An adjoint lattice Boltzmann topology optimization method targeting comprehensive performance goodness factor

Abstract

The present invention relates to the field of efficient heat dissipation technology, and discloses an adjoint lattice Boltzmann topology optimization method with a comprehensive performance goodness factor as the target, and solves the problem using the lattice Boltzmann method as the solver. The flow field and temperature field data are calculated by solving the lattice Boltzmann equation, and the velocity field and temperature field adjoint lattice Boltzmann equation are further solved to obtain their adjoint variables; the design sensitivity is solved, and the design variables are updated according to the design sensitivity to obtain the fluid-solid distribution of the current iteration step; the above steps are repeated until the value of the optimization objective function reaches stability. The comprehensive performance goodness factor proposed by the evaluation index of the porous skeleton is used as the optimization target for topology optimization, the temperature and the inlet and outlet pressure drop are multiplied, and the weight coefficient is placed on the inlet and outlet pressure drop index, thereby achieving strong coupling of the temperature field and the flow field, which has advantages over traditional targets in terms of optimization efficiency and optimization results, thereby reducing the cost of topology optimization calculations.

Description

Technical Field

[0001] The present invention relates to the field of efficient heat dissipation technology, and in particular to an adjoint lattice Boltzmann topology optimization method with a comprehensive performance goodness factor as a target. Background Art

[0002] Miniaturization, integration, high performance and scale of high-end, high-power, aerospace, high-precision devices, microelectronic devices, high-intensity lasers, etc. are inevitable trends in the development of social economy and science and technology. However, the heat dissipation problem caused by the continuous increase in the heat power of precision devices seriously restricts the safe and reliable operation of the system. Especially in the heat dissipation design of engines and electronic chips, improper design of cooling channels will cause serious overheating of parts and limit the performance and durability of devices. Therefore, in order to achieve the goal of high performance and low cost, it is necessary to optimize the design of efficient cooling systems to effectively remove more heat from confined spaces.

[0003] Traditional design methods, namely size / shape optimization, rely on the engineering experience of designers and often fail to achieve the optimal performance of the structure. Topology optimization (TO) design technology is based on mathematical optimization methods. By changing the size, shape or structural configuration of the engineering structure, it can achieve a new structural concept design with optimal performance under given constraints. In particular, the revolutionary progress of 3D printing technology has enabled topology optimization design technology to be applied to many cutting-edge scientific fields such as new compact heat dissipation structures and micro-nano structures. Topology optimization technology has the advantages of high design freedom and strong flexibility. Therefore, topology optimization technology can be applied to the flow channel shape design of cooling structures, thereby effectively solving the thermal management problem of high-heat flux and high-temperature devices.

[0004] The optimization goal in current research is usually to simply add the two together. This goal is more inclined to basic research and is not suitable for direct application in engineering. In fields such as aircraft thermal protection, porous skeletons with high thermal conductivity are usually added to the cooling channel to improve the heat transfer performance, and the indicator for evaluating the heat transfer capacity of the skeleton is usually the area goodness factor (AeraGoodness Factor). For the above optimization problems, traditional finite volume methods, finite element methods, or finite difference methods can be used to solve them, but this method requires complex processing of the grid when facing complex fluid-solid boundaries and has poor parallelism. After each update of the flow-solid distribution in topology optimization, the flow field and temperature field must be solved, and the computational cost is extremely high. Summary of the invention

[0005] The object of the present invention is to provide an adjoint lattice Boltzmann topology optimization method with a comprehensive performance goodness factor as a target in order to improve the problem of high computational cost of existing topology optimization.

