摘要
为改进车载数据采集及分析主机中GPU的散热器结构设计,提高系统主机在高工作频率下的散热效果,采用机器学习进行仿真结果预测,得到优选的结构参数组,加快模型仿真效率,探究不同参数对散热效果的影响。首先在Ansys软件中导入散热器的几何模型,根据流体动力学建立热仿真模型,对模型进行热仿真,确定各部件的温度分布情况得到温度仿真数据。然后建立以支持向量机为基础框架结构,采用了灰狼优化算法更新种群策略的一种高效的模型,通过与实际仿真结果对比验证了仿真预测模型的准确性,同时能使仿真过程达到79.59%的约简。对散热器进行结构参数优化,如散热器位置、长度、宽度和高度以及翅片片数、厚度和高度,获得了降低GPU温度的较优结构参数取值。
In order to improve the radiator structure design of GPU in the vehicle inspection collection and analysis host,and improve the heat dissipation effect of the system host under high working frequency,machine learning was adopted to predict the simulation results,accelerate the simulation efficiency of the model,and explore the influence of different parameters on the heat dissipation effect.Firstly,the geometric model of radiator was imported into Ansys software,and the thermal simulation model was established according to fluid dynamics.After the pre-processing and post-processing of the model,the thermal simulation was carried out to determine the temperature distribution of each component and obtain the temperature simulation data.Then,an efficient model was established based on support vector machine and gray Wolf optimization algorithm was used to update the population strategy.The accuracy of the simulation prediction model is verified by comparing with the actual simulation results,and the simulation process can reach 79.59%reduction.By optimizing the structural parameters of the radiator,such as the position,length,width and height of the radiator,as well as the number,thickness and height of the fins,the optimal structural parameters to reduce the GPU temperature are obtained.
作者
陈锬
刘丹丹
刘宜胜
周在福
CHEN Tan;LIU Dan-dan;LIU Yi-sheng;ZHOU Zai-fu(Faculty of Mechanical Engineering and Automation,Zhejiang Sci-Tach Univercity,Hangzhou,Zhejiang 310018,China;College of Biomedical Engineering&Instrument Science,Zhejiang University,Hangzhou Zhejiang 310027,China;Academia Sinica,United Science&Technology Co.,Ltd.,Hangzhou Zhejiang 310051,China)
出处
《计算机仿真》
北大核心
2023年第10期187-192,共6页
Computer Simulation
基金
城市轨道交通自主驾驶DTO列车运行控制系统技术研究及应用(2019C01144)
城轨数字化关键技术研究与系统开发及示范应用(2022C01064)。
