convergence conditions is found. The optimized pressure distribution pattern is then determined as the target pressure application scheme to guide the directional bonding of the rubber to the meat.

[0013] Furthermore, using a finite element analysis engine, multiple preset candidate pressure distribution patterns are sequentially applied to the outer surface of the rubber 3D model. Based on the material property parameters, the deformation process of the rubber under each candidate pressure distribution pattern is simulated and calculated to obtain the corresponding rubber deformation sequence, including:

[0014] The three-dimensional model of the rubber is imported into the finite element analysis engine, and a material constitutive relation is assigned to the three-dimensional model of the rubber based on the material property parameters;

[0015] The three-dimensional model of the meat block is set as a rigid body with fixed constraints, and the contact properties between the three-dimensional model of the rubber and the three-dimensional model of the meat block are defined in the finite element analysis engine. The contact properties include the friction coefficient and the contact stiffness.

[0016] The nonlinear solver of the finite element analysis engine is invoked, and the current candidate pressure distribution pattern is used as a dynamic load in a time-step manner to be gradually applied to the specified stress area of ​​the rubber three-dimensional model.

[0017] After the calculation of each time step is completed, the spatial position coordinates of each node on the rubber 3D model at the current time step are recorded to form the rubber morphology data of the current frame.

[0018] Arrange the rubber morphology data corresponding to all time steps in chronological order to generate a rubber deformation sequence corresponding to the current candidate pressure distribution pattern.

[0019] Furthermore, before invoking the nonlinear solver of the finite element analysis engine and gradually applying the current candidate pressure distribution pattern as a dynamic load to the designated stress area of ​​the rubber 3D model in a time-step manner, the method also includes:

[0020] Multiple predefined pressure application points are obtained, each pressure application point corresponding to a specific node or region on the outer surface of the three-dimensional rubber model;

[0021] The current candidate pressure distribution pattern is parsed into a pressure loading table containing multiple time steps. The pressure loading table records the pressure value that needs to be applied at each pressure application point in each time step.

[0022] Before the nonlinear solver starts calculation, the pressure loading table is bound to the corresponding pressure application point. At the beginning of each time step, the nonlinear solver automatically reads the pressure value required for the current time step from the pressure loading table and applies it to the corresponding pressure application point.

[0023] Further, from the rubber deformation sequence, the relative displacement between the rubber node and the meat block node at each contact unit under each candidate pressure distribution mode is extracted, including:

[0024] For each candidate pressure distribution pattern, the final rubber morphology data corresponding to the last time step is selected from the rubber deformation sequence corresponding to that pattern.

[0025] Traverse each contact unit on the virtual contact interface to obtain the final spatial coordinates of the rubber node associated with the contact unit in the final rubber shape data, and the fixed spatial coordinates of the meat node associated with the contact unit.

[0026] Calculate the spatial vector difference between the final spatial coordinates and the fixed spatial coordinates, and determine the magnitude of the spatial vector difference as the absolute displacement at the contact unit;

[0027] Obtain the initial spatial coordinates of the rubber node at the contact unit in its initial undeformed state, and calculate the initial distance between the initial spatial coordinates and the fixed spatial coordinates;

[0028] Subtracting the initial distance from the absolute displacement yields the net relative displacement of the rubber node relative to the meat node at the contact unit. The net relative displacements of all contact units are then summed to form a set of displacements corresponding to the current candidate pressure distribution pattern.

[0029] Further, based on the relative displacement, the overall fit coefficient corresponding to each candidate pressure distribution pattern is calculated, including:

[0030] Obtain a preset ideal fit threshold range, which defines the allowable range of net relative displacement between the rubber node and the meat block node;

[0031] Iterate through each net relative displacement in the set of displacements corresponding to the current candidate pressure distribution pattern, and determine whether the net relative displacement falls within the ideal fitting threshold range.

[0032] The first number of net relative displacements falling within the ideal fit threshold range is counted, and the second number of all net relative displacements in the displacement set is counted.

[0033] The ratio of the first quantity to the second quantity is used as the overall fit coefficient of the current candidate pressure distribution pattern. The larger the overall fit coefficient, the higher the tightness of the fit.

[0034] Furthermore, by optimizing the engine to maximize the overall fit coefficient, the multiple candidate pressure distribution patterns are iteratively optimized, including:

[0035] The optimization engine employs an evolutionary strategy, encoding each candidate stress distribution pattern as an individual composed of stress values, and randomly generating an initial population containing multiple individuals.

[0036] For each individual in the initial population, the corresponding overall fit coefficient is calculated by the finite element analysis engine, and the overall fit coefficient is used as the fitness value of that individual.

[0037] Based on the fitness value, at least two individuals are selected from the current population as parents, and crossover and mutation operations are performed on the codes of the parents to generate new offspring individuals;

[0038] The offspring individuals are added to the population, and the fitness values ​​of all individuals in the population are recalculated to form a new generation population.

[0039] The process of selection, crossover, mutation, and recalculation is repeated until the maximum fitness value of individuals in the new generation population no longer increases or reaches the preset number of generations. The pressure distribution pattern represented by the individuals with the maximum fitness value is then used as the optimized pressure distribution pattern.

[0040] Furthermore, by optimizing the engine to maximize the overall fit coefficient, the multiple candidate pressure distribution patterns are iteratively optimized, including:

[0041] The optimization engine employs a Bayesian optimization strategy to construct a probabilistic proxy model that describes the mapping relationship between the pressure distribution pattern and the overall fit coefficient.

[0042] Initialize a set of observation points containing multiple candidate pressure distribution patterns, and use the finite element analysis engine to calculate the overall fit coefficient corresponding to each observation point;

[0043] Based on the set of observation points and their corresponding overall fit coefficients, the probabilistic surrogate model is updated so that it can predict the mean and variance of the fit coefficients for any unknown pressure distribution pattern.

[0044] Based on the prediction results of the probabilistic proxy model, a collection function is constructed, which is used to balance the exploration of unknown regions and the utilization of known high-fit regions;

[0045] By maximizing the acquisition function, the next most promising candidate pressure distribution pattern is selected from the design space of pressure distribution patterns as a new observation point.

[0046] The new observation point is added to the observation point set, and its corresponding overall fit coefficient is calculated again using the finite element analysis engine to update the probabilistic surrogate model.

[0047] Repeat the steps of selecting new observation points and updating the surrogate model until the preset number of iterations is reached or the overall fit coefficient found meets the preset requirements, and determine the candidate pressure distribution pattern with the highest overall fit coefficient in the set of observation points as the optimized pressure distribution pattern.