[0006] The technical solution adopted by the present invention is as follows:

[0007] A method for adjoint lattice Boltzmann topology optimization with a comprehensive performance goodness factor as a target, the method uses a lattice Boltzmann method LBM as a solver for solving, and calculates flow field and temperature field data by solving the lattice Boltzmann equation. The method comprises the following steps: S100: determining a design domain and boundary conditions; S200: discretizing the entire design domain, and characterizing the fluid as 1 and the solid as 0; initializing the entire design domain to 1; S300: solving the flow field lattice Boltzmann equation LBE to obtain flow field data; S400: solving the temperature field lattice Boltzmann equation LBE to obtain temperature field data; S500: solving the velocity field adjoint lattice Boltzmann equation LBE according to the data obtained in S300 and S400 to obtain velocity field adjoint variables ; S600: Solve the temperature field adjoint lattice Boltzmann equation LBE to obtain the temperature field adjoint variable ; S700: solve the design sensitivity according to the data obtained from S300 to S600; S800: update the design variables according to the obtained design sensitivity to obtain the fluid-solid distribution of the current iteration step; S900: repeat S200 to S700 until the objective function is optimized J value reaches stability.

[0008] Furthermore, the flow field data includes velocity and pressure, and the flow field lattice Boltzmann equation LBE is:

[0009] ;

[0010] ;

[0011] in, is the lattice Boltzmann equation for the flow field, is the discrete velocity direction, is the ratio of the grid step size to the time step size, is the weight coefficient, is the density, For speed, is the velocity distribution function, is the initial density, It's relaxation time. is the velocity equilibrium distribution function.

[0012] Furthermore, the temperature field lattice Boltzmann equation LBE is:

[0013] ;

[0014] in, is the lattice Boltzmann equation for the temperature field, is the grid time, is the grid direction, is the ratio of the grid step size to the time step size, is the temperature relaxation time, is the temperature distribution function, is the temperature equilibrium distribution function;

[0015] ;

[0016] in, is the grid direction, is the weight coefficient, is the temperature, is the ratio of the grid step size to the time step size, For speed, is the temperature equilibrium distribution function.

[0017] Furthermore, the velocity field is accompanied by a variable for:

[0018] ;

[0019] in, is the grid direction, is the grid time, For relaxation time, is the gradient operator, and is the equilibrium distribution function and source term of the velocity adjoint variable derived from the KKT condition, is the velocity adjoint variable, is the ratio of the grid step size to the time step size.

[0020] Furthermore, the temperature field is accompanied by a variable for:

[0021] ;

[0022] ;

[0023] in, is the grid direction, is the gradient operator, is the adjoint source term, is the temperature distribution function, is the temperature field adjoint variable, is the grid time, is the ratio of the grid step size to the time step size, is the temperature relaxation time. is the equilibrium distribution function of the temperature adjoint variable, is the time for the positive problem to reach steady state, To optimize the objective function.

[0024] Furthermore, the design sensitivity is:

[0025] ;

[0026] in, is the control equation of the positive problem, is the lattice Boltzmann equation for the flow field, is the lattice Boltzmann equation for the temperature field, is the design variable, is the adjoint variable of the velocity field, is the temperature field adjoint variable, is the volume constraint adjoint variable.

[0027] Furthermore, the fluid-solid distribution of the current iteration step is:

[0028] ;

[0029] in, is the level set function, For virtual time, is the update step size, Design sensitivity.

[0030] Furthermore, in the solutions of the lattice Boltzmann equation LBE and the adjoint lattice Boltzmann equation LBE of the flow field and the temperature field in S400 to S600, the core collision part and the migration step solution are accelerated by combining the graphics processor GPU, and each process of the graphics processor GPU corresponds to each step.

[0031] Furthermore, the form of the adjoint lattice Boltzmann equation LBE in S500 and S600 is closely related to the objective function. The optimization objective function obtained by S900 is J The optimization target is obtained by substituting the optimization target into the boundary formula of the adjoint equation, and the adjoint equation boundaries corresponding to the traditional optimization target and the newly proposed optimization target are obtained as follows, taking the entrance as an example:

[0032]

[0033]

[0034] in, is the weight coefficient of traditional goals and new goals, is the grid direction, is the velocity distribution function, is the temperature distribution function, is the design domain boundary, is the velocity adjoint variable distribution function at the inlet boundary, is the entrance boundary, is the heat flow boundary.

