A parameter optimization design method of a fresh food delivery gradient composite heat preservation bag and a heat preservation bag

Abstract

The application relates to the technical field of heat preservation packaging, and discloses a parameter optimization design method of a gradient composite heat preservation bag for fresh food takeout and the heat preservation bag. The method generates a dynamic convection boundary matrix by taking the vehicle driving speed and the environmental temperature as inputs; a two-dimensional tensile strain field is solved by combining with a load parameter, and a thermal resistance and heat capacity reduction matrix is output; a transient heat conduction control equation is established to perform forward numerical integration to solve a space temperature field; a Lagrange functional is constructed based on a cold preservation target and material consumption, and reverse integration is performed to obtain a concomitant temperature field; a sensitivity gradient is extracted to iteratively update a space design variable field until convergence, and finally, discrete manufacturing parameters are mapped and output. The heat preservation bag designed in this way has an asymmetric thickness distribution of a windward surface and a leeward surface and a stress deformation compensation step structure of a bottom surface and a side wall. The application precisely matches the dynamic convection working condition of takeout, prevents local heat preservation performance attenuation caused by bearing deformation, and realizes the lightweight and easy manufacturing of the heat preservation bag.

Description

Technical Field

[0001] This invention relates to the field of thermal packaging technology, specifically to a parameter optimization design method for a gradient composite thermal insulation bag for fresh food delivery and the thermal insulation bag itself. Background Technology

[0002] Temperature control during last-mile delivery of fresh food plays a crucial role in ensuring food safety and quality. Composite insulated bags are currently a widely used, low-cost cold-keeping packaging solution. Current fresh food delivery insulated bags are typically made of foamed material of uniform thickness, and their structural parameters are often designed based on empirical formulas under static conditions or steady-state thermal conductivity tests.

[0003] However, the actual fresh food delivery process faces complex external environments and internal loads, and existing design solutions have several shortcomings. The transient wind speeds generated by delivery vehicles create complex flow fields on the outside of the insulated bag, resulting in significant differences in convective heat transfer coefficients between the windward and leeward sides. Existing uniform thickness designs cannot match this dynamic and asymmetric convective heat transfer boundary, leading to severe heat loss in high-wind-speed areas and material redundancy in low-wind-speed areas. Furthermore, existing insulated bags fail to consider the coupled effects of structural mechanics and heat transfer in their design.

[0004] Once the food container is placed in the insulated bag, its weight and geometric shape exert continuous normal pressure and uniformly distributed gravitational load on the bottom and sidewalls, causing two-dimensional tensile deformation of the material in the load-bearing area. This stress-induced deformation directly leads to localized material thinning and internal foam pore compression, resulting in a decrease in the equivalent thermal resistance and specific heat capacity of that area, making it a weak point for localized heat leakage in the insulated bag. Furthermore, relying solely on increasing the overall wall thickness to compensate for the lack of localized insulation performance not only increases the total material consumption and manufacturing cost, but existing design processes also lack topology optimization algorithms that can combine environmental dynamic variables, material deformation attenuation, and spatially non-uniform variable fields. Moreover, they cannot convert continuous optimization calculation results into discrete manufacturing parameters suitable for actual mold-making and foaming processes. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a parameter optimization design method for a gradient composite insulated bag for fresh food delivery and an insulated bag itself. This solves the problems that the uniform thickness design of existing insulated bags cannot match the dynamic and asymmetric convective heat transfer conditions during vehicle operation, do not consider the local material stretching and thinning caused by the load of the food container, which leads to a decrease in insulation performance, and that traditional designs are difficult to calculate and output discrete structural parameters suitable for actual processing and manufacturing while controlling the total material consumption.

[0006] To achieve the above objectives, the present invention provides the following technical solution: The first aspect of this invention provides a method for optimizing the parameters of a gradient composite insulated bag for fresh food delivery, comprising the following steps: Acquire vehicle speed data and ambient temperature sequence to generate a dynamic convection boundary matrix; Obtain the geometric dimensions and load weight parameters of the food container to be delivered, solve the two-dimensional tensile strain field, and output the local thermal resistance reduction matrix and heat capacity reduction matrix; Initialize the spatial design variable field, receive the dynamic convection boundary matrix, the local thermal resistance reduction matrix, and the heat capacity reduction matrix to establish the continuous medium transient thermal conduction partial differential control equation, perform forward numerical integration on the continuous medium transient thermal conduction partial differential control equation, and store the spatial temperature field data; Based on the space temperature field data, an optimization objective function is defined, a Lagrange functional of the optimization objective function is constructed, and inverse time integration is performed to obtain the accompanying temperature field. Based on the space temperature field data and the accompanying temperature field, the sensitivity gradient is extracted, and the space design variable field is updated until the optimization objective function satisfies the convergence threshold. The spatial design variable field that satisfies the convergence threshold is mapped to discrete manufacturing parameters, and the thermal insulation bag structure design parameters are output.

[0007] Preferably, the step of acquiring vehicle driving time-series speed data and ambient temperature sequence to generate a dynamic convection boundary matrix includes: The vehicle driving time-series speed data is converted into transient Reynolds numbers; The transient convective heat transfer coefficient time series is calculated based on the ambient temperature sequence, the transient Reynolds number, the air Prandtl number, and the correlation of the convective heat transfer criterion number. The ambient temperature sequence and the transient convective heat transfer coefficient time sequence are aligned in time steps and combined to generate the dynamic convective boundary matrix.

[0008] Preferably, the steps of obtaining the geometric dimensions and load weight parameters of the food container to be delivered, solving the two-dimensional tensile strain field, and outputting the local thermal resistance reduction matrix and heat capacity reduction matrix include: The geometric dimensions of the food box to be delivered are mapped to the boundary constraints of the internal space force contact surface; The load weight parameters are converted into uniformly distributed gravity loads and normal pressures on the contact surfaces acting on the bottom and sidewalls. The two-dimensional tensile strain field is solved based on the boundary constraints of the stressed contact surface, the uniformly distributed gravity load, and the normal pressure on the contact surface. By combining the Poisson's ratio of the insulation bag wall material and the preset pore compression empirical function, the two-dimensional tensile strain field is transformed into a physical property attenuation coefficient, thereby generating the local thermal resistance reduction matrix and the thermal capacity reduction matrix.

[0009] Preferably, the steps of initializing the spatial design variable field, receiving the dynamic convection boundary matrix, the local thermal resistance reduction matrix, and the heat capacity reduction matrix, establishing the transient thermal conduction partial differential control equation for the continuous medium, performing forward numerical integration on the transient thermal conduction partial differential control equation for the continuous medium, and storing the spatial temperature field data include: Based on the aforementioned spatial design variable field, the basic thermal conductivity function and the basic volumetric specific heat capacity function are obtained; The equivalent thermal conductivity is obtained by multiplying the basic thermal conductivity function with the local thermal resistance reduction matrix. The equivalent volumetric specific heat capacity is obtained by multiplying the basic volumetric specific heat capacity function by the heat capacity reduction matrix; The equivalent thermal conductivity and the equivalent volumetric specific heat capacity are embedded as spatial variation coefficients into the transient thermal conductivity partial differential control equation of the continuous medium. Apply convective heat transfer boundary conditions defined by the dynamic convection boundary matrix at the outer boundary of the overall three-dimensional spatial domain and perform the forward numerical integration.

[0010] Preferably, the step of defining an optimization objective function based on the space temperature field data and constructing a Lagrange functional of the optimization objective function includes: Calculate the deviation between the spatial temperature field data and the target temperature for fresh food preservation; Construct a time integral term for the internal cavity temperature deviation that characterizes the cumulative degree of deviation between the actual temperature and the target temperature for fresh food preservation; The total material consumption is obtained by integrating the basic density of the material within the spatial domain of the solid structure of the insulation bag wall. A quality penalty weighting coefficient is introduced to scale the total material consumption to generate a total material consumption penalty term; The optimization objective function is constructed by adding the time integral term of the internal cavity temperature deviation to the total material consumption penalty term; The Lagrange functional is constructed by combining the transient thermal conductivity partial differential control equation of the continuous medium, which includes the equivalent thermal conductivity and the equivalent volumetric specific heat capacity, with the objective function as the state constraint.

[0011] Preferably, the step of performing inverse time integration to obtain the accompanying temperature field includes: Solving the variational equation for the temperature state variable using the Lagrange functional yields the transient adjoint thermal conduction partial differential equation. The partial derivative of the optimization objective function with respect to the temperature field state variables is set as the adjoint source term of the transient adjoint heat conduction partial differential equation; Force the associated temperature variable value in the entire spatial domain to be set to zero at the delivery terminal time as the inverse integration start state; The transient accompanying thermal conduction partial differential equation is solved in the reverse direction of the time axis to obtain the accompanying temperature field.

