论文部分内容阅读
为了更好地监测钢液成分,利用激光诱导击穿光谱(LIBS)技术,建立了基于径向基函数(RBF)核函数的支持向量机模型,采用粒子群算法优化支持向量机的参数,通过计算元素特征谱线的积分强度和Fe元素内标归一化来降低仪器和环境带来的干扰。将实验数据进行主成分降维后,对钢液中Mn元素的浓度进行定量分析,得到均方根误差(MSE)为0.599%,相对标准偏差(RSD)为8.26%,相关系数为0.997。结果显示,粒子群优化支持向量机回归定量分析方法可以用于LIBS钢液成分分析,其分析性能较传统的定标方法有一定提高。
In order to better monitor the composition of molten steel, a support vector machine model based on Radial Basis Function (RBF) kernel function is established by laser induced breakdown spectroscopy (LIBS). Particle swarm optimization is used to optimize the parameters of SVM. Calculate the integral strength of the element signature line and the normalization of the Fe element internal standard to reduce the interference caused by the instrument and the environment. After the experimental data were reduced in principal components, the concentration of Mn in the liquid steel was quantitatively analyzed. The results showed that the root mean square error (MSE) was 0.599%, the relative standard deviation (RSD) was 8.26% and the correlation coefficient was 0.997. The results show that Particle Swarm Optimization Support Vector Machine regression quantitative analysis method can be used for LIBS analysis of liquid steel, and its analytical performance has some improvement over the traditional calibration methods.