[0048] Furthermore, after determining the optimized pressure distribution pattern as the target pressure application scheme for guiding the directional adhesion of the rubber to the meat piece, the method further includes:

[0049] The target pressure application scheme and the three-dimensional model of the rubber are input into a pre-built deformation simulation module;

[0050] The deformation simulation module drives the rubber three-dimensional model to deform according to the pressure value and its point of application in the target pressure application scheme, generating a final rubber shape model that is closely attached to the surface of the meat block three-dimensional model.

[0051] Extract the inner surface of the final rubber morphology model that contacts the three-dimensional meat block model, and construct the mating surface mesh;

[0052] The mesh of the bonding surface is compared with the mesh of the outer surface of the three-dimensional model of the meat block, and the gap distance distribution between the two is calculated.

[0053] The gap distance distribution map is rendered as a heat map and overlaid on the three-dimensional model of the meat block to visually demonstrate the simulated fitting effect.

[0054] Furthermore, before obtaining the 3D models of the rubber sheet and meat to be processed, the following steps are also included:

[0055] Receive the initial rubber triangular mesh model and the initial meat triangular mesh model imported by the user through the graphical user interface;

[0056] The mesh quality inspection tool was used to calculate the aspect ratio and distortion of each triangle face in the initial rubber triangular mesh model and the initial meat triangular mesh model, respectively.

[0057] Triangular faces with an aspect ratio greater than the first preset threshold or a distortion greater than the second preset threshold are judged as unqualified meshes;

[0058] A local re-partitioning operation is performed on the region containing the defective mesh, and adjacent triangular faces are subdivided or merged to generate the rubber 3D model and the meat block 3D model with better mesh quality.

[0059] Furthermore, based on a preset bonding area, a virtual contact interface is constructed between the rubber 3D model and the meat block 3D model, and the virtual contact interface is discretized into multiple contact units, including:

[0060] From the outer surface of the three-dimensional model of the meat block, select a continuous curved surface region specified by the user as the preset fitting region;

[0061] All meat chunk nodes on the bonding area are marked, and a bonding area surface is generated based on the spatial position of these meat chunk nodes.

[0062] The perpendicular points of all nodes on the three-dimensional model of the rubber on the surface of the bonding area are calculated using a spatial projection algorithm.

[0063] If the distance from a rubber node to its perpendicular point is less than a preset contact distance threshold, then the rubber node is paired with the meat node to which its perpendicular point belongs to form a contact pair.

[0064] All contact pairs that meet the distance condition are aggregated, and each contact pair is defined as a contact unit. The collection of all contact units constitutes the virtual contact interface.

[0065] This invention discloses a computer simulation and optimization method for achieving tight bonding between rubber and meat, belonging to the field of computer-aided engineering and virtual simulation technology. The method involves constructing three-dimensional models of the rubber and meat components and defining material properties; building a virtual contact interface and discretizing it into multiple contact units; using a finite element analysis engine to simulate the deformation process of the rubber under different candidate pressure distribution modes, generating corresponding deformation sequences; extracting the relative displacement between rubber and meat nodes at each contact unit and calculating the overall bonding coefficient; iteratively optimizing the candidate pressure distribution modes until an optimized pressure distribution mode that meets the convergence condition is obtained, and outputting it as the target pressure application scheme. This invention can automatically find the optimal pressure application scheme that achieves the tightest bonding between the rubber and meat, effectively replacing the traditional process debugging method that relies on repeated trials based on experience, significantly improving bonding quality and efficiency, and providing precise computer guidance for the formulation of actual production processes. Attached Figure Description

[0066] Figure 1 A flowchart illustrating the computer simulation and optimization method for oriented rubber-flesh bonding tightness claimed in an embodiment of the present invention;

[0067] Figure 2 The second flowchart is a computer simulation and optimization method for oriented adhesive flesh bonding tightness claimed in an embodiment of the present invention.

[0068] Figure 3 The third flowchart is a computer simulation and optimization method for oriented adhesive flesh bonding tightness claimed in an embodiment of the present invention. Detailed Implementation

[0069] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0070] The terms "first," "second," and "third" in this application are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first," "second," or "third" may explicitly or implicitly include at least one of those features. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified. All directional indications in the embodiments of this application, such as up, down, left, right, front, back, etc., are only used to explain the relative positional relationships and movements between components in a specific orientation as shown in the accompanying drawings. If the specific orientation changes, the directional indications will change accordingly. Furthermore, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0071] References to embodiments herein mean that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0072] According to a first embodiment of the present invention, the present invention claims protection for a computer simulation and optimization method for oriented rubber-flesh adhesion tightness, referring to Figure 1 ,include:

[0073] Obtain the three-dimensional model of the rubber sheet and the three-dimensional model of the meat block to be processed, and define material property parameters for the three-dimensional model of the rubber sheet and the three-dimensional model of the meat block respectively. The material property parameters include at least the elastic modulus and Poisson's ratio.

[0074] According to the preset bonding area, a virtual contact interface is constructed between the rubber three-dimensional model and the meat three-dimensional model, and the virtual contact interface is discretized into multiple contact units, each contact unit being associated with a rubber node on the rubber three-dimensional model and a meat node on the meat three-dimensional model;

[0075] Using the finite element analysis engine, multiple preset candidate pressure distribution patterns are sequentially applied to the outer surface of the rubber three-dimensional model. Based on the material property parameters, the deformation process of the rubber under each candidate pressure distribution pattern is simulated and calculated to obtain the corresponding rubber deformation sequence.

[0076] From the rubber deformation sequence, extract the relative displacement between the rubber node and the meat node at each contact unit under each candidate pressure distribution mode;

[0077] Based on the relative displacement, the overall fit coefficient corresponding to each candidate pressure distribution pattern is calculated. The overall fit coefficient is used to characterize the tightness of the fit between the rubber and the meat on the virtual contact interface.

[0078] By optimizing the engine to maximize the overall fit coefficient, the multiple candidate pressure distribution patterns are iteratively optimized until an optimized pressure distribution pattern that meets the preset convergence conditions is found. The optimized pressure distribution pattern is then determined as the target pressure application scheme to guide the directional bonding of the rubber to the meat.

[0079] In this embodiment, the operator first starts a dedicated rubber-meat bonding simulation and optimization software on a computer. Through the software's graphical user interface, the operator imports pre-built 3D rubber model files and 3D meat model files from local storage. These two model files are usually in standard 3D mesh formats, such as STL or OBJ files. After importing, the software interface will simultaneously display 3D renderings of the two models. The operator can rotate and pan the viewpoint to confirm whether the model is correct.

