Research On Collaborative Optimization Of Heat Flow In Micro Radiator For High Heat Flux Components

With the development of miniaturization and high power of electrical and electronic equipment,more strict requirements have been put forward for the heat dissipation capacity.The micro-radiator is becoming an important equipment widely used in the area of modern cooling.The heat transfer performance of micro-radiator is affected by many factors,and there is a complex coupling effect among these factors,which has not been expounded yet.This is not conducive to the exploration of the true principle of the flow heat transfer process in the micro-channel,nor to the realization of the multi-parameter collaborative optimization design of the micro-radiator.Therefore,how to reveal the internal correlation and improve the heat dissipation performance is a difficult problem to be solved.Therefore,attached to the NSFC project [52006058],this paper adopts the combination of numerical simulation method and Machine Learning(ML)to conduct collaborative optimization research on the heat flow in the micro radiator of high heat flux components from the perspective of data-driven.The main contents are introduced as follows:(1)Modeling and parameter sensitivity of flow and heat transfer processes in a typical microchannel were carried out.Firstly,the influence of three typical sections on the channel performance was analyzed,and the rectangular section was determined as the research object in the subsequent micro-radiator.In this paper,the influence of main structural parameters and key physical parameters of fluid medium on heat transfer of rectangular micro heat sink is explored,and Gradient Boosting Tree(GBT)is used to set up a fast mapping between parameter variables under study and two indicators: Nu and Pumping power.At the same time,the Sobol method was used in combination with GBT to quantify the influence of each study variable on the flow and heat transfer characteristics of microchannels from a global perspective.(2)The overall thermal performance evaluation and collaborative optimization of a Gallium Nitride(Ga N)high-electron-mobility Transistors(HEMTs)chip microradiator were studied.Combined with the temperature field and the velocity field,the effect of the transition central Angle β on the performance of the micro-radiator and the heat dissipation performance of different heat sources Q under the condition of structural change are analyzed emphatically.(3)Combined with the CFD calculation data set,the ML prediction model was established and the multi-objective optimization equation was determined.The classical difference algorithm in the GEATPY library was used to optimize the parameters of the multi-objective equation,and the optimal objective function value and the optimal control variable value were obtained.In the case of different weights,the thermal resistance and pressure drop were considered separately,and the optimal objective function value and the reference value were compared and analyzed.Finally,the optimized thermal resistance and pressure drop value could be obtained in reverse.