[0012] Preferably, the step of extracting the sensitivity gradient based on the space temperature field data and the accompanying temperature field, and updating the space design variable field, includes: Calculate the analytical functions of the partial derivatives of the equivalent thermal conductivity and the equivalent volumetric specific heat capacity with respect to the spatial design variable field; The sensitivity gradient is generated by multiplying the time rate of change and spatial gradient vector of the spatial temperature field data with the state value and spatial gradient vector of the accompanying temperature field. Define the boundary constraints of the spatial design variable field; The nonlinear optimization solver is invoked to iteratively update the spatial design variable field by combining the sensitivity gradient with the boundary constraints.

[0013] Preferably, the step of updating the spatial design variable field until the optimization objective function satisfies the convergence threshold includes: The updated spatial design variable field is used as the basis for recalculating the total optimization objective function value under the current state; Calculate the absolute value of the difference between the total optimization objective function value and the previous state, and divide it by the previous baseline state to obtain the relative decrease scalar. When the relative decrease scalar is less than the convergence threshold, the cyclic iteration channel is determined to meet the convergence condition.

[0014] Preferably, the step of mapping the spatial design variable field that satisfies the convergence threshold to discrete manufacturing parameters and outputting the thermal insulation bag structural design parameters includes: Based on the contour distribution characteristics and spatial gradient change rate of the continuous spatial design variable field, the overall entity space domain is divided into multiple non-overlapping sub-regions. For each of the divided sub-regions, the integral error between the distribution of continuous variables and the preset candidate parameter set is calculated, and a local mapping cost function is constructed. The discrete specification that minimizes the integral of the local mapping cost function is selected as the local optimal discrete specification of the sub-region. The locally optimal discrete specifications of all the sub-regions are summarized to generate the discrete manufacturing parameters that include the discrete thickness distribution and discrete porosity distribution of each sub-region.

[0015] A second aspect of the present invention provides a gradient composite insulated bag for fresh food delivery, comprising: The insulated bag body, the wall structure of the insulated bag body includes, from the outside to the inside, an outer protective layer, a middle gradient foamed insulation layer and an inner food-grade contact layer. The thickness of the intermediate gradient foamed insulation layer has a high value set in the windward area of ​​the insulation bag body and a low value set in the leeward area of ​​the insulation bag body. The intermediate gradient foamed insulation layer has a deformation compensation topology in the bottom bearing area and the side wall stretching area of ​​the insulation bag body. The deformation compensation topology is manifested as a discrete thickened step structure that protrudes inward or outward, and the porosity value corresponding to the deformation compensation topology is lower than the porosity value of the top and free side wall areas of the insulation bag body that are not under stress.

[0016] This invention provides a method for optimizing the design parameters of a gradient composite insulated bag for fresh food delivery, as well as the insulated bag itself. It offers the following advantages: 1. This invention generates a dynamic convection boundary matrix by acquiring vehicle speed data and ambient temperature sequences, and performs inverse time integration optimization in conjunction with the accompanying temperature field, overcoming the shortcomings of conventional insulated bag designs that rely solely on static environmental assumptions. The insulated bag generated using this method has a higher thickness in the windward region and a lower thickness in the leeward region, thus accurately matching the real dynamic wind speed convection heat transfer conditions during food delivery, specifically reducing heat loss in high-convection areas, and improving the overall cold-keeping effect during delivery.

[0017] 2. This invention transforms the geometric dimensions and load weight parameters of the food container to be delivered into a two-dimensional tensile strain field, and extracts the local thermal resistance and heat capacity reduction matrices accordingly, introducing them into the governing equations. This allows the optimization calculation to compensate for the material thinning caused by load-bearing stress in advance. The resulting insulated bag generates a deformation-compensating topology in the bottom load-bearing area and the side wall tensile area, namely, discrete thickened steps with low local porosity. This avoids the local attenuation of insulation performance caused by load-bearing deformation, ensuring the stability of the insulation effect under full load.

[0018] 3. This invention introduces a total material consumption penalty term when constructing the Lagrange functional of the objective function, and transforms the continuous spatial design variable field that satisfies convergence into discrete manufacturing parameters based on a local mapping cost function. This limits material usage while solving for the optimal insulation structure, avoiding excessive design redundancy. The final output discrete thickness and discrete porosity distribution parameters solve the processing problem of accurately molding continuous gradient foamed structures in actual foaming molds, realizing the transformation of parameter optimization results into actual industrial manufacturing. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method of the present invention; Figure 2This is a system architecture diagram of the present invention; Figure 3 This is a schematic diagram of the gradient composite insulated bag for fresh food delivery according to the present invention; Figure 4 This is a comparison diagram of the bottom surface thickness collapse before and after the present invention is subjected to full load. Figure 5 This is a comparison chart of heat penetration under dynamic wind resistance conditions according to the present invention. Figure 6 This is a curve showing the evolution of the load-based cooling effect duration of the present invention. The components include: 1. the insulated bag body; 2. the protective layer; 3. the foamed insulation layer; and 4. the contact layer. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] See attached document Figure 1 and 2 This invention provides a parameter optimization design system for a gradient composite insulated bag for fresh food delivery. The system operates on a computer device with data processing capabilities and includes: The module includes a boundary generation module, a mechanical mapping module, a forward solution module, an inverse solution module, a sensitivity analysis module, and a discrete mapping module.

[0022] The boundary generation module receives meteorological data and vehicle speed data from the food delivery route, and calculates the time series of the transient convective heat transfer coefficient of the outer boundary. The mechanical mapping module obtains the geometric dimensions and load weight parameters of the food containers to be delivered, solves the two-dimensional tensile strain field of the insulated bag wall, and generates the local thermal resistance reduction matrix and heat capacity reduction matrix.

[0023] The forward solution module is used to establish the transient thermal conduction partial differential equations of the continuous medium based on the local thermal resistance reduction matrix and the heat capacity reduction matrix, and to perform forward integration in the time domain, storing the spatial temperature field data. The inverse solution module is used to define the optimization objective function and construct the Lagrange functional, derive the transient adjoint thermal conduction partial differential equations, call the spatial temperature field data and perform numerical integration in reverse along the time axis to obtain the adjoint temperature field.

[0024] The sensitivity analysis module extracts the sensitivity gradient of the objective function based on the space temperature field data and its accompanying temperature field data, and inputs it into the optimization solver to update the space design variable field. The discrete mapping module maps the space design variable field to discrete manufacturing parameters after the objective function iteratively converges, and outputs the structural model file.

[0025] See attached document Figure 2 This invention also provides a method for optimizing the parameters of a gradient composite insulated bag for fresh food delivery. The specific steps of executing this design method in the above-mentioned system for optimizing the parameters of a gradient composite insulated bag for fresh food delivery are as follows: In step S100, the boundary generation module acquires the vehicle driving time-series speed data, converts it into transient Reynolds number, and calculates the transient convective heat transfer coefficient time series based on the air Prandtl number and convective correlation, outputting the dynamic convective boundary matrix.

[0026] In step S200, the mechanical mapping module performs unidirectional coupling calculations of the physical field. The mechanical mapping module obtains the two-dimensional tensile strain field of the insulation bag wall through thin-shell mechanical calculations, and outputs a local thermal resistance reduction matrix and a heat capacity reduction matrix embedded with deformation loss weights.

[0027] In step S300, the forward solver receives the dynamic convection boundary matrix, the local thermal resistance reduction matrix, and the heat capacity reduction matrix, establishes the three-dimensional heat conduction control equation, performs forward numerical integration in the spatiotemporal domain, and stores the spatial temperature field data of each time step in the register.

[0028] In step S400, the inverse solver module defines an optimization objective function based on the spatial temperature field data output by the forward solver module. This objective function includes a time integral term for the internal cavity temperature deviation and a penalty term for the total material consumption. The inverse solver module constructs the Lagrange functional of the objective function and performs the construction and inverse time integration of the transient adjoint partial differential equation.

[0029] In step S500, the sensitivity analysis module extracts the global parameter sensitivity gradient through analytical formulas and updates the spatial design variable field of material porosity and thickness.

[0030] The system forms an iterative path for parameter optimization through data interaction between the forward solution module, the inverse solution module, and the sensitivity analysis module. After each iteration, the system determines whether the relative decrease in the objective function meets the set convergence threshold.

[0031] Step S600: When the convergence threshold is met, the discrete mapping module obtains the converged spatial design variable field, maps the continuously distributed spatial design variable field to discrete manufacturing parameters in the preset discrete thickness and density level set, and outputs the final gradient composite insulation bag structure design parameters.