[0080] Next, the operator enters the material property definition module. In this module, the operator needs to input material property parameters for the selected rubber 3D model in the pop-up property panel. For example, the operator enters a value for the elastic modulus, such as 5 MPa, and a value for Poisson's ratio, such as 0.45, in the corresponding input boxes. Similarly, define material properties for the meat block 3D model. Assuming the meat block is considered a relatively hard material in this simulation, an elastic modulus much larger than that of the rubber, such as 500 MPa, and a corresponding Poisson's ratio, such as 0.35, can be entered. These parameters will be used in subsequent finite element calculations.

[0081] Then, the operator begins constructing the virtual contact interface. In the software interface, using the mouse to select or click, they define an area on the outer surface of the 3D meat model. This area is the desired bonding area where the rubber sheet will ultimately adhere. Upon receiving this area information, the software automatically identifies all meat nodes within that area. Next, the software initiates a spatial search algorithm, traversing every node on the 3D rubber model and calculating the shortest distance from each rubber node to the selected meat area. When the software detects that the distance from a rubber node to its projection point on the meat area is less than a preset minimum threshold, such as 0.01 mm, it pairs the two nodes to form a contact unit. The system continues this process until all matching node pairs are found. The collection of all these contact units constitutes a virtual contact interface.

[0082] After completing the above preparations, the operator enters the main simulation and optimization process. The software has a built-in candidate pressure distribution pattern generator, which can generate multiple initial candidate schemes based on several preset basic patterns such as center point concentrated pressure, ring-shaped distributed pressure, and uniform pressure. Each scheme defines a set of pressure values ​​and specifies which external nodes of the rubber model these pressure values ​​will be applied to.

[0083] Next, the software's core processing engine begins its work. It extracts each candidate pressure distribution pattern one by one and inputs it as boundary conditions into the built-in finite element analysis engine. Taking the first candidate pattern as an example, the engine progressively loads the pressure value under that pattern onto the rubber model according to a specified time step. In each tiny loading step, the engine solves a complex set of mechanical equations based on the previously defined material properties of the rubber and the meat block, such as the elastic modulus and Poisson's ratio, to calculate the displacement and deformation of the rubber model under the current load. After the calculation is complete, the engine records the spatial coordinates of all nodes on the rubber at that moment. When all loading steps are completed, the engine obtains a sequence file consisting of rubber morphology data from multiple time steps.

[0084] For the rubber deformation sequence that has just been calculated, the software's data extraction module begins its work. It first locates the last time step of the sequence, that is, the state after the pressure has been fully applied. Then, it iterates through each previously constructed contact unit again. For each contact unit, it extracts the coordinates of the associated rubber node in its final state, as well as the original fixed coordinates of the associated meat node. By calculating the straight-line distance between these two coordinates in three-dimensional space, the software obtains the final distance between the rubber node and the meat node.

[0085] Finally, the optimization engine intervenes. Based on the final distances calculated from all contact elements, it evaluates the candidate pressure distribution pattern. The optimization engine aims to minimize these final distances, even to zero, for the tightest fit. After evaluating the first pattern, the optimization engine uses its built-in optimization logic to determine how to adjust the pressure values ​​to generate the next potentially better candidate pattern. It then calls the finite element analysis engine again for simulation, and this process iterates automatically multiple times. When the optimization engine finds that the evaluation metrics no longer show significant improvement after several iterations, or when it has reached the preset maximum number of iterations, it stops calculating and uses the best-performing pressure distribution pattern found historically as the final target pressure application scheme. This is output to the software interface along with a report containing detailed pressure values ​​and application locations.

[0086] Furthermore, using a finite element analysis engine, multiple preset candidate pressure distribution patterns are sequentially applied to the outer surface of the rubber 3D model. Based on the material property parameters, the deformation process of the rubber under each candidate pressure distribution pattern is simulated and calculated to obtain the corresponding rubber deformation sequence, including:

[0087] The three-dimensional model of the rubber is imported into the finite element analysis engine, and a material constitutive relation is assigned to the three-dimensional model of the rubber based on the material property parameters;

[0088] The three-dimensional model of the meat block is set as a rigid body with fixed constraints, and the contact properties between the three-dimensional model of the rubber and the three-dimensional model of the meat block are defined in the finite element analysis engine. The contact properties include the friction coefficient and the contact stiffness.

[0089] The nonlinear solver of the finite element analysis engine is invoked, and the current candidate pressure distribution pattern is used as a dynamic load in a time-step manner to be gradually applied to the specified stress area of ​​the rubber three-dimensional model.

[0090] After the calculation of each time step is completed, the spatial position coordinates of each node on the rubber 3D model at the current time step are recorded to form the rubber morphology data of the current frame.

[0091] Arrange the rubber morphology data corresponding to all time steps in chronological order to generate a rubber deformation sequence corresponding to the current candidate pressure distribution pattern.

[0092] Furthermore, before invoking the nonlinear solver of the finite element analysis engine and gradually applying the current candidate pressure distribution pattern as a dynamic load to the designated stress area of ​​the rubber 3D model in a time-step manner, the method also includes:

[0093] Multiple predefined pressure application points are obtained, each pressure application point corresponding to a specific node or region on the outer surface of the three-dimensional rubber model;

[0094] The current candidate pressure distribution pattern is parsed into a pressure loading table containing multiple time steps. The pressure loading table records the pressure value that needs to be applied at each pressure application point in each time step.

[0095] Before the nonlinear solver starts calculation, the pressure loading table is bound to the corresponding pressure application point. At the beginning of each time step, the nonlinear solver automatically reads the pressure value required for the current time step from the pressure loading table and applies it to the corresponding pressure application point.

[0096] In this embodiment, when the software begins processing a candidate pressure distribution pattern, the finite element analysis engine performs the following operations:

[0097] First, the engine creates a new computational task. It assigns predefined material properties to each mesh element of the imported rubber 3D model, establishing the relationship between stress and strain. At the same time, it sets the meat block 3D model as a rigid body, meaning that the meat block will not undergo any deformation during the calculation, and the degrees of freedom of all its nodes are locked. The engine then defines the interaction between the rubber and the meat block, sets the coefficient of friction of their surfaces to a constant, such as 0.3, and defines a contact stiffness to control the degree to which they can penetrate each other.

