CN-122020920-A - Tesla valve type micro-channel heat sink optimal design method and Tesla valve type micro-channel heat sink optimal design system

CN122020920ACN 122020920 ACN122020920 ACN 122020920ACN-122020920-A

Abstract

The invention discloses a Tesla valve type micro-channel heat sink optimal design method and system, and belongs to the technical field of micro-channel heat dissipation. The non-uniform change of the top width of the valve core and the valve core height along the flow direction is introduced as an optimal design variable, the maximum number of Knoop and the minimum pressure drop of the system are taken as optimization targets, the optimal design variable is subjected to optimizing solution through a multi-target optimizing algorithm to obtain a pareto optimal solution set, and a scheme with the maximum relative closeness is selected from the pareto optimal solution set through an approaching ideal solution sequencing method to serve as a comprehensive optimal design scheme. The performance prediction agent model is constructed by using Latin hypercube sampling and numerical simulation, so that the rapid mapping between structural parameters and heat exchange performance is realized, the calculation cost in the optimization process is obviously reduced, the design efficiency is improved, the flow resistance is effectively controlled while the heat exchange capability is finally ensured, and the further improvement of the comprehensive heat dissipation performance is realized.

Inventors

YIN YING
ZHU YUNXIN
GONG LIANG
DUAN XINYUE
LIU SHUO
ZHAN WANLIN

Assignees

中国石油大学(华东)

Dates

Publication Date: 20260512
Application Date: 20260414

Claims (10)

1. A Tesla valve type micro-channel heat sink optimization design method is characterized by comprising the following steps: Obtaining an optimal design variable and a preset value range thereof, wherein the optimal design variable comprises a cooling medium inlet speed u in , and a valve core top width W t and a valve core height H which are unevenly changed along the flow direction in a Tesla valve type micro-channel heat sink; Inputting the optimal design variable and the preset value range thereof into a pre-constructed performance prediction proxy model, wherein the performance prediction proxy model is a back propagation neural network model, and is internally provided with a mapping relation between an input variable and an output performance index and is used for outputting a corresponding Nu number and a corresponding system pressure drop delta P according to the input valve core top width W t , the valve core height H and the input speed u in ; Optimizing and solving the optimization design variable by using the output Nu-Cork number to be maximized and the system pressure drop delta P to be minimized as optimization targets through a multi-target optimization algorithm to obtain a pareto optimal solution set; Comprehensively evaluating the pareto optimal solution set by an approximation ideal solution sorting method, taking each solution in the pareto optimal solution set as a candidate solution, calculating Euclidean distances from each candidate solution to a positive ideal solution and a negative ideal solution, calculating the relative closeness of each candidate solution according to the Euclidean distances, and selecting the candidate solution with the maximum relative closeness as a comprehensive optimal design solution; And outputting the valve core top width W t , the valve core height H and the inlet speed u in corresponding to the comprehensive optimal design scheme to serve as a final design result of the Tesla valve type micro-channel heat sink non-uniform valve core structure.
2. The tesla valve type micro-channel heat sink optimization design method according to claim 1, wherein the mapping relation between the input variable and the output performance index built in the back propagation neural network model is established by the following modes: The method comprises the steps of generating sample points in a preset value range of an optimal design variable by using a Latin hypercube sampling method, obtaining a Nu number and a system pressure drop delta P corresponding to each group of sample points through numerical simulation, constructing a sample database, and training the back propagation neural network model based on the sample database, wherein the numerical simulation adopts a periodic calculation model as an analysis object, the periodic calculation model is obtained by intercepting according to an actual microchannel integral model and is used for reflecting periodic flow and heat transfer characteristics in a complete microchannel, the number of the sample points generated by Latin hypercube sampling is 60, and the sample database is divided into a training set, a verification set and a test set and is used for training and performance evaluation of the back propagation neural network model.
3. The optimized design method of the Tesla valve type micro-channel heat sink according to claim 1, wherein the basic structure of the Tesla valve type micro-channel heat sink adopts an oval valve core and is arranged in an incremental mode along the flowing direction.
4. The tesla valve type micro-channel heat sink optimal design method according to claim 1, wherein the multi-objective optimization algorithm is a non-dominant ordering genetic algorithm II.
5. The tesla valve type micro-channel heat sink optimal design method according to claim 1, wherein the valve core top width W t is 0.20-0.60 mm, the valve core height H is 0.50-1.10 mm, and the inlet speed u in is 0.60-1.80 m/s.
6. The tesla valve type micro-channel heat sink optimal design method according to claim 1, wherein the back propagation neural network model is of a three-layer feedforward neural network structure, an input layer corresponds to a valve core top width W t , a valve core height H and an inlet speed u in , an output layer corresponds to a Nu number and a system pressure drop delta P, a hidden layer excitation function adopts a Tansig function, an output layer excitation function adopts a Purelin function, and a training algorithm adopts a Bayesian regularization algorithm.
7. The tesla valve type micro-channel heat sink optimal design method according to claim 1, wherein a positive ideal solution a + and a negative ideal solution a - satisfy: ; 。
8. The optimized design method of a Tesla valve type micro-channel heat sink according to claim 1, further comprising comparing the comprehensive optimal design scheme with the Tesla valve type micro-channel heat sink with uniform valve core size, wherein the comparison indexes comprise an average Nu number, a system pressure drop delta P and a comprehensive performance evaluation factor PEC.
9. A tesla valve type microchannel heat sink optimizing design system, characterized in that the optimizing design method according to any one of claims 1 to 8 is used for designing, comprising: The parameter input module is used for setting preset value ranges of the valve core top width W t , the valve core height H and the inlet speed u in ; The agent model storage module is in communication connection with the parameter input module and is used for storing a pre-constructed performance prediction agent model, the performance prediction agent model is a back propagation neural network model, and a mapping relation between an input variable and an output performance index is built in the performance prediction agent model and is used for outputting a corresponding Nu number and a system pressure drop delta P according to the input valve core top width W t , the valve core height H and the inlet speed u in ; The multi-objective optimization module is in communication connection with the agent model storage module, and is used for receiving the preset value range output by the parameter input module, calling a performance prediction agent model in the agent model storage module, optimizing and solving an optimization design variable by adopting a multi-objective optimization algorithm with the maximum number Nu of the Knoop and the minimum pressure drop delta P of the system as optimization targets to obtain a pareto optimal solution set; the decision evaluation module is in communication connection with the multi-objective optimization module and is used for receiving the pareto optimal solution set output by the multi-objective optimization module, comprehensively evaluating the pareto optimal solution set by adopting an approximation ideal solution sequencing method, and selecting a scheme with the maximum relative closeness as a comprehensive optimal design scheme; The result output module is in communication connection with the decision evaluation module and is used for receiving the comprehensive optimal design scheme output by the decision evaluation module and outputting the valve core top width W t , the valve core height H and the inlet speed u in corresponding to the comprehensive optimal design scheme.
10. The tesla valve type micro-channel heat sink optimal design system according to claim 9, wherein the proxy model storage module is constructed by adopting a Latin hypercube sampling method to generate sample points in a preset value range of an optimal design variable, obtaining a Nu number and a system pressure drop delta P corresponding to each group of sample points through numerical simulation, constructing a sample database, and training the back propagation neural network model based on the sample database.