[0032] The boundary generation module performs the specific workflow of dynamic convection boundary time series parameterization.

[0033] The boundary generation module acquires external environmental parameters and kinematic parameters during the food delivery process. Specifically, the external environmental parameters may include: Meteorological data such as ambient temperature and wind speed are collected. Kinematic parameters are specifically represented by vehicle speed data over time. In specific implementation scenarios, vehicle speed data can be obtained by calculating the time-varying position coordinates collected by the GPS receiver on the delivery vehicle, or by obtaining the acceleration integral output by the inertial measurement unit. The boundary generation module extracts the temperature information from the meteorological data into an ambient temperature sequence and the vehicle speed into a speed sequence. This is a time variable on the delivery timeline.

[0034] The boundary generation module determines the thermal properties of the air outside the boundary layer based on the ambient temperature sequence. These thermal properties include: The kinematic viscosity, air density, and thermal conductivity of air are calculated. The boundary generation module simultaneously extracts the dimensions of the windward side from the 3D geometric model of the insulation bag as the feature length. For the values ​​of air kinematic viscosity, air density, and thermal conductivity at different temperatures and the interpolation calculation methods, those skilled in the art can consult standard heat transfer property handbooks. The property interpolation process is well-known in the field and will not be elaborated upon here.

[0035] The boundary generation module converts the velocity sequence into a transient Reynolds number to characterize the ratio of inertial force to viscous force in the external flow field of the insulation bag. This calculation is achieved through algebraic operations on air density, vehicle speed, characteristic length, and air kinematic viscosity. Subsequently, the boundary generation module solves for the transient convective heat transfer coefficient time series of the outer boundary based on the Prandtl number and the correlation formula of the convective heat transfer criterion number. Specifically, the boundary generation module uses the forced convection empirical formula, combined with the convective correlation empirical coefficient determined by the approximately bluff body flow model around the windward side of the insulation bag, to calculate the transient Nusselt number. The acquisition of the Prandtl number is a well-known technique in this field and will not be elaborated here. The boundary generation module calculates the transient convective heat transfer coefficient time series by dividing the air thermal conductivity by the characteristic length and multiplying it by the obtained transient Nusselt number.

[0036] The boundary generation module aligns the calculated ambient temperature sequence with the transient convective heat transfer coefficient time sequence in time steps, and combines them to generate a dynamic convective boundary matrix. This dynamic convective boundary matrix characterizes the time-domain fluctuations in the heat exchange rate between the outer wall of the insulated bag and the surrounding air under dynamic conditions such as stationary, accelerating, constant speed driving, and decelerating vehicles. This dynamic convective boundary matrix is ​​then input into the forward solution module as boundary condition data for solving subsequent partial differential equations.

[0037] The mechanical mapping module executes a specific workflow based on non-uniform mechanical mapping and thermal resistance reduction logic under load strain. When a heavy load is loaded onto the insulation bag, the flexible bag wall undergoes tensile deformation, leading to a decrease in the porosity and thickness of the internal foam material. This deformation causes a non-uniform decrease in local thermal resistance and heat capacity in the corresponding area.

[0038] The mechanics mapping module acquires the geometric dimensions and load weight parameters of the food container to be delivered. The geometric dimensions specifically include the length, width, and height matrix data of the container. The load weight parameter is the mass scalar of the container after it is loaded with food. The mechanics mapping module maps the container's geometric dimensions to the boundary constraints of the force-bearing contact surfaces within the insulated bag, and converts the load weight parameters into uniformly distributed gravity loads acting on the bottom and sidewalls, and normal pressures on the contact surfaces.

[0039] The mechanics mapping module solves for the two-dimensional tensile strain field of the insulation bag wall based on the boundary constraints and pressure data of the stressed contact surface. The module calls its internal thin-shell elasticity mechanics solver to obtain the two-dimensional tensile principal strain field in the spatial domain. For methods of calculating the tensile strain field of flexible materials based on thin-shell theory, those skilled in the art can use commercial finite element solvers to assemble the element stiffness matrix and solve for nodal displacements. The specific construction of the mechanical equilibrium equations and the numerical solution process are well-known techniques in this field and will not be elaborated upon here.

[0040] The mechanical mapping module constructs spatially distributed local thermal resistance reduction matrices and heat capacity reduction matrices based on the two-dimensional tensile principal strain field. Combining the Poisson's ratio of the insulation bag wall material and a preset empirical function for pore compression, the mechanical mapping module directly maps tensile strain data into property attenuation coefficients. The values ​​of the aforementioned reduction matrices range from 0 to 1; the larger the stress deformation in a region, the smaller the corresponding reduction matrix coefficient value due to material compaction and thinning.

[0041] The mechanical mapping module further uses the local thermal resistance reduction matrix and heat capacity reduction matrix as penalty weights to define the equivalent thermal conductivity and equivalent volumetric specific heat capacity after deformation. The system defines a basic design variable field to be optimized, specifically indicating the initial porosity and initial thickness distribution of the insulation bag wall surface under unstressed natural conditions. Based on this basic design variable field, the mechanical mapping module obtains the basic thermal conductivity function and basic volumetric specific heat capacity function for the unstressed state. The mechanical mapping module performs a unidirectional coupling operation of the physical field, multiplying the basic coefficient function by the corresponding reduction matrix to obtain the equivalent thermal conductivity and equivalent volumetric specific heat capacity after coupled stress deformation.

[0042] Through the aforementioned algebraic coupling operations, the mechanical mapping module hard-codes the thermal insulation performance degradation characteristics caused by structural mechanics into the thermal property parameters in the form of multiplicative coefficients, accurately characterizing the reduction in equivalent thermal resistance in the stressed tensile region. This equivalent thermal conductivity and equivalent volumetric specific heat capacity data are then output to the forward solution module as property coefficients for establishing the partial differential governing equations.

[0043] The forward solver module executes the specific workflow for establishing the three-dimensional thermal conduction control equations for coupled thermal resistance reduction. After acquiring the physical property data output by the mechanical mapping module, the system needs to construct a mathematical model describing the heat transfer process in three-dimensional space to provide a basis for subsequent parameter optimization.

[0044] The forward solver module defines the overall three-dimensional spatial domain for the system's thermodynamic integral calculations. This overall three-dimensional spatial domain consists of two subdomains: a spatial domain representing the solid structure of the insulated bag's walls, and a spatial domain representing the cavity space inside the insulated bag that holds the food and air. Within this overall three-dimensional spatial domain, the forward solver module defines a scalar temperature field that evolves spatially and temporally.

[0045] The forward solver module in the overall three-dimensional space domain The partial differential governing equations for transient thermal conduction in continuous media are established internally. The forward solver module calls the equivalent thermal conductivity output by the mechanical mapping module. With equivalent volumetric specific heat capacity We embed this as the spatial variation coefficient of the equation into the energy conservation derivation to construct the following governing equation: ; in, Indicates spatial location In Temperature value at any given time; This represents the partial derivative of temperature with respect to time. This represents the vector differential operator (Nabla operator). The left side of the above equation characterizes the rate of change of internal energy per unit volume of material with time, while the right side characterizes the net heat flowing into that volume unit through a combination of divergence and gradient operators. By directly internalizing the stress-deformation reduction matrix into the property coefficients of the partial differential equation, this governing equation naturally possesses the mathematical analytical capability to capture the phenomenon of local heat flux surges caused by thermal resistance collapse in stress concentration regions.

[0046] The forward solver module sets external boundary conditions for the aforementioned partial differential governing equations. It extracts the dynamic convection boundary matrix output by the boundary generation module. At the outer boundary of the overall three-dimensional spatial domain, the forward solver module applies time-varying third-type boundary conditions, namely the Robin boundary conditions.

[0047] The convective heat transfer boundary condition controls the heat flux density conducted from the interior to the outer surface of the insulation bag, which is equal to the heat flux density dissipated by the outer surface and the surrounding flowing air through convection. The external heat loss flux is determined by the transient convective heat transfer coefficient, the instantaneous surface temperature, and the ambient temperature sequence. The forward solution module establishes the initial time conditions and the internal interface continuity conditions of the system.

[0048] The forward solution module is set so that the temperature distribution within the overall three-dimensional spatial domain satisfies the initial state field at the initial moment, where the initial state field consists of a preset constant room temperature and the initial cold-state temperature distribution of the food. For the convective-radiative coupled heat transfer process between the food and air within the cavity space domain, those skilled in the art can use the equivalent comprehensive thermal conductivity amplification method to perform equivalent thermal resistance treatment of the fluid-solid interface. The specific setting of the equivalent parameters is a well-known technique in this field and will not be elaborated here.