[0098] To apply pressure, the software first analyzes the candidate pattern, assuming it's named Pattern A. This pattern defines two pressure application points, located at the top center and edge of the rubber model, respectively, and specifies a pressure-time curve. The software discretizes this time curve, generating a pressure loading table. This table is a two-dimensional data structure; each row corresponds to a time step, and each column corresponds to a pressure application point. The values ​​in the table represent the pressure to be applied at that point at that time step. Subsequently, the software associates and binds this pressure loading table with the corresponding two pressure application point nodes on the rubber model.

[0099] Once preparation is complete, the engine activates its core nonlinear solver. The solver begins its cyclical calculation with extremely short time steps, such as 0.01 seconds. In the first time step, the solver queries the pressure loading table to obtain the pressure that needs to be applied to the two application points. It then applies these pressures as loads to the corresponding nodes. Next, based on these loads, the solver calculates the minute deformation of the rubber model and updates the positions of all nodes. After the calculation converges, the solver records the deformation result of this first step and then proceeds to the second time step. It queries the pressure loading table again, and the pressure value may have increased. It applies the updated load, continues to calculate the deformation, and records the results. This process is repeated until the preset total duration, such as 1 second, is reached, meaning all time steps specified by the pressure loading table have been executed. Finally, the solver outputs a deformation sequence file containing 100 rubber morphology data at different times with an assumed time step of 0.01 seconds and a total duration of 1 second.

[0100] Furthermore, referring to Figure 2 From the rubber deformation sequence, the relative displacement between the rubber node and the meat block node at each contact unit under each candidate pressure distribution mode is extracted, including:

[0101] For each candidate pressure distribution pattern, the final rubber morphology data corresponding to the last time step is selected from the rubber deformation sequence corresponding to that pattern.

[0102] Traverse each contact unit on the virtual contact interface to obtain the final spatial coordinates of the rubber node associated with the contact unit in the final rubber shape data, and the fixed spatial coordinates of the meat node associated with the contact unit.

[0103] Calculate the spatial vector difference between the final spatial coordinates and the fixed spatial coordinates, and determine the magnitude of the spatial vector difference as the absolute displacement at the contact unit;

[0104] Obtain the initial spatial coordinates of the rubber node at the contact unit in its initial undeformed state, and calculate the initial distance between the initial spatial coordinates and the fixed spatial coordinates;

[0105] Subtracting the initial distance from the absolute displacement yields the net relative displacement of the rubber node relative to the meat node at the contact unit. The net relative displacements of all contact units are then summed to form a set of displacements corresponding to the current candidate pressure distribution pattern.

[0106] Furthermore, referring to Figure 3 Based on the relative displacement, the overall fit coefficient corresponding to each candidate pressure distribution pattern is calculated, including:

[0107] Obtain a preset ideal fit threshold range, which defines the allowable range of net relative displacement between the rubber node and the meat block node;

[0108] Iterate through each net relative displacement in the set of displacements corresponding to the current candidate pressure distribution pattern, and determine whether the net relative displacement falls within the ideal fitting threshold range.

[0109] The first number of net relative displacements falling within the ideal fit threshold range is counted, and the second number of all net relative displacements in the displacement set is counted.

[0110] The ratio of the first quantity to the second quantity is used as the overall fit coefficient of the current candidate pressure distribution pattern. The larger the overall fit coefficient, the higher the tightness of the fit.

[0111] In this embodiment, after the finite element analysis engine completes the simulation of a candidate pressure distribution pattern, such as pattern A, it generates a rubber deformation sequence file containing multiple time steps. The software's post-processing module is then activated and begins executing the fit evaluation task.

[0112] The post-processing module first opens the sequence file and directly navigates to the last file in the sequence, which is the final form of the rubber after the simulation ends and pressure is fully applied. Then, it retrieves the list of contact elements built during the preprocessing stage. This list contains tens of thousands of contact elements, each of which records the IDs of a pair of rubber nodes and meat nodes.

[0113] The post-processing module begins to traverse this list. For the first contact unit, it extracts the final coordinates x1_end, y1_end, and z1_end of the rubber node P1 from the final shape data based on the recorded ID, and extracts the coordinates x1_m, y1_m, and z1_m of the meat node M1 from the fixed data of the meat. The software calculates the Euclidean distance between the two points and obtains D_end = 0.15 mm.

[0114] At the same time, the software also recorded the initial coordinates x1_start, y1_start, and z1_start of the same rubber node P1 before the simulation started and before any deformation of the rubber had occurred. The software calculated that the distance from the initial position of P1 to M1 was D_start = 2.00 mm.

[0115] To measure the net effect of the bonding, the software calculates the net relative displacement D_net = D_end - D_start. In this example, D_net = 0.15 - 2.00 = -1.85 mm. This negative value indicates that the rubber node P1 has moved 1.85 mm towards the meat node M1, which is very close to the target. The software stores this net displacement in the result field corresponding to the contact element.

[0116] After traversing all contact units, the software obtains a large dataset of net displacement. Next, the software reads a preset ideal fit threshold range. This range can be preset by the operator, for example, [-0.01 mm, +0.05 mm]. This means that as long as the final position of the rubber node is closer to the meat node than the initial state (negative value) or slightly further than the initial state (positive value), as long as it falls within this small range, the fit is considered acceptable.

[0117] The software iterates through all contact elements again, retrieves each previously calculated net displacement D_net, and determines whether it falls within the interval [-0.01, +0.05]. Internally, the software maintains two counters: a pass / fail counter A and a total counter B. Each time a contact element is traversed, the total counter B is incremented by 1. If the current element's D_net falls within the interval, the pass / fail counter A is also incremented by 1.

[0118] After the traversal is complete, the software calculates the overall fit coefficient S=A / B. Assuming there are a total of 10,000 contact units, and the net displacement of 9,500 of them is within the ideal threshold range, then the overall fit coefficient S is 0.95. The higher this value, the tighter and more uniform the rubber and meat are bonded under this pressure distribution mode. The software uses this coefficient S as the final score of mode A and returns it to the optimization engine to guide the next round of solution search.

[0119] Furthermore, by optimizing the engine to maximize the overall fit coefficient, the multiple candidate pressure distribution patterns are iteratively optimized, including:

[0120] The optimization engine employs an evolutionary strategy, encoding each candidate stress distribution pattern as an individual composed of stress values, and randomly generating an initial population containing multiple individuals.

[0121] For each individual in the initial population, the corresponding overall fit coefficient is calculated by the finite element analysis engine, and the overall fit coefficient is used as the fitness value of that individual.

[0122] Based on the fitness value, at least two individuals are selected from the current population as parents, and crossover and mutation operations are performed on the codes of the parents to generate new offspring individuals;

[0123] The offspring individuals are added to the population, and the fitness values ​​of all individuals in the population are recalculated to form a new generation population.