[0035] Among them, the boundary formula of the adjoint equation is:

[0036]

[0037] Among them, the boundary formula of the adjoint equation is the existing formula, is the differential symbol, , and is the velocity distribution function at the inlet boundary, , , , , and is the velocity adjoint variable distribution function at the inlet boundary, , and is the temperature adjoint variable distribution function at the inlet boundary, J is the objective function, is the grid time, is the time after the positive problem stabilizes, , and is the velocity boundary format for the positive problem, , and is the temperature boundary format for the direct problem, is the entrance boundary.

[0038] Furthermore, the design sensitivity Set to 0.1.

[0039] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0040] The present invention performs topology optimization by taking the comprehensive performance goodness factor proposed by the evaluation index of the porous skeleton as the optimization target. In essence, the temperature and the inlet and outlet pressure drops are directly multiplied, and the weight coefficient is placed on the inlet and outlet pressure drop index, thereby achieving strong coupling of the temperature field and the flow field. While having greater engineering practical value, it has certain advantages over traditional targets in terms of optimization efficiency and optimization results, thereby reducing the cost of topology optimization calculations.

[0041] The present invention uses the lattice Boltzmann method as a solver to solve the problem, and greatly improves the efficiency of topology optimization based on CUDA combined with GPU acceleration technology, thereby improving the heat dissipation efficiency to a certain extent. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 is a flow chart of the method of the present invention;

[0043] Figure 2 It is a schematic diagram of the design domain of the present invention;

[0044] Figure 3 It is a calculation flow chart of the present invention;

[0045] Figure 4 This is a comparison chart of GPU accelerated computing efficiency of the present invention;

[0046] Figure 5 (a) shows the convergence speed of the adjoint field of the traditional target, and Figure 5 (b) shows the convergence speed of the adjoint field of the new target;

[0047] Figure 6 The Pareto frontier corresponding to the traditional goal and the new goal of the present invention;

[0048] in, is the inlet and outlet pressure drop, is the heat flow surface temperature;

[0049] Figure 7(a) shows the traditional target optimization result, and Figure 7(b) shows the new target optimization result. DETAILED DESCRIPTION

[0050] The present invention will be described in detail below in conjunction with the accompanying drawings.

[0051] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0052] Topology optimization technology has the advantages of high design freedom and strong flexibility. Therefore, topology optimization technology can be applied to the flow channel shape design of cooling structures, thereby effectively solving the thermal management problems of high-temperature devices with high heat flux density.

[0053] A lattice Boltzmann topology optimization method with comprehensive performance goodness factor as the target is proposed. The performance goodness factor is used as the optimization target (Eq(2)), and the GPU acceleration technology is combined to realize the efficient topology optimization of the flow-heat transfer coupling problem. The optimized configuration with this parameter as the target can be directly applied to practical engineering applications. The weight factor is used as the optimization target. Focusing on the inlet and outlet pressure drop index, the final optimized configuration can be regulated by adjusting the weight coefficient;

[0054] (1);

[0055] (2);

[0056] in, and They are the old optimization target and the new optimization target respectively. is the heat flow boundary, is the temperature, and is the weight factor, is the inlet and outlet pressure drop. is the density, It's speed. and are the characteristic length and characteristic height, q is the heat flux density, Re and Pr are the Reynolds number and the Planck number, and is the heat flow surface temperature and inlet temperature. When the working conditions are fixed, , ,in, is the inlet pressure, is the outlet pressure.

[0057] It can be concluded from formula (1) and formula (2) that, compared with the old optimization objective, the new optimization objective strengthens the coupling effect of velocity and temperature in the adjoint field solution, and has a certain improvement in optimization efficiency and optimization results compared with the commonly used optimization objectives. Therefore, the optimization objective has a profound impact on the adjoint field boundary form and gradient field solution in topology optimization.

