X—射线荧光光谱测定钨基合金粉中Cu、Ni、W、Fe—经验系数法
来源期刊：中南大学学报(自然科学版)1981年第1期
论文作者：罗重庆 康叔常 赵新那
文章页码：22 - 31
关键词：联立方程组； 经验系数法； 准确度； 合金粉； 数据处理； 行列式； 化学法； 标准样品； 数学模型； 经验方程
摘 要：本文法采用了Criss和Birks提出的标准与被测元素数目相同的数学模型。在用Fatemi和Birks的零行列式法进行计算时,经验方程的几个联立方程组的解一致性提高了,但在我们的实验中准确度（以化学法为准）一般未得到改善。我们发现,选择所解的联立方程组对结果的准确与否具有一定的规律性。解一个联立方程组即可得到与化学法更为接近的结果,且计算过程简单、快速。对本实验而言,解舍去经验方程中含Rw方程后的联立方程组结果最好。当用Fatemi和Birks的实验数据解舍去含RFe方程后的联立方程时,其结果也证实了上面的规律。方法的标准偏差（10号样）对Cu、Ni、W、Fe分别为0.03,0.013,0.08,0.001。变动系数分别为0.32%,0.63%,0.036%,10.8%。数据处理采用自编计算程序。
Abstract: The mathematical model for data processing used in the method is the regression equations described by riss and Birks,in which the numbers ofthe standards are the same as those of the elements to be determined in thesample.By using Fatemi and Birks null-determinant technique,a better consi-stency for the solutions of empirical quations is obtained,but the agreementwith chemical analysis is not generally improved in our work.We found from ur experiment that,as a rule,by the proper choice ofsubset equations,we need to solve only one set of simultaneous quations.Calculated compositions,in general,agree more closely with the chemicalanalysis,and the process of the alculation is simpler and faster.In our case,omitting the empirical equation representing Rw gives best result,the amerule is also shown by Fatemi and Birks′ experimental data by omitting theequation representing RFe.The tandard deviations for Cu,Ni,W and Fe are（for No.10 sample）0.03,0.13,0.08 and 0.001 respectively,and the coefficient f variation is0.32%,0.63%,0.036% and 10.8% respectively.Data processing is done by our own programs using either I-59 program-mable calculator or TRS-80 LEVEL Ⅱ micro-computer（BASIC）.