[0124] The process of selection, crossover, mutation, and recalculation is repeated until the maximum fitness value of individuals in the new generation population no longer increases or reaches the preset number of generations. The pressure distribution pattern represented by the individuals with the maximum fitness value is then used as the optimized pressure distribution pattern.

[0125] In this embodiment, after the optimization engine is started, the decision variable to be optimized—that is, the pressure distribution pattern—is first encoded. Each pressure distribution pattern is represented as a one-dimensional array, and each element in the array represents the pressure value at a specific pressure point. For example, if 10 pressure points are defined, then each candidate pattern is an individual containing 10 pressure values.

[0126] The optimization engine then generates an initial population, which generates 50 individuals in a random manner, that is, 50 different combinations of pressure values, each of which is randomly selected within a reasonable physical range, such as 0 Newtons to 100 Newtons.

[0127] Next, the optimization engine submits each of the 50 individuals to the finite element analysis engine for simulation calculations, as described in Examples 1 and 2. After the calculations are completed, the overall fit coefficient corresponding to each individual is obtained from the evaluation module described in Example 3. This coefficient is regarded as the fitness value of that individual. The optimization engine records each individual and its fitness value.

[0128] After evaluating all individuals, the optimization engine enters the evolutionary cycle. It first performs a selection operation based on fitness values, using a roulette wheel selection method, where individuals with higher fitness values ​​have a greater probability of being selected. In this way, it selects two individuals from the current 50 individuals as parents, for example, individual A with a fitness of 0.8 and individual B with a fitness of 0.75.

[0129] Then, the optimization engine performs a crossover operation on the selected parent. It randomly selects a crossover point with a certain probability, for example, 0.9, and concatenates the part of individual A before the crossover point with the part of individual B after the crossover point to generate a new offspring individual C; for example, the stress array of individual A is [10, 20, 30, 40], and that of individual B is [15, 25, 35, 45]. If the crossover point is selected after the second element, then the array of offspring C will be [10, 20, 35, 45].

[0130] After generating offspring C, the optimization engine performs a mutation operation. It randomly selects an element from the offspring C array with a very small probability, such as 0.01, and adds a tiny random perturbation to it, such as changing the third element 35 to 36, thus obtaining a completely new individual.

[0131] The optimization engine repeats the selection, crossover, and mutation process until the number of new offspring individuals reaches 50, which is the same as the size of the parent population, thus forming a new generation population.

[0132] Next, the optimization engine calls upon the finite element analysis engine and evaluation module again to calculate the fitness values ​​of these 50 newly generated individuals. At this point, it possesses complete evaluation data for the new generation of the population.

[0133] The optimization engine compares the maximum fitness value in the current generation of the population with the maximum fitness value in the previous generation. If this maximum value has not increased for 10 consecutive generations, or if the total number of generations has reached the preset 100 generations, the optimization engine stops iterating. It then identifies the individual with the highest historical fitness value from all generations of the population, decodes it into a set of stress values, and outputs this as the final optimization stress distribution pattern.

[0134] Furthermore, by optimizing the engine to maximize the overall fit coefficient, the multiple candidate pressure distribution patterns are iteratively optimized, including:

[0135] The optimization engine employs a Bayesian optimization strategy to construct a probabilistic proxy model that describes the mapping relationship between the pressure distribution pattern and the overall fit coefficient.

[0136] Initialize a set of observation points containing multiple candidate pressure distribution patterns, and use the finite element analysis engine to calculate the overall fit coefficient corresponding to each observation point;

[0137] Based on the set of observation points and their corresponding overall fit coefficients, the probabilistic surrogate model is updated so that it can predict the mean and variance of the fit coefficients for any unknown pressure distribution pattern.

[0138] Based on the prediction results of the probabilistic proxy model, a collection function is constructed, which is used to balance the exploration of unknown regions and the utilization of known high-fit regions;

[0139] By maximizing the acquisition function, the next most promising candidate pressure distribution pattern is selected from the design space of pressure distribution patterns as a new observation point.

[0140] The new observation point is added to the observation point set, and its corresponding overall fit coefficient is calculated again using the finite element analysis engine to update the probabilistic surrogate model.

[0141] Repeat the steps of selecting new observation points and updating the surrogate model until the preset number of iterations is reached or the overall fit coefficient found meets the preset requirements, and determine the candidate pressure distribution pattern with the highest overall fit coefficient in the set of observation points as the optimized pressure distribution pattern.

[0142] In this embodiment, the optimization engine first initializes a Gaussian process as a probabilistic surrogate model, which is used to predict the average value and uncertainty variance of the overall fit coefficient that any unknown pressure distribution pattern may achieve.

[0143] Next, the optimization engine needs to train this surrogate model using some initial observation points. It uses the Latin hypercube sampling method to uniformly select 10 points, i.e. 10 candidate pressure distribution patterns, from the design space of pressure values. Then, these 10 patterns are submitted to the finite element analysis engine for simulation, and their corresponding overall fit coefficients are obtained.

[0144] With these 10 observation point pressure patterns and their corresponding fit coefficients, the optimization engine updates the Gaussian process model. The updated model can now predict any untried pressure pattern in space, providing a mean of the predicted fit coefficient and a variance representing the uncertainty of the prediction.

[0145] Then, the optimization engine builds a sampling function, such as the expected improvement function, which takes into account that the predicted mean of a certain point may be high and the prediction variance uncertainty is large, which may lead to surprises, and calculates a score worth trying for each point in the design space.

[0146] The optimization engine then searches the design space of stress values, aiming to find the point that maximizes the acquisition function value. It uses an internal genetic algorithm to maximize the acquisition function. After the search, it finds a new candidate stress distribution pattern, which the current surrogate model considers to be the point with the greatest potential to improve the fit coefficient.

[0147] The optimization engine takes this newly found pattern as a new observation point and submits it to the finite element analysis engine for simulation to obtain its true fit coefficient. Then, it adds this new real data point to the existing 10 observation points, expanding the observation point set to 11, and updates the Gaussian process surrogate model again.

[0148] The optimization engine repeats the above loop of finding the next most promising point by maximizing the acquisition function -> performing realistic simulation evaluation -> updating the surrogate model. After each loop, it checks whether the highest observed true fit coefficient has reached the preset target value, or whether the preset total number of iterations, such as 40, has been reached. If either stopping condition is met, the optimization engine stops the loop and determines the pattern with the highest true fit coefficient among all historically observed pressure distribution patterns as the final optimized pressure distribution pattern.