[0058] Example 1

[0059] like Figure 1-Figure 3 As shown, another embodiment of the present invention is a method for adjoint lattice Boltzmann topology optimization with comprehensive performance goodness factor as the target, the method uses lattice Boltzmann method LBM as the solver for solving, and calculates flow field and temperature field data by solving lattice Boltzmann equation, the method comprises the following steps: S100: determining the design domain and boundary conditions, finding the optimal force transmission path (i.e. the optimal material distribution method) in the design area, so as to effectively transfer the load to the support, and optimizing the shape of the structure by changing the coordinates of the boundary points of the structure; S200: establishing a mathematical model, discretizing the entire design domain, and characterizing the fluid as 1 and the solid as 0; initializing the entire design domain to 1; S300: solving the flow field lattice Boltzmann equation LBE, and obtaining flow field data, including velocity, pressure, etc.; S400: solving the temperature field lattice Boltzmann equation LBE, and obtaining temperature field data; S500: solving the velocity field adjoint lattice Boltzmann equation LBE according to the data obtained in S300 and S400, and obtaining velocity field adjoint variables ; S600: Solve the temperature field adjoint lattice Boltzmann equation LBE to obtain the temperature field adjoint variable ; S700: solve the design sensitivity according to the data obtained from S300 to S600; S800: update the design variables according to the design sensitivity obtained from S700 to obtain the fluid-solid distribution of the current iteration step; S900: repeat S200 to S700 until the objective function is optimized J value reaches stability.

[0060] This embodiment uses the lattice Boltzmann method as a solver to solve the problem. Based on CUDA, which is a parallel computing platform and programming model, combined with GPU acceleration technology, the efficiency of topology optimization is greatly improved, thereby improving the heat dissipation efficiency to a certain extent.

[0061] Example 2

[0062] Another embodiment of the present invention is an adjoint lattice Boltzmann topology optimization method with a comprehensive performance goodness factor as a target, wherein the flow field data of S300 includes velocity and pressure, etc., and the flow field lattice Boltzmann equation LBE is:

[0063] ;

[0064] ;

[0065] in, is the lattice Boltzmann equation for the flow field, is the discrete velocity direction, is the ratio of the grid step size to the time step size, is the weight coefficient, is the density, For speed, is the velocity distribution function, is the initial density, It's relaxation time. is the velocity equilibrium distribution function.

[0066] The traditional existing methods use traditional finite volume method, finite element method or finite difference method for solution, but this method needs to process the grid in a complex way when facing complex fluid-solid boundary and has poor parallelism. In addition, the flow field must be solved after each update of the fluid-solid distribution in topology optimization, so the computational cost is extremely high. The embodiment uses the lattice Boltzmann method (LBM) as the solver for solution, and the fluid-solid wall is directly processed using the rebound boundary. Since LBM has the characteristics of local operation, it is very compatible with the single instruction multi-threaded architecture and GPU. Therefore, this study combines GPU for acceleration, thereby reducing the computational cost and achieving efficient optimization.

[0067] Example 3

[0068] Another embodiment of the present invention is a method for adjoint lattice Boltzmann topology optimization with comprehensive performance goodness factor as the target, wherein the temperature field data in S400 is solved according to the temperature field lattice Boltzmann equation LBE,

[0069] ;

[0070] in, is the lattice Boltzmann equation for the temperature field, is the grid time, is the grid direction, is the ratio of the grid step size to the time step size, is the temperature relaxation time, is the temperature distribution function, is the temperature equilibrium distribution function;

[0071] The corresponding relaxation times for fluids and solids are as follows:

[0072] ;

[0073] in, is the grid direction, is the weight coefficient, is the temperature, is the ratio of the grid step size to the time step size, For speed.

[0074] The traditional existing methods use traditional finite volume method, finite element method or finite difference method for solution, but this method needs to process the grid in a complex way when facing the complex fluid-solid boundary and has poor parallelism. The temperature field must be solved after each update of the fluid-solid distribution in topology optimization, so the computational cost is extremely high. The embodiment uses the lattice Boltzmann method (LBM) as the solver for solution, and the fluid-solid wall is directly processed using the rebound boundary. Since LBM has the characteristics of local operation, it is very compatible with the single instruction multi-threaded architecture and GPU. Therefore, this study combines GPU for acceleration, thereby reducing the computational cost and achieving efficient optimization.