[0049] The forward solver module has now completed the construction of the three-dimensional thermal conductivity control equations, including dynamic boundaries and deformation property losses. The forward solver module executes the specific workflow of forward numerical integration in the spatiotemporal domain and the state quantity storage strategy. After establishing the transient thermal conductivity partial differential control equations and boundary conditions for the continuous medium, the system obtains the temperature field evolution data under discrete states through a discretization algorithm. The forward solver module performs spatial discretization of the geometric domain on the entire three-dimensional spatial domain.

[0050] The forward solver module calls a mesh generation program to divide the overall three-dimensional spatial domain into a finite number of discrete computational mesh elements. Based on the finite volume method or the finite element method, the forward solver module performs spatial integration on the transient thermal conduction partial differential equations of the continuous medium in each computational mesh element, transforming the continuous partial differential equations containing differential operators into a system of nonlinear algebraic equations containing spatial nodal temperature vectors. For the Jacobian matrix assembly and linearization operations in the spatial discretization process of the partial differential equations, those skilled in the art can use standard finite element stiffness matrix assembly methods. The specific numerical discretization process is well-known in the field and will not be elaborated here.

[0051] The forward solver module performs forward numerical integration calculations in the time domain. It sets the time step for computation and divides the total delivery cycle into several consecutive discrete time steps leading to the delivery terminal. Considering the nonlinear characteristics of the deformation equivalent property coefficients, the forward solver module employs an absolutely stable backward Euler implicit difference scheme to solve the discretized governing equations in the time-shifting direction. By evaluating the nodal differences in the temperature field state variables between the current iteration time step and the previous time step, and combining this with the transient convection heat transfer conditions at the outer boundary, the forward solver uses a nonlinear algebraic equation solver to iteratively calculate from the initial time until the delivery terminal time, thus acquiring the forward time evolution trajectory of the physical field.

[0052] The forward solving module employs a caching and storage strategy for historical state data. Since subsequent solutions to the adjoint partial differential equations must be based on the original process state variables, and the mathematical derivation of the adjoint state method requires inverse integration along the time axis, the system needs to retain the forward solving results for each discrete time step. The forward solving module completes the solution at any time step... After iteratively solving the temperature field, the spatial temperature field distribution vector of the global grid nodes at that moment is obtained. State variables are sequentially written to the computer device's memory registers or high-speed hard disk storage area according to a time index sequence. The forward solver establishes address pointers to these state variable data blocks. This state variable storage strategy ensures that the reverse solver subsequently writes these state variable data blocks along the time axis. return At any time, the forward thermodynamic state can be read, thus providing complete data support for the inverse gradient calculation of complex parameter domains.

[0053] The inverse algorithm module executes the specific workflow of defining the mathematical objective function and setting the penalty term. During the inverse parameter optimization process, the system needs to transform the engineering performance indicators and manufacturing cost constraints of the thermal insulation bag into a continuous scalar objective functional that can be calculated by the algorithm.

[0054] The reverse engineering module defines the target region and preset temperature parameters for thermodynamic performance evaluation. It extracts the overall three-dimensional spatial domain defined by the forward engineering module, setting the internal cavity of the insulated bag containing the food and air as the temperature evaluation region. Target temperature for fresh food preservation Specifically, it is a constant scalar value that characterizes the optimal storage environment for food ingredients. Its subordinate characteristics include, but are not limited to, the refrigeration temperature threshold above zero degrees Celsius under the delivery conditions of chilled fresh meat, or the temperature threshold below freezing point in Celsius under the delivery conditions of frozen foods such as ice cream.

[0055] The reverse engineering module constructs a time integral term for the internal cavity temperature deviation. This integral term quantifies the cumulative degree to which the actual thermodynamic evolution of the food cavity deviates from the ideal cold-keeping state over the entire delivery cycle. The reverse engineering module calculates the deviation between the actual temperature and the target temperature for fresh food preservation based on spatial temperature field data. The module constructs the integrand by calculating the square of the difference. Using the squared difference ensures that both overheating deviations above the target temperature and undercooling deviations below the target temperature are accumulated as positive penalties, preventing positive and negative errors from canceling each other out. The reverse engineering module performs a double integral of this squared difference function in both the time and spatial domains.

[0056] The inverse solver module constructs a penalty term for total material consumption and combines it to generate the overall optimization objective function. At the manufacturing level, to prevent the solver from outputting meaningless solutions with infinitely thick walls and zero porosity, lightweighting and cost constraints need to be introduced. The inverse solver module calls the fundamental density function characterizing the distribution of material density with design variables. The reverse engineering module performs volume integration on the basic material density within the spatial domain of the solid structure of the insulation bag wall to obtain the total material consumption. The reverse engineering module introduces a mass penalty weighting coefficient. The volume integral term is then weighted and scaled to generate a penalty term for total material consumption. For conventional scalarization methods that achieve multi-objective Pareto optimization by setting penalty weight coefficients of different magnitudes, those skilled in the art can adjust the settings according to specific manufacturing cost limits. The adjustment of these weight factors is a well-known technique in the field and will not be elaborated upon here.

[0057] The reverse engineering module adds the time integral term of the internal cavity temperature deviation to the penalty term of the total material consumption to construct the overall optimization objective function. : ; in, This represents the overall optimization objective function of the system; Indicates the delivery terminal time. Represents the cavity space domain. Indicates spatial location In The actual solution of the spatial temperature field data at that time. Indicates the target temperature for keeping fresh food cold. This represents the quality penalty weighting coefficient. This represents the spatial domain of the solid structure of the insulated bag wall. Indicates spatial location Subject to basic design variables The fundamental density function is determined.

[0058] The reverse engineering module, by defining the aforementioned optimization objective function, unifies the originally conflicting requirements for extreme thermal insulation performance and lightweight cost reduction within a single differentiable mathematical framework. This optimization objective function is not only used to determine whether the system optimization iteration has converged, but also serves as the fundamental mathematical object for subsequently constructing the transient adjoint state-law Lagrange functional and deriving the sensitivity of the partial differential equation.

[0059] The inverse solver module executes the specific workflow of constructing the transient adjoint partial differential equation and performing inverse time integration. When dealing with high-dimensional optimization problems involving design variables with tens of thousands of spatial grid nodes, using the finite difference method to solve for the perturbation of each variable would consume enormous computational resources. The system introduces the adjoint state method to transform the gradient calculation in the high-dimensional parameter domain into the process of solving a single adjoint equation.

[0060] The inverse solution module introduces the adjoint temperature field as an intermediate state variable in the mathematical derivation. Mathematically, this adjoint temperature field is represented by Lagrange multipliers. The inverse solution module uses the transient thermal conductivity partial differential control equations for the continuous medium constructed by the forward solution module as state constraints, combining them with the overall optimization objective function defined by the system to construct a Lagrange functional for variational derivation. This functional construction method internalizes the system's energy conservation law within the mathematical optimization framework.

[0061] The inverse solver module derives the transient adjoint heat conduction partial differential equation based on the constructed Lagrange functional. It then solves for the variational function of the Lagrange functional with respect to the temperature state variable. Using the integration-by-parts rule in the spatial and temporal domains, the module transfers the spatial gradient differential operator and the time partial derivative operator from the original temperature variable to the adjoint temperature field variable. By setting the variation of the functional with respect to the temperature state variable to zero, the inverse solver module obtains the corresponding transient adjoint partial differential equation.

[0062] The left-hand side of the adjoint equation includes the partial derivative of the adjoint temperature field with respect to time and the equivalent volumetric specific heat capacity, while the right-hand side includes the divergence of the spatial gradient of the adjoint temperature field and the adjoint source term. The numerical distribution of the physical property parameter matrix controlling this equation is strictly consistent with the solution process for forward heat conduction. The specific mathematical form of the adjoint source term is determined by the partial derivative of the overall optimization objective function with respect to the temperature field state variables.

[0063] The inverse solver module configures the spatial distribution characteristics of the accompanying source term. When the location is within the defined cavity space domain, the accompanying source term is assigned the deviation between the actual solved spatial temperature field data and the target temperature for fresh food preservation. When the location is on the outer wall structure of the cavity space domain, the accompanying source term is truncated to zero. Through this segmented setting, the inverse solver module directly transforms the cold and heat deviation values ​​within the evaluation area into spatial heat sources driving the evolution of the accompanying temperature field. The inverse solver module further derives the accompanying boundary conditions and time terminal conditions of the transient accompanying partial differential equation.