[0149] Furthermore, after determining the optimized pressure distribution pattern as the target pressure application scheme for guiding the directional adhesion of the rubber to the meat piece, the method further includes:

[0150] The target pressure application scheme and the three-dimensional model of the rubber are input into a pre-built deformation simulation module;

[0151] The deformation simulation module drives the rubber three-dimensional model to deform according to the pressure value and its point of application in the target pressure application scheme, generating a final rubber shape model that is closely attached to the surface of the meat block three-dimensional model.

[0152] Extract the inner surface of the final rubber morphology model that contacts the three-dimensional meat block model, and construct the mating surface mesh;

[0153] The mesh of the bonding surface is compared with the mesh of the outer surface of the three-dimensional model of the meat block, and the gap distance distribution between the two is calculated.

[0154] The gap distance distribution map is rendered as a heat map and overlaid on the three-dimensional model of the meat block to visually demonstrate the simulated fitting effect.

[0155] In this embodiment, after the optimization engine determines and outputs the target pressure application scheme, the software enters the result verification and visualization stage.

[0156] The software first sends the target pressure scheme, such as a set of pressure values ​​and their corresponding coordinates of the application point, along with the original 3D model of the rubber, into a dedicated deformation simulation module. This module is similar to the previous finite element analysis engine, but it performs a single, high-precision final simulation. Based on the target pressure scheme, it accurately calculates the final shape of the rubber under ideal pressure and generates a final rubber shape model that fits closely to the surface of the meat.

[0157] Next, the software extracts the inner surface of this final rubber-shaped model. This is typically achieved by selecting all mesh surfaces whose normal directions roughly point inwards from the model. The software then copies these inner surface meshes to create a separate mating surface mesh.

[0158] Then, the software launches the geometric comparison tool. It takes each node on the mating mesh and calculates the shortest distance from that node to the outer surface of the original meat block 3D model. This calculation process is equivalent to firing a ray from each point on the mating surface to the meat block surface, finding the intersection point with the meat block surface and measuring the distance. The software records this distance as the gap value of that node.

[0159] After traversing all nodes on the bonding surface, the software obtains a dataset consisting of tens of thousands of gap values. These values ​​can be positive or negative. Positive values ​​indicate that there is still a gap between the inner surface of the rubber and the surface of the meat block, while negative values ​​indicate that the rubber has penetrated the surface of the meat block, which is physically impossible but may occur during geometric comparison and needs to be filtered or corrected according to the contact definition.

[0160] Finally, the visualization module is activated. It uses color mapping technology to map these gap values ​​onto a color spectrum from blue to red. For example, the largest gap value, such as 0.5 mm, is mapped to red; the smallest gap value, such as -0.1 mm, representing slight penetration, is mapped to dark blue; and intermediate values ​​are green and yellow. The software displays the original 3D model of the meat block and renders the mesh of the bonding surface according to the gap values ​​at its nodes. This rendering is then overlaid on the meat block model in a semi-transparent manner, or the colors are directly projected onto the surface of the meat block model. The operator ultimately sees a meat block model with a color heatmap on the software interface, visually demonstrating which areas between the rubber and the meat block are tightly bonded (green areas) and which areas still have gaps (red areas), thus verifying the effectiveness of the optimization scheme.

[0161] Furthermore, before obtaining the 3D models of the rubber sheet and meat to be processed, the following steps are also included:

[0162] Receive the initial rubber triangular mesh model and the initial meat triangular mesh model imported by the user through the graphical user interface;

[0163] The mesh quality inspection tool was used to calculate the aspect ratio and distortion of each triangle face in the initial rubber triangular mesh model and the initial meat triangular mesh model, respectively.

[0164] Triangular faces with an aspect ratio greater than the first preset threshold or a distortion greater than the second preset threshold are judged as unqualified meshes;

[0165] A local re-partitioning operation is performed on the region containing the defective mesh, and adjacent triangular faces are subdivided or merged to generate the rubber 3D model and the meat block 3D model with better mesh quality.

[0166] In this embodiment, after importing the model and before defining material properties, the operator can choose to perform a mesh optimization step. When the operator clicks the "Check Mesh Quality" button, the software starts to process automatically.

[0167] First, the software receives the initial rubber triangular mesh model and the initial meat triangular mesh model imported by the operator through the graphical user interface. These two models may come from 3D scanning or CAD design, and the mesh quality varies.

[0168] Next, the software invokes the built-in mesh quality inspection tool. This tool iterates through each triangular facet in the rubber mesh model. For each facet, it extracts the coordinates of the three vertices, calculates the lengths of the three sides of the triangle, and then calculates the aspect ratio of the triangle, that is, the ratio of the longest side length to the shortest side length. At the same time, it calculates the twist of the triangle based on the angular deviation between the normal vector of the plane containing the triangle and the ideal normal vector.

[0169] The software compares the calculated aspect ratio with a first preset threshold, for example, a threshold of 10. If the aspect ratio of a triangle is greater than 10, the triangle is judged as an unqualified mesh. Simultaneously, the software compares the calculated distortion with a second preset threshold, for example, a threshold of 30 degrees. If the distortion of a triangle is greater than 30 degrees, the triangle is also judged as an unqualified mesh.

[0170] The inspection tool marks each triangle that is deemed unqualified. After traversing all triangles, the software highlights the areas where these unqualified grids are located for operator confirmation.

[0171] If the operator agrees to optimization, the software initiates a local re-mesh operation. It first identifies the adjacent regions of all defective meshes. For elongated triangles with excessively large aspect ratios, the software inserts a new node on its longest side, subdividing the triangle with its neighboring triangles to generate new, more symmetrical triangles. For triangles with excessive distortion, the software merges several adjacent triangles within the region and then regenerates a new set of more regularly arranged triangles based on the average position of surrounding nodes. This process iterates several times until all newly generated triangles within the region pass the quality check. After processing all defective regions, the software generates 3D models of the rubber and meat with improved mesh quality, suitable for subsequent finite element analysis.

[0172] Furthermore, based on a preset bonding area, a virtual contact interface is constructed between the rubber 3D model and the meat block 3D model, and the virtual contact interface is discretized into multiple contact units, including:

[0173] From the outer surface of the three-dimensional model of the meat block, select a continuous curved surface region specified by the user as the preset fitting region;

[0174] All meat chunk nodes on the bonding area are marked, and a bonding area surface is generated based on the spatial position of these meat chunk nodes.

[0175] The perpendicular points of all nodes on the three-dimensional model of the rubber on the surface of the bonding area are calculated using a spatial projection algorithm.