[0075] like Figure 4 As shown in the figure, the computation time is reduced by nearly an order of magnitude when combined with GPU.

[0076] Example 4

[0077] like Figure 1As shown, another embodiment of the present invention is a method for topological optimization based on a combined performance factor. The core of the method is to use the LBE to solve the velocity in the flow field data, and calculate the design sensitivity based on the velocity and pressure, and then achieve topological optimization. The flow field data includes velocity and pressure. The velocity field LBE is solved based on the data of S300 and S400. The velocity field variable for:

[0078] ;

[0079] in, is the grid direction, is the grid time, For relaxation time, is the gradient operator, and is the equilibrium distribution function and source term of the velocity adjoint variable derived from the KKT condition, is the velocity adjoint variable, is the ratio of the grid step size to the time step size.

[0080] By accompanying variables Solving the design sensitivity, and the application of the adjoint method significantly improves the computational efficiency. Traditional topology optimization methods often require multiple iterations to calculate flow field data, while the adjoint method can obtain sensitivity information about design variables by solving the adjoint equation once, greatly reducing the amount of calculation. The solution process of adjoint LBE utilizes part of the information in the original LBE calculation, avoiding repeated calculations, further improving the computational efficiency, and reducing the cost of calculations.

[0081] At the same time, the introduction of adjoint variables (such as adjoint equilibrium distribution function and source term) makes the sensitivity calculation more accurate. These adjoint variables are derived based on KKT conditions, ensuring mathematical rigor and accuracy; by accurately solving the velocity field adjoint LBE, a more accurate velocity field distribution can be obtained, and then a more accurate design sensitivity can be obtained, providing a reliable basis for topology optimization.

[0082] Example 5

[0083] Another embodiment of the present invention is a method for adjoint lattice Boltzmann topology optimization with comprehensive performance goodness factor as the target, which solves the temperature field adjoint variable according to the temperature field data obtained in S400. :

[0084] ;

[0085] ;

[0086] in, is the grid direction, is the gradient operator, is the adjoint source term, is the temperature distribution function, is the grid time, is the ratio of the grid step size to the time step size, is the temperature relaxation time. is the equilibrium distribution function of the temperature adjoint variable, is the time when the positive problem reaches steady state, To optimize the objective function.

[0087] The adjoint lattice Boltzmann equation (LBE) is used to solve the temperature in the flow field data, and the design sensitivity is calculated based on this to achieve topology optimization. This adjoint method is not only applicable to the optimization of velocity field, but can also be extended to the optimization of other flow field parameters such as temperature field, and has wide applicability.

[0088] As a general numerical method, the adjoint lattice Boltzmann method can handle complex geometries and boundary conditions and provide solutions to various practical engineering problems.

[0089] The adjoint method has broad application prospects in the engineering field. For example, in the fields of aerospace, automotive engineering, energy, etc., the flow characteristics of the fluid are crucial to the performance of the product. By using this method to perform topological optimization, more efficient and energy-saving products can be designed. In addition, the adjoint method can also be applied to microfluidic chips, biomedical engineering and other fields, providing strong support for research and development in these fields.

[0090] Figure 5 (a) shows the convergence speed of the adjoint field of the traditional objective, and Figure 5 (b) shows the convergence speed of the adjoint field of the new objective. The adjoint field of the new objective only needs 5400 steps to converge, while the traditional objective needs 10100 steps, which shows that the proposed new objective can effectively shorten the convergence time and improve the optimization efficiency.

[0091] The convective heat transfer coefficient in the study is as follows:

[0092] ;

[0093] So the heat flux q Heat flow surface temperature under fixed conditions The smaller it is, the larger the corresponding convective heat transfer coefficient h is.