[0064] At the outer boundary of the insulation bag, the inverse solution module sets the convection boundary conditions of the accompanying field to ensure that the outward transfer of accompanying energy is also governed by the transient convection heat transfer coefficient. Since the accompanying partial differential equation exhibits ill-posedness due to the negative time derivative term in its forward evolution, the solution process must proceed in the reverse time direction. Based on this, the inverse solution module sets the system's evolution terminal conditions, forcing the accompanying temperature variable value across the entire spatial domain to be set to zero at the delivery terminal time as the inverse integration initiation state.

[0065] The reverse solving module performs numerical integration calculations in reverse order along the time axis. It establishes a time traversal loop that is the opposite of the forward solving module, starting from the delivery terminal time. The process begins by iterating backwards to the initial time step. Within each backward time step, the backward solution module retrieves the forward spatial temperature field data stored in the register for that corresponding moment. Used to calculate the accompanying source term at the current time. For the construction of the inverse implicit difference discretization scheme for partial differential equations and the iterative solution of linear algebraic equation systems, those skilled in the art can use the corresponding finite element inverse calculation package for processing. The specific matrix solving operations are well-known techniques in this field and will not be elaborated upon here. The inverse solution module, through complete inverse time extrapolation, obtains the accompanying temperature field data in the entire spatiotemporal domain. This data contains the mathematical mapping information of the objective function error back to each spatial grid node.

[0066] The sensitivity analysis module performs the differential-free analysis and extraction of global parameter sensitivity gradients. After obtaining the system's thermodynamic forward and inverse evolution states, the system needs to quantify the impact of hundreds of thousands of design variables distributed on various grid nodes on the overall performance into a directly computable gradient vector field.

[0067] The sensitivity analysis module retrieves the full-temporal spatial temperature field data cached during the forward solution process from the computer's storage space. And the accompanying temperature field data obtained from the reverse solution process. The sensitivity analysis module simultaneously extracts the fundamental design variable fields set on each spatial grid cell. .

[0068] The sensitivity analysis module obtains the analytical functions of the partial derivatives of the equivalent thermal conductivity and equivalent volumetric specific heat capacity with respect to the basic design variable field. Since the mechanical mapping module introduces a reduction matrix strongly correlated with stress deformation when establishing the physical property model, the sensitivity analysis module calculates the corresponding parameter change rate matrix using the chain rule. and Because the thickness of the stressed tensile region decreases, the absolute value of the partial derivative in that region is automatically amplified. This algebraic correlation method transforms the penalty for weak load-bearing links into optimization weights in the mathematical gradient space.

[0069] The sensitivity analysis module performs differenceless analytical calculation of the sensitivity gradient through spatial operators and time integration operations. It combines the time rate of change of the forward physical field with the spatial gradient vector, and the state value of the inverse adjoint field with the spatial gradient vector. In conjunction with the material's fundamental density penalty term, the module uses the following analytical partial derivative formula to calculate the unconditional partial derivatives of the objective function with respect to the design variables at various spatial locations: ; in, Indicates spatial location The sensitivity gradient vector of the objective function; Indicates the delivery terminal time; This represents the spatial gradient vector accompanying the temperature field data; The matrix representing the partial derivatives of the equivalent thermal conductivity with respect to the basic design variables; Represents the spatial gradient vector of spatial temperature field data; This represents the accompanying temperature field data; The matrix representing the partial derivatives of the equivalent volumetric specific heat capacity with respect to the basic design variables; This represents the partial derivative of the spatial temperature field data with respect to time. This represents the quality penalty weighting coefficient; This represents the partial derivative of the fundamental density function with respect to the fundamental design variables.

[0070] The sensitivity analysis module outputs sensitivity gradient field data covering the entire spatial domain. Conventional finite difference methods require perturbing each node parameter and repeatedly solving forward partial differential equations when obtaining gradients for tens of thousands of dimensional variables, leading to the curse of dimensionality. The system, relying on the aforementioned analytical partial derivative formula, only needs to perform one forward thermal solution of the physical field and one inverse integral derivation of the adjoint field to simultaneously extract the high-dimensional sensitivity gradient of all grid nodes within a single mathematical expression. This computational mechanism reduces the computational complexity of spatial parameter optimization from large-scale operations positively correlated with the dimension of design variables to constant-level matrix multiplication and addition operations. For the Gaussian numerical integration and difference operator approximation of spatially distributed partial derivatives on a discrete computational grid, those skilled in the art can utilize existing numerical computation function libraries; the specific discretization algebraic operations are well-known techniques in the field and will not be elaborated upon here. The sensitivity gradient field data extracted by the sensitivity analysis module is subsequently input into the system's nonlinear programming optimizer for parameter update iteration.

[0071] The sensitivity analysis module performs a gradient-based nonlinear optimization process. After acquiring the sensitivity gradient field data across the entire spatial domain, the system updates the design variables distributed across each grid node according to the descent direction indicated by the mathematical gradient, thereby driving the insulation bag structure toward its optimal performance state.

[0072] The sensitivity analysis module sets the boundary constraints for the spatial design variable field. Basic design variable field. In space, the porosity of the foam material in the insulation bag and the thickness of the solid wall are represented. To ensure that the structural parameters output by the optimization solver are meaningful in actual manufacturing, the sensitivity analysis module defines upper and lower bounds for all spatial nodes. Specifically, the sensitivity analysis module sets the lower bound of the porosity variable to the intrinsic porosity of the solid, unfoamed material, and the upper bound to the critical point of maximum porosity achievable by the multi-layer foaming process. The sensitivity analysis module also constrains the thickness variable between the minimum safe thickness determined by the structural stiffness and the maximum forming thickness determined by the mold size. These operations transform the unconstrained gradient descent process into a nonlinear programming problem with boundary constraints.

[0073] The sensitivity analysis module invokes a nonlinear optimization solver with boundary constraints to iteratively update the design variable field. Since the spatial grid node variables reach hundreds of thousands or even millions, directly calculating the second derivative matrix of the objective function is extremely memory-intensive. Therefore, the sensitivity analysis module employs a quasi-Newton algorithm with limited memory to perform dimensionality reduction iterative calculations. Specifically, the system calls the L-BFGS-B algorithm, inputting the fundamental design variable field of the current iteration step and the obtained sensitivity gradient vector into the solver.

[0074] The solver implicitly constructs the inverse of the approximate Hessian matrix of the objective function by extracting and storing the variable displacement differences and gradient differences from the most recent iterations. The sensitivity analysis module combines this approximate inverse matrix, the current iteration step size determined by line search, and the analytically obtained sensitivity gradient vector to calculate the structural displacement. It then uses a truncated projection operator to force the deviated variables back within the boundary constraints, thereby generating the updated variable field for the next iteration.

[0075] For the specific derivation of the approximate Hessian matrix update in the quasi-Newton algorithm and the line search rule satisfying the Wolf condition, those skilled in the art can refer to standard numerical optimization manuals, as these are well-known techniques in the field and will not be elaborated here. The discrete mapping module performs iterative convergence control and termination condition determination. After the sensitivity analysis module outputs the updated basic design variable field, the discrete mapping module uses it as the new physical property basis and calls the forward solution module to recalculate the total optimization objective function value under the current state.

[0076] The discrete mapping module extracts the numerical parameters of the function from the previous iteration. By calculating the absolute value of the difference between the current state and the previous state and dividing it by the previous baseline state, it obtains the relative decrease scalar of the objective function between two adjacent iterations, which drives the system to make a rigorous determination of the convergence threshold in the form of positive real numbers.

[0077] The discrete mapping module compares the relative decrease with the system's preset convergence threshold. If the relative decrease is less than the convergence threshold, or if the current number of iterations exceeds the preset maximum cycle limit parameter, the discrete mapping module determines that the system's iterative cycle meets the convergence condition and terminates the nonlinear optimization process. If the convergence condition is not met, the discrete mapping module feeds the updated fundamental design variable field back into the forward solution module, and the system restarts a new round of optimization loops, including forward evolution of the physical field, inverse derivation of the adjoint field, and re-extraction of sensitivity.

[0078] The discrete mapping module executes the specific workflow of industrial-tolerance discretization and dimensionality reduction mapping of continuous physical fields. After the nonlinear optimization iteration converges, the spatial design variable field obtained by the system exhibits an infinite-order continuous distribution in the mathematical space. Due to the inherent engineering tolerance limitations of multi-layer foaming equipment and mold processing technology, it is impossible to directly manufacture entities with parameters that change continuously in space. Therefore, the system needs to transform this continuous field into a set of discrete parameters of finite order.