[0176] If the distance from a rubber node to its perpendicular point is less than a preset contact distance threshold, then the rubber node is paired with the meat node to which its perpendicular point belongs to form a contact pair.

[0177] All contact pairs that meet the distance condition are aggregated, and each contact pair is defined as a contact unit. The collection of all contact units constitutes the virtual contact interface.

[0178] In this embodiment, before the simulation, the operator draws on the rendered view of the 3D meat block model using the mouse on the software interface. For example, the operator can use the brush tool to draw a closed curve on a specific surface at the top of the model. The area inside this curve is designated as the preset fitting area. After receiving this area information, the software marks all the triangular faces corresponding to the area and extracts all the nodes on these triangular faces to form a subset of meat block nodes. Subsequently, the software uses the spatial coordinates of these nodes to generate a smooth spatial surface representing the fitting area through a surface fitting algorithm.

[0179] Next, the software begins to find the corresponding point on the surface of the mating area for each node on the rubber model. It initiates a spatial projection algorithm. For a node P on the rubber, the algorithm calculates the perpendicular line drawn from point P along the normal direction or the shortest distance direction of the surface. The intersection of the perpendicular line and the surface is the foot point Q of the perpendicular.

[0180] The algorithm calculates the spatial distance d from point P to the perpendicular point Q. The software compares this distance d with a preset contact distance threshold, which can be set by the operator based on factors such as the initial thickness of the rubber, for example, 0.5 mm.

[0181] If the distance d is less than or equal to 0.5 mm, the software considers the rubber node P to be very close to the meat block surface before the simulation starts, belonging to the initial contact area. The software then needs to find the meat block node to which the perpendicular point Q belongs. Since point Q is usually located inside a meat block triangle rather than exactly on a node, the software will determine a virtual meat block node by calculating the weights of the three vertices of the triangle. Alternatively, the software can directly select the vertex of the meat block triangle closest to point Q as the meat block node M paired with P.

[0182] Once the software confirms the pairing relationship, it will generate a new data structure, namely a contact pair P,M. The software will store this contact pair and assign it a unique ID.

[0183] The software performs the above projection, distance judgment and pairing operations on each node on the rubber model. After all nodes have been traversed, the software summarizes all successfully paired contact pairs. These thousands of contact pairs constitute a virtual contact interface composed of discrete point pairs that covers the entire predetermined bonding area. This interface will serve as the basis for all subsequent bonding evaluations.

[0184] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0185] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated units described above can be implemented in hardware or as software functional units. The above are merely embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made based on the description and drawings of this application, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

[0186] The specific embodiments of the invention have been described in detail above, but they are only examples, and this application is not limited to the specific embodiments described above. For those skilled in the art, any equivalent modifications or substitutions to the invention are also within the scope of this application. Therefore, all equivalent changes, modifications, and improvements made without departing from the spirit and principles of this application should be covered within the scope of this application.

Claims

1. A computer simulation and optimization method for achieving tight adhesion between oriented rubber and flesh, characterized in that, include: Obtain the three-dimensional model of the rubber sheet and the three-dimensional model of the meat block to be processed, and define material property parameters for the three-dimensional model of the rubber sheet and the three-dimensional model of the meat block respectively. The material property parameters include at least the elastic modulus and Poisson's ratio. According to the preset bonding area, a virtual contact interface is constructed between the rubber three-dimensional model and the meat three-dimensional model, and the virtual contact interface is discretized into multiple contact units, each contact unit being associated with a rubber node on the rubber three-dimensional model and a meat node on the meat three-dimensional model; Using the finite element analysis engine, multiple preset candidate pressure distribution patterns are sequentially applied to the outer surface of the rubber three-dimensional model. Based on the material property parameters, the deformation process of the rubber under each candidate pressure distribution pattern is simulated and calculated to obtain the corresponding rubber deformation sequence. From the rubber deformation sequence, extract the relative displacement between the rubber node and the meat node at each contact unit under each candidate pressure distribution mode; Based on the relative displacement, the overall fit coefficient corresponding to each candidate pressure distribution pattern is calculated. The overall fit coefficient is used to characterize the tightness of the fit between the rubber and the meat on the virtual contact interface. By optimizing the engine to maximize the overall fit coefficient, the multiple candidate pressure distribution patterns are iteratively optimized until an optimized pressure distribution pattern that meets the preset convergence conditions is found. The optimized pressure distribution pattern is then determined as the target pressure application scheme to guide the directional bonding of the rubber to the meat.

2. The method according to claim 1, characterized in that, The process involves using a finite element analysis engine to sequentially apply multiple preset candidate pressure distribution patterns to the outer surface of the rubber 3D model. Based on the material property parameters, the deformation process of the rubber under each candidate pressure distribution pattern is simulated and calculated to obtain the corresponding rubber deformation sequence, including: The three-dimensional model of the rubber is imported into the finite element analysis engine, and a material constitutive relation is assigned to the three-dimensional model of the rubber based on the material property parameters; The three-dimensional model of the meat block is set as a rigid body with fixed constraints, and the contact properties between the three-dimensional model of the rubber and the three-dimensional model of the meat block are defined in the finite element analysis engine. The contact properties include the friction coefficient and the contact stiffness. The nonlinear solver of the finite element analysis engine is invoked, and the current candidate pressure distribution pattern is used as a dynamic load in a time-step manner to be gradually applied to the specified stress area of ​​the rubber three-dimensional model. After the calculation of each time step is completed, the spatial position coordinates of each node on the rubber 3D model at the current time step are recorded to form the rubber morphology data of the current frame. Arrange the rubber morphology data corresponding to all time steps in chronological order to generate a rubber deformation sequence corresponding to the current candidate pressure distribution pattern.

3. The method according to claim 2, characterized in that, Before invoking the nonlinear solver of the finite element analysis engine to progressively apply the current candidate pressure distribution pattern as a dynamic load to the designated stress area of ​​the rubber 3D model in a time-step manner, the method further includes: Multiple predefined pressure application points are obtained, each pressure application point corresponding to a specific node or region on the outer surface of the three-dimensional rubber model; The current candidate pressure distribution pattern is parsed into a pressure loading table containing multiple time steps. The pressure loading table records the pressure value that needs to be applied at each pressure application point in each time step. Before the nonlinear solver starts calculation, the pressure loading table is bound to the corresponding pressure application point. At the beginning of each time step, the nonlinear solver automatically reads the pressure value required for the current time step from the pressure loading table and applies it to the corresponding pressure application point.