[0094] Figure 6 shows the Pareto frontier (Pareto frontier refers to the set of all possible optimal solutions in a multi-objective decision-making problem, which achieves the best balance between multiple objectives) when the fluid volume constraint is 50%. In the figure, the closer the Pareto frontier is to the lower left corner, the better the optimized configuration is. Therefore, the corresponding inlet and outlet pressure drop is smaller, and the heat flow surface temperature is lower (the heat transfer coefficient is higher). It can be found that the optimization effect of the new objective is superior to that of the traditional objective.

[0095] Example 6

[0096] like Figure 1 As shown, another embodiment of the present invention is an adjoint lattice Boltzmann topology optimization method with a comprehensive performance goodness factor as a target, and the design sensitivity is calculated according to the adjoint variables of the velocity field and the adjoint variables of the temperature field:

[0097] ;

[0098] in, is the control equation of the positive problem, is the lattice Boltzmann equation for the flow field, is the lattice Boltzmann equation for the temperature field, is the design variable, is the adjoint variable of the velocity field, is the temperature field adjoint variable, is the volume constraint adjoint variable.

[0099] Since the velocity field and the temperature field are coupled and influence each other, more comprehensive and accurate optimization results can be obtained by considering multiple physical fields at the same time. And the accuracy of the design sensitivity can more accurately reflect the influence of the design variables on the objective function, thereby guiding the optimization process in the right direction and improving the accuracy of the optimization results. When calculating the design sensitivity, the adjoint method takes into account the influence of multiple physical fields, which reduces the sensitivity of the optimization process to the initial conditions and boundary conditions and improves the robustness of the optimization process; in addition, the adjoint method can also reduce the influence of numerical errors on the optimization results to a certain extent, further improving the stability and reliability of the optimization process.

[0100] Example 7

[0101] Another embodiment of the present invention is an adjoint lattice Boltzmann topology optimization method with a comprehensive performance goodness factor as a target, wherein the fluid-solid distribution of the current iteration step is:

[0102] ;

[0103] in, is the level set function, For virtual time, For the update step size, it can usually be chosen as 0.1. Design sensitivity.

[0104] In the adjoint lattice Boltzmann topology optimization method, each iteration adjusts the design variables (such as the comprehensive performance goodness factor) according to the current optimization objectives and constraints, and recalculates the flow field (such as the velocity field, temperature field) and adjoint variables. Fluid-solid distribution refers to the distribution position of the fluid and solid in the design domain after each iteration, that is, the corresponding optimized topological structure.

[0105] The fluid-solid distribution of the current iteration provides a clear guide for the next iteration. By observing and analyzing the fluid-solid distribution, we can understand which parts of the structure or material have important contributions to the optimization target, so that these parts can be retained or strengthened in subsequent iterations. This saves computing resources and improves optimization efficiency.

[0106] Example 8

[0107] Another embodiment of the present invention is a lattice Boltzmann topology optimization method with a comprehensive performance goodness factor as a target. In the steps S400 to S600, in solving the lattice Boltzmann equation LBE of the flow field and the temperature field and the lattice Boltzmann equation LBE of the adjoint, the core collision part and the migration step solution are accelerated by combining the graphics processing unit GPU, and each process of the graphics processing unit GPU corresponds to each step.

[0108] Figure 7 (a) shows the traditional target optimization result, and Figure 7 (b) shows the new target optimization result. This figure is when the fluid volume constraint is 50% and the inlet and outlet pressure drops are the same ( ), the heat flow surface temperatures corresponding to the traditional target and the new target are 19.39 and 18.71 respectively. The new target has a lower temperature, which means it has a stronger heat transfer capacity.