[0079] The discrete mapping module acquires the converged continuous spatial design variable field. This module extracts a set of discrete thickness and density levels pre-stored within the computer equipment. This set is specifically represented as a series of discrete numerical sequences, characterizing a list of material specifications that can be practically used on the production line. The system defines this as a candidate parameter set. ,in The total number of discrete specifications available, to These represent different constant levels determined by the manufacturing precision of the factory.

[0080] The discrete mapping module performs region clustering and segmentation on the 3D solid space model of the thermal insulation bag. Based on the contour distribution characteristics and spatial gradient change rate of the continuous spatial design variable field, the discrete mapping module divides the overall solid space domain into multiple non-overlapping sub-regions. Within each sub-region, the material thickness and porosity are forcibly unified to a single discrete specification. For conventional algorithms for 3D spatial region clustering and geometric segmentation based on contour data, those skilled in the art can use spatial clustering methods such as K-means. The unsupervised clustering process of the model mesh is a well-known technique in this field and will not be elaborated here.

[0081] The discrete mapping module performs dimensionality reduction mapping operations based on the least squares method. For each defined entity sub-region, the discrete mapping module calculates the integral error between the distribution of continuous variables within that region and the discrete specifications in the candidate parameter set. The system constructs a local mapping cost function by performing three-dimensional spatial integration on the squared difference between the scalar values ​​of the continuous spatial design variable field and the values ​​of specific candidate discrete specifications.

[0082] The discrete mapping module iterates through all candidate discrete specifications and selects the specification that minimizes the integral of the local mapping cost function as the locally optimal discrete specification for that block. The discrete mapping module then assigns this locally optimal specification to the corresponding 3D sub-region, completing the discretization and shaping of the region's structural parameters.

[0083] The discrete mapping module aggregates the locally optimal specification data of all sub-regions, completing the dimensionality reduction replacement of the overall wall parameters of the insulation bag. The discrete mapping module then reassembles these discrete manufacturing parameters into the system's geometric topology framework, generating and outputting a structural model file containing these discrete manufacturing parameters.

[0084] The structural model file is specifically manifested as a computer-aided design model file containing the discrete thickness dimensions of each layer and the three-dimensional coordinates of the foam density distribution, along with the corresponding bill of materials. Through the above operations, the system safely transitions the purely mathematical theoretical solutions derived from the partial differential equations into engineering drawing data adapted for industrial mass production lines.

[0085] See attached document Figure 3 , Figure 3 According to an embodiment of the present invention, the gradient composite insulated bag for fresh food delivery defines the specific physical characteristics of the gradient composite insulated bag in the structural model file output by the discrete mapping module.

[0086] The entity's characteristics are manifested as a multi-layered topological structure and a spatially non-uniform thickness and porosity distribution. Based on the system's solution and mapping results, the specific structural feature configuration steps and attribute display are as follows: The insulated bag body 1 has a multi-layer topological structure. The solid structure of the insulated bag body 1, from the outside to the inside, includes an outer protective layer 2, a middle gradient foamed insulation layer 3, and an inner food-grade contact layer 4. The outer protective layer is characterized by a polymer tear-resistant fiber fabric with a reflective coating. This material is used to isolate long-wave heat radiation from the external environment and block forced convection airflow during dynamic driving.

[0087] The inner food-grade contact layer 4 is constructed using a flexible polyethylene-vinyl acetate copolymer film, providing a waterproof and leak-proof interface. The intermediate gradient foamed insulation layer 3 is made of polyurethane or foamed polypropylene material, and its three-dimensional spatial structural parameters directly inherit the discrete manufacturing parameters output by the discretization dimension reduction mapping step. The hot-pressing bonding process between the outer protective layer 2 and the inner food-grade contact layer 4 with the intermediate gradient foamed insulation layer 3 can be achieved by those skilled in the art using conventional multi-layer flexible material composite equipment. The multi-layer roll bonding and edge sealing operations are well-known techniques in the field and will not be elaborated upon here.

[0088] Spatial discrete thickness distribution data of the intermediate gradient foamed insulation layer 3 are extracted. Considering the influence of the dynamic convection boundary matrix output by the forward solver module, the intermediate gradient foamed insulation layer 3 exhibits an asymmetric step thickness gradient in three-dimensional space. In the windward region of the insulation bag, because the integral value of the transient convective heat transfer coefficient time series is high, the system assigns its thickness to a high-value set through sensitivity analysis. In the leeward side and back-fitting area of ​​the insulation bag body 1, the convective heat dissipation effect is weakened, and the system assigns its thickness to a low-value set.

[0089] The aforementioned non-uniform size distribution in the spatial dimension avoids material redundancy in the leeward and top areas caused by the globally uniform thickness of traditional thermal insulation bags. Spatially discrete porosity distribution data of the intermediate gradient foamed insulation layer 3 are extracted. Combined with the two-dimensional tensile principal strain field calculated by the loading strain mapping and thermal resistance reduction module, the intermediate gradient foamed insulation layer 3 exhibits local adaptive density reinforcement characteristics in the load-bearing and stress-concentrated areas.

[0090] In the bottom bearing area and the side wall tension area connected to the handle of the thermal insulation bag body 1, the internal foam structure of the material is compressed due to the uniformly distributed gravity load and the normal pressure of the contact surface, posing a risk of local thermal resistance collapse. The system configures a lower porosity and a higher intrinsic density of the solid material in these stress-concentrated areas to compensate for the increase in the equivalent thermal conductivity caused by deformation by increasing the thickness of the solid phase thermal conductivity of the material.

[0091] In the unstressed top and free sidewall regions, the system employs a higher porosity to utilize the still air sealed within the foam pores, thereby enhancing insulation performance and reducing the overall weight of the insulation bag. A discrete mathematical correlation is established between the thickness of the intermediate gradient foam insulation layer 3 and its porosity in geometric space. This correlation, based on the dimensionality reduction mapping result after the iterative convergence of the objective function, describes the multi-tiered solid distribution state under actual industrial tolerance constraints.

[0092] The system extracts the indicator distribution rules defined on the overall entity space domain, and combines them with the local optimal discrete thickness specifications and local optimal discrete porosity specifications of each sub-region obtained by least squares mapping to construct discrete thickness distribution functions and discrete porosity distribution functions that characterize the spatial step characteristics.

[0093] When the entity's spatial coordinates are contained within a specified sub-region, its corresponding discrete step parameters are strictly activated and mapped. The mathematical model based on region indicator assembly described above precisely solidifies the continuous solution results into entity processing data distributed in a step-like block pattern.

[0094] In the structural model file generated by the discrete mapping module, a deformation compensation topology with thermal resistance and collapse resistance is generated for the loaded stress area. This deformation compensation topology is a response feature extracted by the inverse solution module based on the transient adjoint state method and objective function optimization, and is mainly achieved by adjusting the initial geometry of the local entity and the micro-foaming parameters.

[0095] The spatial distribution region of the deformation compensation topology on the insulated bag entity is defined. Based on the two-dimensional tensile principal strain field distribution calculated by the mechanical mapping module, the system identifies high-strain concentration areas in the load-bearing area on the bottom surface of the insulated bag, the normal extension area connecting the sidewalls and the lifting strap, and the edge seam area. These high-strain concentration areas bear the uniformly distributed load of the food container's weight and tensile forces under actual delivery and loading conditions, and are weak points where thermal resistance and heat loss are prone to non-uniform decay.

[0096] The initial thickness spatial shaping features of the deformation compensation topology are extracted. The system captures the sensitivity of high strain concentration areas to the target temperature deviation of fresh food preservation through high-dimensional sensitivity gradient analysis, and then superimposes the initial thickness increment in this area. In the unstressed state of the insulated bag under natural rest, the intermediate foamed insulation layer in the bottom bearing area and the side wall tension area exhibits a discrete thickened step structure that bulges inward or outward. When the takeout container is placed into the insulated bag, causing elastic tension of the material, these thickened steps are compressed and thinned. The actual working thickness after deformation falls back to the nominal insulation thickness required to maintain the target heat flux density, avoiding local heat dissipation cold bridges caused by thinning due to stress.

[0097] The foam porosity and density configuration characteristics of the deformation compensation topology were extracted. In the aforementioned high-strain concentration region, the deformation compensation topology was configured with the lowest level of discrete porosity distribution. This parameter configuration manifests macroscopically as higher material hardness and tensile stiffness in the foam layer of this region; microscopically, it manifests as a reduction in the diameter of closed-cell foam cells and an increase in the volume ratio of the polymer solid-phase skeleton. By increasing the intrinsic density of the solid material, this region can resist greater normal pressure and tangential stress, suppressing excessive thickness attenuation. For the determination of the intrinsic thermal conductivity and mechanical strength of polymer foam materials based on the closed-cell microstructure, those skilled in the art can use the steady-state flat plate heat flow meter method and a universal testing machine for testing. The testing standards and instrument operation are well-known techniques in this field and will not be elaborated here.

