摘要
为确定铁路风屏障的最优参数,基于代理模型方法对风屏障防风效果进行了优化.首先,改进了网格搜索法以优化支持向量机回归模型的参数,并通过算例进行了验证.其次,以设置风屏障时的车辆气动特性为目标函数,建立了风屏障防风效果的优化模型.最后,利用风屏障的风洞试验结果,采用支持向量机回归建立了目标函数的代理模型,对风屏障的高度和透风率进行了优化.研究结果表明:改进的网格搜索法提高了支持向量机模型参数选择的准确性,当风屏障高度为1.91~2.90m时,最优的风屏障透风率为0.00~0.17;当风屏障高度超过2.50m后,增加风屏障的高度对防风效果的提高较为有限.
To determine the optimal parameters of railway wind barriers, the protective effects of wind barriers were optimized by surrogate model methods. Firstly, the grid search method was improved to optimize the parameter of support vector machine regression (SVR) model, and an example was used to validate the parameter selection method. Then, the aerodynamic characteristics of a vehicle in the presence of wind barriers were considered as the objective functions, and the optimization model of the protective effects of wind barriers was presented. Lastly, according to the wind tunnel test results, the surrogate models of the objective functions were obtained by the SVR to optimize the heights and porosities of wind barriers. The results show that the modified grid search methods improve the accuracy of the parameter selections of SVR model. The optimal porosity of wind barriers is 0.00 - 0. 17 when the wind barrier height is 1.91 -2.90 m. If the height of the wind barrier is more than 2.50 m, the improvement of protective effect is limited with increase of the wind barrier height.
出处
《西南交通大学学报》
EI
CSCD
北大核心
2016年第6期1098-1104,共7页
Journal of Southwest Jiaotong University
基金
国家自然科学基金资助项目(51408503
U1334201)
四川省青年科技创新团队资助项目(15CXTD0004)
中央高校基本科研业务费专项资金资助项目(2682014BR049)
关键词
支持向量机回归
风屏障
防风效果
代理模型
SVR(vector machine regression)
wind barrier
protective effect
surrogate model