Description

Tesla valve type micro-channel heat sink optimal design method and Tesla valve type micro-channel heat sink optimal design system Technical Field The invention relates to the technical field of micro-channel heat dissipation, in particular to a Tesla valve type micro-channel heat sink optimal design method and system. Background With the continuous improvement of chip integration level and power consumption density, the heat flow density of unit area is continuously increased, and the traditional micro-channel cooling mode with simple geometric configuration has difficulty in meeting the requirements of enhanced heat exchange and low-resistance operation. The tesla valve type micro-channel can induce fluid to generate periodic diversion, confluence and backflow disturbance under the condition of no moving part by virtue of the asymmetric geometric structure of the tesla valve type micro-channel, so that the heat exchange of a near wall surface is enhanced, and the tesla valve type micro-channel is regarded as a novel potential structure for heat dissipation of a high heat flux chip. However, the structural design of the existing tesla valve type micro-channel heat sink mostly adopts uniform valve core size configuration, and the possibility of changing the valve core size parameter along the flow direction is not fully considered, so that it is difficult to effectively control the flow resistance while improving the heat exchange capability. Meanwhile, obvious coupling effect exists between structural parameters of the micro-channel heat sink, an optimal scheme which takes heat exchange strengthening and resistance control into consideration cannot be obtained only by means of single factor analysis, and if high-precision numerical simulation is adopted to carry out direct exhaustive optimization on multi-parameter combination, the problems of large calculated amount, long optimization period and easiness in sinking into local optimum are faced. Disclosure of Invention The invention aims to provide a Tesla valve type micro-channel heat sink optimal design method and system for solving the problems. The technical scheme of the invention is as follows: a Tesla valve type micro-channel heat sink optimization design method comprises the following steps: And obtaining an optimal design variable and a preset value range thereof, wherein the optimal design variable comprises a cooling medium inlet speed u in, and a valve core top width W t and a valve core height H which are unevenly changed along the flow direction in the Tesla valve type micro-channel heat sink. And inputting the optimal design variable and the preset value range thereof into a pre-constructed performance prediction proxy model, wherein the performance prediction proxy model is a back propagation neural network model, and is internally provided with a mapping relation between an input variable and an output performance index and is used for outputting a corresponding Nu number and a system pressure drop delta P according to the input valve core top width W t, the valve core height H and the inlet speed u in. And optimizing and solving the optimization design variable through a multi-objective optimization algorithm by using the output Nu-Sai-Er number Nu as the maximum and the system pressure drop delta P as the optimization objective to obtain the pareto optimal solution set. And comprehensively evaluating the pareto optimal solution set by approaching an ideal solution sorting method, taking each solution in the pareto optimal solution set as a candidate solution, calculating Euclidean distances from each candidate solution to a positive ideal solution and a negative ideal solution, calculating the relative closeness of each candidate solution according to the Euclidean distances, and selecting the candidate solution with the maximum relative closeness as a comprehensive optimal design solution. And outputting the valve core top width W t, the valve core height H and the inlet speed u in corresponding to the comprehensive optimal design scheme to serve as a final design result of the Tesla valve type micro-channel heat sink non-uniform valve core structure. Further, the mapping relationship between the input variable and the output performance index built in the back propagation neural network model is established by the following modes: The method comprises the steps of generating sample points in a preset value range of an optimal design variable by using a Latin hypercube sampling method, obtaining a Nu number and a system pressure drop delta P corresponding to each group of sample points through numerical simulation, constructing a sample database, and training the back propagation neural network model based on the sample database, wherein the numerical simulation adopts a periodic calculation model as an analysis object, the periodic calculation model is obtained by intercepting according to an actual microchannel integral model and is us