[0098] An equivalent local thermal resistance mathematical evaluation model for the deformation-compensated topology under loaded operating conditions is established. This evaluation model is used to verify the thermodynamic stability capability of the insulation bag under deformation failure conditions after the structure is finalized. The local thermal resistance function of the system under operating conditions is constructed as follows: ; in, Indicates spatial location Actual local thermal resistance under loading and stress conditions; This represents the discrete thickness distribution function of the intermediate gradient foamed insulation layer under stress-free conditions. This indicates the Poisson's ratio of the foamed material; Indicates spatial location The tensile principal strain field at the location; This represents the thickness compression coefficient under stress. Represents the discrete porosity distribution function The fundamental thermal conductivity is determined by this factor. This represents the local thermal resistance reduction matrix affected by deformation.

[0099] The numerator of the equation represents the actual thickness of the material after deformation, while the denominator represents the actual equivalent thermal conductivity of the material after deformation. The discrete thickness is hard-coded and implanted through system inverse parameter optimization. With discrete porosity Even with the intervention of the multiplicative decay term, it can still ensure The calculated values ​​maintain a balanced evolution throughout the entire physical space domain, eliminating high-speed heat leakage channels caused by local forces, and achieving uniformity of thermal insulation performance across the entire domain under varying working conditions.

[0100] Specific application examples: The following section will elaborate on the workflow of the parameter optimization design method of this invention, using the scenario of chilled beef delivery under high-temperature conditions in summer.

[0101] Initial parameter input and boundary condition setting: Set target temperature for fresh food cooling The critical failure temperature is 4℃. The average delivery time is 2℃. The time is 45 minutes, the external ambient temperature is kept constant at 35℃, and the boundary generation module obtains the time-series speed data of the delivery rider's electric vehicle (frequently fluctuating between 0km / h and 40km / h).

[0102] Through conversion, the transient convective heat transfer coefficient on the windward side is approximately 5 W / (m²). 2 K) (when the traffic light is at a standstill) up to 45W / (m 2 The data oscillates violently between K (at full speed). This data generates a dynamic convection boundary matrix.

[0103] Load mapping and physical field evolution: The delivered item is a chilled fresh meat container weighing 1.5kg and with a base area of ​​20cm x 15cm. The mechanics mapping module calculates: The bottom bearing area of ​​the insulated bag experiences a maximum of 15% normal compressive strain, while the sidewall connected to the handle experiences 8% tangential tensile strain. These deformations cause the local thermal resistance reduction matrix in this area to drop to 0.65 (i.e., a nominal thermal resistance loss of 35%). Based on this, the forward solver module establishes a three-dimensional thermal conductivity control equation with property decay and calculates the space temperature field under this initial state.

[0104] Inverse optimization and discrete mapping output: The inverse algorithm module constructs a Lagrange functional that includes temperature deviation and material mass penalty terms to calculate the associated temperature field. The sensitivity analysis module reveals that the gradient sensitivity is extremely high on the windward side and the bottom bearing region. After 120 iterations of optimization, the objective function converges.

[0105] The discrete mapping module maps it to discrete specifications that can be manufactured in a factory, outputting the final structure: Windward side area: Discrete thickness distribution of 15mm (to resist high wind speed convection heat penetration).

[0106] Leeward side area: Discrete thickness reduced to 8mm (reducing unnecessary redundant material).

[0107] Bottom load-bearing area and side wall tensile zone (deformation compensation topology): A 12mm thickened step with an inward protrusion was designed, and the porosity was set to the lowest level of 60% (the porosity of the remaining unloaded areas is 85%). Due to the low porosity and high density, the thickness of this step remained at about 10mm after being subjected to a 1.5kg load, and the equivalent thermal resistance did not show significant collapse.

[0108] Experimental verification and effect comparison: To verify the actual technical effect of the parameter optimization design method of the present invention, a solid heat-insulating bag was prepared based on the system output parameters of the present invention, and a comparative test was conducted with the traditional design scheme.

[0109] Experimental sample preparation: Experimental group (optimized design group of this invention): The insulation bag is manufactured using parameters output by the discrete mapping of the system of this invention. The thickness of the insulation bag is 15mm on the windward side and 8mm on the leeward side. A 12mm deformation compensation thickened step with an inward protrusion and a porosity of 60% is provided in the bearing area of ​​the bottom surface; The porosity of the non-stressed area is 85%. The total mass of the foamed layer is 120g.

[0110] Control group (traditional uniform design group): Using a standard design, the thickness of the foam layer in the insulation bag remains constant at 10mm in all areas, with an overall porosity of 85%. The total material weight of the foam layer is also controlled to 120g.

[0111] Experimental testing conditions: The experimental group and the control group were placed in a walk-in artificial climate chamber for simultaneous testing.

[0112] The external environment parameters are set as follows: Temperature conditions: The ambient temperature in the climate chamber is kept constant at 35℃.

[0113] Load conditions: Each bag contains a rectangular temperature-measuring water bag with dimensions of 20cm×15cm and a total mass of 1.5kg. The initial core temperature of the water bag is 2℃, the target temperature for fresh food preservation is set at 2℃, and the critical temperature for safe failure is set at 4℃.

[0114] Dynamic convection conditions: A programmable wind tunnel was set up directly in front of the windward side of the insulated bag. The wind speed fluctuated periodically according to a sine law between 0 m / s and 11.1 m / s (about 40 km / h) to simulate the dynamic wind resistance of the delivery electric vehicle. The total test time was 60 minutes.

[0115] Experimental data recording: During the test, state changes were recorded using a high-precision heat flow meter patch, a laser displacement sensor, and an internal thermocouple matrix. The specific test data results are summarized in the table below.

[0116] Table 1. Comparison of measured results between the optimized design group of the present invention and the traditional uniform design group. in conclusion: Based on the measured data in Table 1, and in conjunction with the attached... Figure 4 To be continued Figure 6 Based on the intuitive graphical representation and the technical features of this invention, the following engineering conclusions can be drawn: As attached Figure 4 The comparison chart of bottom surface thickness collapse before and after full load is shown. The light gray bars represent the nominal thickness before the test, and the dark gray bars represent the actual working thickness after full load. It can be clearly seen that after loading a 1.5kg weight, the bottom surface of the control group with traditional uniform design is severely deformed under pressure, and the actual working thickness is sharply reduced from 10.0mm to 5.2mm. The dark bars are lower than the light bars, and the thermal resistance in this area is reduced, making it very easy to form a bottom heat exchange channel.

[0117] Conversely, the bottom surface of the experimental group, designed according to the present invention, features a deformation compensation step with low porosity, resulting in higher material compressive stiffness. Under load, the actual thickness remains at 9.8 mm, with an extremely low thickness collapse rate. This comparative result demonstrates that the present invention cuts off the localized heat leakage path caused by stress concentration, confirming the necessity of extracting the reduction matrix using the mechanical mapping module.

[0118] Combined with appendix Figure 5 The comparison chart of heat penetration under dynamic wind resistance conditions shows that the average heat flux density on the windward side of the control group reaches 28.5 W / m². 2 This indicates that under dynamic high wind speeds, a uniform thickness of 10mm cannot withstand high-intensity forced convection heat transfer.

[0119] The experimental group, through reverse optimization, tilted the material towards the region of high convective heat transfer coefficient. The dark gray column representing heat penetration was lower in this group than in the control group, with a value reduced to 16.2 W / m. 2 The overall heat penetration of the windward side decreased by 43.1%. This indicates that the present invention can perform boundary matching and thermal resistance enhancement according to the external dynamic wind resistance environment.

[0120] In the appendix Figure 6 The internal load insulation duration evolution curve provides an excellent dynamic representation. The horizontal dotted line represents the critical failure temperature of 4℃, while the solid black line with circles and the dashed black line with squares simulate the internal temperature evolution trajectories of the control and experimental groups, respectively. Under the constraint of a strictly equal total foaming material mass of 120g, the experimental group achieved a flatter temperature rise curve (dashed line) compared to the control group (solid line) by reducing the redundant thickness on the leeward side and transferring the saved material to the bottom stress zone and the windward flow zone. The crosshairs in the graph visually indicate that the time point for the temperature curve to break through the critical failure line was extended from 28 minutes in the traditional design to 48 minutes. This objective graph demonstrates that the Lagrange functional with a mass penalty term and the associated optimization algorithm constructed by the system can indeed adjust the discrete manufacturing parameters to the optimal spatial distribution state in a thermodynamic sense.