[0109] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for adjoint lattice Boltzmann topology optimization with comprehensive performance goodness factor as the target, characterized in that: The method uses the lattice Boltzmann method LBM as a solver to solve the flow field and temperature field data by solving the lattice Boltzmann equation. The method comprises the following steps: S100: Determine the design domain and boundary conditions; S200: Discretize the entire design domain and characterize the fluid as 1 and the solid as 0; initialize the entire design domain to 1; S300: Solve the flow field lattice Boltzmann equation LBE to obtain flow field data; S400: Solve the temperature field lattice Boltzmann equation LBE to obtain temperature field data; S500: Solve the velocity field adjoint lattice Boltzmann equation LBE based on the data obtained from S300 and S400 to obtain the velocity field adjoint variables ; S600: Solve the temperature field adjoint lattice Boltzmann equation LBE and obtain the temperature field adjoint variable ; S7 00: Solve the design sensitivity based on the data obtained from S300 to S600; S800: updating the design variables according to the obtained design sensitivity to obtain the fluid-solid distribution of the current iteration step; S900: Repeat S200 to S700 until the objective function is optimized J The value of reaches stability; The flow field data includes velocity and pressure, and the flow field lattice Boltzmann equation LBE is: ； ； in, is the lattice Boltzmann equation for the flow field, is the discrete velocity direction, is the ratio of the grid step size to the time step size, is the weight coefficient, is the grid time, is the density, For speed, is the velocity distribution function, is the initial density, is the velocity relaxation time, is the velocity equilibrium distribution function; The temperature field lattice Boltzmann equation LBE is: ； in, is the lattice Boltzmann equation for the temperature field, is the grid time, is the grid direction, is the ratio of the grid step size to the time step size, is the temperature relaxation time, is the temperature distribution function, is the temperature equilibrium distribution function; ； in, is the grid direction, is the weight coefficient, is the temperature, is the ratio of the grid step size to the time step size, For speed, is the temperature equilibrium distribution function; The velocity field is accompanied by variables for: ； in, is the grid direction, is the grid time, For relaxation time, is the gradient operator, and is the velocity adjoint variable equilibrium distribution function and source term derived from the KKT condition, is the adjoint variable of the velocity field, is the ratio of the grid step to the time step; The temperature field is accompanied by variables for: ； ； in, is the grid direction, is the gradient operator, is the adjoint source term, is the temperature distribution function, is the temperature field adjoint variable, is the grid time, is the ratio of the grid step size to the time step size, is the temperature relaxation time, is the equilibrium distribution function of the temperature adjoint variable, is the time after the positive problem stabilizes, To optimize the objective function.

2. The adjoint lattice Boltzmann topology optimization method with comprehensive performance goodness factor as the target according to claim 1, characterized in that: The design sensitivity is: ； in, is the positive problem equation, is the lattice Boltzmann equation for the flow field, is the lattice Boltzmann equation for the temperature field, is the design variable, is the adjoint variable of the velocity field, is the temperature field adjoint variable, is the volume constraint adjoint variable.

3. The adjoint lattice Boltzmann topology optimization method with comprehensive performance goodness factor as the target according to claim 1, characterized in that: The fluid-solid distribution of the current iteration step is: ； in, is the level set function, For virtual time, is the update step size, Design sensitivity.

4. The adjoint lattice Boltzmann topology optimization method with comprehensive performance goodness factor as the target according to claim 1, characterized in that: In the above-mentioned S400 to S600, in solving the lattice Boltzmann equation LBE of the flow field and the temperature field and the adjoint lattice Boltzmann equation LBE, the core collision part and the migration step solution are accelerated by combining the graphics processing unit GPU, and each process of the graphics processing unit GPU corresponds to each step.

5. The adjoint lattice Boltzmann topology optimization method with comprehensive performance goodness factor as the target according to claim 1, characterized in that: The form of the adjoint lattice Boltzmann equation LBE in S500 and S600 is closely related to the objective function. The optimization objective function obtained by S900 is J The optimization target is obtained by taking the value of , and the optimization target is brought into the boundary formula of the adjoint equation to obtain the adjoint equation boundaries corresponding to the traditional optimization target and the newly proposed optimization target as follows: in, is the weight coefficient of traditional goals and new goals, is the grid direction, is the velocity distribution function, is the temperature distribution function, is the design domain boundary, is the velocity adjoint variable distribution function at the inlet boundary, is the entrance boundary, is the heat flow boundary.

6. The adjoint lattice Boltzmann topology optimization method targeting the comprehensive performance goodness factor according to claim 3, characterized in that: The design sensitivity Set to 0.1.

Images