论文部分内容阅读
通过解析发射光谱和脉冲分辨分析联用技术(OES/PDA)数据确定钢铁中夹杂物的类型及计算其粒度分布需要借助于高等数学方法,以便在大量的数据中滤出那些与夹杂物相关的OES/PDA数据,从而找到可靠的该数据集的离群数据。强度及大多的质量分布其分布函数都是非正态分布的,因此,通过对这些数据进行拟合即可得到非属于该正态分布的离群数据。离群数据确定后,通过这些信息可找到与一些特定夹杂物类型相关联的数据群。基于此目的,发现一种称为“自组织特征映射(SOM)”的数学方法非常适于进行此项研究。根据离群数据及夹杂物类型的相关信息,采用该方法可以计算出质量分布并能进一步进行不同夹杂物类型的粒度分布计算。
Determining the type of inclusions in steel and calculating their particle size distribution by analyzing emission spectroscopy and pulse-resolved analytical techniques (OES / PDA) data requires advanced mathematical methods to filter out large amounts of data associated with inclusions OES / PDA data to find reliable outlier data for that data set. Intensity and Most Mass Distributions The distribution functions are all non-normal distributions, so outliers that do not belong to this normal distribution can be obtained by fitting these data. After the outlier data is determined, this information is used to find the data set associated with some of the specific inclusion types. For this purpose, it was found that a mathematical method called “self-organizing feature mapping (SOM) ” is very suitable for this research. Based on the information about the outlier and the types of inclusions, this method can be used to calculate the mass distribution and to further calculate the particle size distribution of different types of inclusions.