4. The method according to claim 1, characterized in that, The step of extracting the relative displacement between the rubber node and the meat node at each contact unit under each candidate pressure distribution pattern from the rubber deformation sequence includes: For each candidate pressure distribution pattern, the final rubber morphology data corresponding to the last time step is selected from the rubber deformation sequence corresponding to that pattern. Traverse each contact unit on the virtual contact interface to obtain the final spatial coordinates of the rubber node associated with the contact unit in the final rubber shape data, and the fixed spatial coordinates of the meat node associated with the contact unit. Calculate the spatial vector difference between the final spatial coordinates and the fixed spatial coordinates, and determine the magnitude of the spatial vector difference as the absolute displacement at the contact unit; Obtain the initial spatial coordinates of the rubber node at the contact unit in its initial undeformed state, and calculate the initial distance between the initial spatial coordinates and the fixed spatial coordinates; Subtracting the initial distance from the absolute displacement yields the net relative displacement of the rubber node relative to the meat node at the contact unit. The net relative displacements of all contact units are then summed to form a set of displacements corresponding to the current candidate pressure distribution pattern.

5. The method according to claim 4, characterized in that, The step of calculating the overall fit coefficient corresponding to each candidate pressure distribution pattern based on the relative displacement includes: Obtain a preset ideal fit threshold range, which defines the allowable range of net relative displacement between the rubber node and the meat block node; Iterate through each net relative displacement in the set of displacements corresponding to the current candidate pressure distribution pattern, and determine whether the net relative displacement falls within the ideal fitting threshold range. The first number of net relative displacements falling within the ideal fit threshold range is counted, and the second number of all net relative displacements in the displacement set is counted. The ratio of the first quantity to the second quantity is used as the overall fit coefficient of the current candidate pressure distribution pattern. The larger the overall fit coefficient, the higher the tightness of the fit.

6. The method according to claim 1, characterized in that, The step of iteratively optimizing the multiple candidate pressure distribution patterns through an optimization engine, with the goal of maximizing the overall fit coefficient, includes: The optimization engine employs an evolutionary strategy, encoding each candidate stress distribution pattern as an individual composed of stress values, and randomly generating an initial population containing multiple individuals. For each individual in the initial population, the corresponding overall fit coefficient is calculated by the finite element analysis engine, and the overall fit coefficient is used as the fitness value of that individual. Based on the fitness value, at least two individuals are selected from the current population as parents, and crossover and mutation operations are performed on the codes of the parents to generate new offspring individuals; The offspring individuals are added to the population, and the fitness values ​​of all individuals in the population are recalculated to form a new generation population. The process of selection, crossover, mutation, and recalculation is repeated until the maximum fitness value of individuals in the new generation population no longer increases or reaches the preset number of generations. The pressure distribution pattern represented by the individuals with the maximum fitness value is then used as the optimized pressure distribution pattern.

7. The method according to claim 1, characterized in that, The step of iteratively optimizing the multiple candidate pressure distribution patterns through an optimization engine, with the goal of maximizing the overall fit coefficient, includes: The optimization engine employs a Bayesian optimization strategy to construct a probabilistic proxy model that describes the mapping relationship between the pressure distribution pattern and the overall fit coefficient. Initialize a set of observation points containing multiple candidate pressure distribution patterns, and use the finite element analysis engine to calculate the overall fit coefficient corresponding to each observation point; Based on the set of observation points and their corresponding overall fit coefficients, the probabilistic surrogate model is updated so that it can predict the mean and variance of the fit coefficients for any unknown pressure distribution pattern. Based on the prediction results of the probabilistic proxy model, a collection function is constructed, which is used to balance the exploration of unknown regions and the utilization of known high-fit regions; By maximizing the acquisition function, the next most promising candidate pressure distribution pattern is selected from the design space of pressure distribution patterns as a new observation point. The new observation point is added to the observation point set, and its corresponding overall fit coefficient is calculated again using the finite element analysis engine to update the probabilistic surrogate model. Repeat the steps of selecting new observation points and updating the surrogate model until the preset number of iterations is reached or the overall fit coefficient found meets the preset requirements, and determine the candidate pressure distribution pattern with the highest overall fit coefficient in the set of observation points as the optimized pressure distribution pattern.

8. The method according to claim 1, characterized in that, After determining the optimized pressure distribution pattern as the target pressure application scheme for guiding the directional bonding of the rubber to the meat piece, the method further includes: The target pressure application scheme and the three-dimensional model of the rubber are input into a pre-built deformation simulation module; The deformation simulation module drives the rubber three-dimensional model to deform according to the pressure value and its point of application in the target pressure application scheme, generating a final rubber shape model that is closely attached to the surface of the meat block three-dimensional model. Extract the inner surface of the final rubber morphology model that contacts the three-dimensional meat block model, and construct the mating surface mesh; The mesh of the bonding surface is compared with the mesh of the outer surface of the three-dimensional model of the meat block, and the gap distance distribution between the two is calculated. The gap distance distribution map is rendered as a heat map and overlaid on the three-dimensional model of the meat block to visually demonstrate the simulated fitting effect.

9. The method according to claim 1, characterized in that, Before obtaining the 3D model of the rubber sheet and the 3D model of the meat block to be processed, the process also includes: Receive the initial rubber triangular mesh model and the initial meat triangular mesh model imported by the user through the graphical user interface; The mesh quality inspection tool was used to calculate the aspect ratio and distortion of each triangle face in the initial rubber triangular mesh model and the initial meat triangular mesh model, respectively. Triangular faces with an aspect ratio greater than the first preset threshold or a distortion greater than the second preset threshold are judged as unqualified meshes; A local re-partitioning operation is performed on the region containing the defective mesh, and adjacent triangular faces are subdivided or merged to generate the rubber 3D model and the meat block 3D model with better mesh quality.

10. The method according to claim 1, characterized in that, The virtual contact interface is constructed between the rubber 3D model and the meat block 3D model according to a preset bonding area, and the virtual contact interface is discretized into multiple contact units, including: From the outer surface of the three-dimensional model of the meat block, select a continuous curved surface region specified by the user as the preset fitting region; All meat chunk nodes on the bonding area are marked, and a bonding area surface is generated based on the spatial position of these meat chunk nodes. The perpendicular points of all nodes on the three-dimensional model of the rubber on the surface of the bonding area are calculated using a spatial projection algorithm. If the distance from a rubber node to its perpendicular point is less than a preset contact distance threshold, then the rubber node is paired with the meat node to which its perpendicular point belongs to form a contact pair. All contact pairs that meet the distance condition are aggregated, and each contact pair is defined as a contact unit. The collection of all contact units constitutes the virtual contact interface.

Images