利用支持向量回归对燃爆单元宽度进行预测的方法研究

发布时间：2018-05-09 01:31

  本文选题：氢气燃爆 + 燃爆单元宽度　； 参考：《原子能科学技术》2017年06期

【摘要】：燃爆单元宽度λ是衡量可燃气体燃爆风险的一项重要参数,通常认为其与特征化学反应区宽度δ的比值是无量纲活化能和无量纲温度的函数。在以上述两个无量纲量为自变量、以λ/δ的对数为因变量对实验数据进行回归的基础上,更进一步引入无量纲压力作为第3个自变量进行回归。另外,针对传统参数回归方法的不足,采用基于机器学习的支持向量回归方法进行数据拟合。比较回归结果发现,与采用二变量模型及参数回归方法相比,采用三变量模型及支持向量回归方法的计算结果与实验数据的拟合度更好,并能更为准确地预测不同初始条件下可燃爆气体的特征单元宽度。
[Abstract]:The width of combustion unit 位 is an important parameter to measure the explosion risk of combustible gas. It is generally considered that the ratio of 位 to the width 未 of characteristic chemical reaction zone is a function of dimensionless activation energy and dimensionless temperature. Taking the two dimensionless variables as independent variables and the logarithm of 位 / 未 as dependent variables, we further introduce dimensionless pressure as the third independent variable to regress the experimental data. In addition, the support vector regression method based on machine learning is used to fit the data. Compared with the two-variable model and the parametric regression method, the results obtained by using the three-variable model and the support vector regression method are better than those of the experimental data. And the characteristic unit width of ignitable gas under different initial conditions can be predicted more accurately.
【作者单位】： 清华大学工程物理系;卡尔斯鲁厄理工学院;中国核动力研究设计院核反应堆系统设计技术重点实验室;
【基金】：核反应堆系统设计技术重点实验室基金资助项目(HT-A100K-02-201406)
【分类号】：TL364.3
，

本文编号：1863988