Claims

1. A method for optimizing the parameters of a gradient composite insulated bag for fresh food delivery, characterized in that, Includes the following steps: Acquire vehicle speed data and ambient temperature sequence to generate a dynamic convection boundary matrix; Obtain the geometric dimensions and load weight parameters of the food container to be delivered, solve the two-dimensional tensile strain field, and output the local thermal resistance reduction matrix and heat capacity reduction matrix; Initialize the spatial design variable field, receive the dynamic convection boundary matrix, the local thermal resistance reduction matrix, and the heat capacity reduction matrix to establish the continuous medium transient thermal conduction partial differential control equation, perform forward numerical integration on the continuous medium transient thermal conduction partial differential control equation, and store the spatial temperature field data; Based on the space temperature field data, an optimization objective function is defined, a Lagrange functional of the optimization objective function is constructed, and inverse time integration is performed to obtain the accompanying temperature field. Based on the space temperature field data and the accompanying temperature field, the sensitivity gradient is extracted, and the space design variable field is updated until the optimization objective function satisfies the convergence threshold. The spatial design variable field that satisfies the convergence threshold is mapped to discrete manufacturing parameters, and the thermal insulation bag structure design parameters are output.

2. The parameter optimization design method for a gradient composite insulated bag for fresh food delivery according to claim 1, characterized in that, The step of acquiring vehicle driving time-series speed data and ambient temperature sequence to generate a dynamic convection boundary matrix includes: The vehicle driving time-series speed data is converted into transient Reynolds numbers; The transient convective heat transfer coefficient time series is calculated based on the ambient temperature sequence, the transient Reynolds number, the air Prandtl number, and the correlation of the convective heat transfer criterion number. The ambient temperature sequence and the transient convective heat transfer coefficient time sequence are aligned in time steps and combined to generate the dynamic convective boundary matrix.

3. The parameter optimization design method for a gradient composite insulated bag for fresh food delivery according to claim 1, characterized in that, The steps of obtaining the geometric dimensions and load weight parameters of the food container to be delivered, solving the two-dimensional tensile strain field, and outputting the local thermal resistance reduction matrix and heat capacity reduction matrix include: The geometric dimensions of the food box to be delivered are mapped to the boundary constraints of the internal space force contact surface; The load weight parameters are converted into uniformly distributed gravity loads and normal pressures on the contact surfaces acting on the bottom and sidewalls. The two-dimensional tensile strain field is solved based on the boundary constraints of the stressed contact surface, the uniformly distributed gravity load, and the normal pressure on the contact surface. By combining the Poisson's ratio of the insulation bag wall material and the preset pore compression empirical function, the two-dimensional tensile strain field is transformed into a physical property attenuation coefficient, thereby generating the local thermal resistance reduction matrix and the thermal capacity reduction matrix.

4. The parameter optimization design method for a gradient composite insulated bag for fresh food delivery according to claim 1, characterized in that, The steps of initializing the spatial design variable field, receiving the dynamic convection boundary matrix, the local thermal resistance reduction matrix, and the heat capacity reduction matrix, establishing the transient thermal conduction partial differential control equation for the continuous medium, performing forward numerical integration on the transient thermal conduction partial differential control equation for the continuous medium, and storing the spatial temperature field data include: Based on the aforementioned spatial design variable field, the basic thermal conductivity function and the basic volumetric specific heat capacity function are obtained; The equivalent thermal conductivity is obtained by multiplying the basic thermal conductivity function with the local thermal resistance reduction matrix. The equivalent volumetric specific heat capacity is obtained by multiplying the basic volumetric specific heat capacity function by the heat capacity reduction matrix; The equivalent thermal conductivity and the equivalent volumetric specific heat capacity are embedded as spatial variation coefficients into the transient thermal conductivity partial differential control equation of the continuous medium. Apply convective heat transfer boundary conditions defined by the dynamic convection boundary matrix at the outer boundary of the overall three-dimensional spatial domain and perform the forward numerical integration.

5. The parameter optimization design method for a gradient composite insulated bag for fresh food delivery according to claim 4, characterized in that, The steps of defining an optimization objective function based on the aforementioned space temperature field data and constructing the Lagrange functional of the optimization objective function include: Calculate the deviation between the spatial temperature field data and the target temperature for fresh food preservation; Construct a time integral term for the internal cavity temperature deviation that characterizes the cumulative degree of deviation between the actual temperature and the target temperature for fresh food preservation; The total material consumption is obtained by integrating the basic density of the material within the spatial domain of the solid structure of the insulation bag wall. A quality penalty weighting coefficient is introduced to scale the total material consumption to generate a total material consumption penalty term; The optimization objective function is constructed by adding the time integral term of the internal cavity temperature deviation to the total material consumption penalty term; The Lagrange functional is constructed by combining the transient thermal conductivity partial differential control equation of the continuous medium, which includes the equivalent thermal conductivity and the equivalent volumetric specific heat capacity, with the objective function as the state constraint.

6. The parameter optimization design method for a gradient composite insulated bag for fresh food delivery according to claim 5, characterized in that, The step of performing inverse time integration to obtain the accompanying temperature field includes: Solving the variational equation for the temperature state variable using the Lagrange functional yields the transient adjoint thermal conduction partial differential equation. The partial derivative of the optimization objective function with respect to the temperature field state variables is set as the adjoint source term of the transient adjoint heat conduction partial differential equation; Force the associated temperature variable value in the entire spatial domain to be set to zero at the delivery terminal time as the inverse integration start state; The transient accompanying thermal conduction partial differential equation is solved in the reverse direction of the time axis to obtain the accompanying temperature field.

7. The parameter optimization design method for a gradient composite insulated bag for fresh food delivery according to claim 6, characterized in that, The step of extracting the sensitivity gradient based on the space temperature field data and the accompanying temperature field, and updating the space design variable field, includes: Calculate the analytical functions of the partial derivatives of the equivalent thermal conductivity and the equivalent volumetric specific heat capacity with respect to the spatial design variable field; The sensitivity gradient is generated by multiplying the time rate of change and spatial gradient vector of the spatial temperature field data with the state value and spatial gradient vector of the accompanying temperature field. Define the boundary constraints of the spatial design variable field; The nonlinear optimization solver is invoked to iteratively update the spatial design variable field by combining the sensitivity gradient with the boundary constraints.

8. The parameter optimization design method for a gradient composite insulated bag for fresh food delivery according to claim 1, characterized in that, The step of updating the spatial design variable field until the optimization objective function satisfies the convergence threshold includes: The updated spatial design variable field is used as the basis for recalculating the total optimization objective function value under the current state; Calculate the absolute value of the difference between the total optimization objective function value and the previous state, and divide it by the previous baseline state to obtain the relative decrease scalar. When the relative decrease scalar is less than the convergence threshold, the cyclic iteration channel is determined to meet the convergence condition.

9. The parameter optimization design method for a gradient composite insulated bag for fresh food delivery according to claim 1, characterized in that, The step of mapping the spatial design variable field that satisfies the convergence threshold to discrete manufacturing parameters and outputting the thermal insulation bag structural design parameters includes: Based on the contour distribution characteristics and spatial gradient change rate of the continuous spatial design variable field, the overall entity space domain is divided into multiple non-overlapping sub-regions. For each of the divided sub-regions, the integral error between the distribution of continuous variables and the preset candidate parameter set is calculated, and a local mapping cost function is constructed. The discrete specification that minimizes the integral of the local mapping cost function is selected as the local optimal discrete specification of the sub-region. The locally optimal discrete specifications of all the sub-regions are summarized to generate the discrete manufacturing parameters that include the discrete thickness distribution and discrete porosity distribution of each sub-region.

10. A gradient composite insulated bag for fresh food delivery, designed by the parameter optimization design method for a gradient composite insulated bag for fresh food delivery as described in any one of claims 1-9, characterized in that, include: The heat-insulating bag body (1) has a wall structure that includes an outer protective layer (2), a middle gradient foamed insulation layer (3), and an inner food-grade contact layer (4) from the outside to the inside. The thickness of the intermediate gradient foamed insulation layer (3) has a high value set in the windward area of ​​the insulation bag body (1) and a low value set in the leeward area of ​​the insulation bag body (1). The intermediate gradient foamed insulation layer (3) has a deformation compensation topology in the bottom bearing area and side wall stretching area of ​​the insulation bag body (1). The deformation compensation topology is manifested as a discrete thickened step structure that protrudes inward or outward, and the porosity value corresponding to the deformation compensation topology is lower than the porosity value of the top and free side wall area of ​​the insulation bag body (1) that is not under force.

Images