激光诱导击穿光谱(LIBS)技术用于钢液成分在线定量分析时,基体效应会对其精确度产生严重影响。在定量分析时,用一种改进的多元非线性模型进行定标,以降低基体效应对待测元素的影响,并与单变量定标和改进前的多元非线性模型定标进行对比。结果表明,与单变量定标相比,多元非线性模型定标的测量精度有所提高,模型改进后,其分析性能得到进一步完善。测量元素Mn、Si的定标曲线的拟合度从0.980、0.984分别提高到0.985、0.989,两个验证样品的预测相对误差分别从6.231%、5.437%和6.912%、6.315%下降到5.510%、5.039%和6.125%、5.919%。 Laser Induced Breakdown Spectroscopy (LIBS) technology is used to quantify the composition of molten steel online, the matrix effect will have a serious impact on its accuracy. In quantitative analysis, an improved multivariate nonlinear model was used to calibrate to reduce the effect of matrix effect on the tested elements, and to compare with the calibration of univariate and multivariate nonlinear model before improvement. The results show that compared with univariate calibration, the measurement accuracy of multivariate nonlinear model calibration is improved. After the model is improved, its analytical performance is further improved. The fitting accuracy of the calibration curves of the measured elements Mn and Si increased from 0.980,0.984 to 0.985,0.989, respectively. The relative errors of the two validation samples decreased from 6.231%, 5.437% and 6.912%, 6.315% to 5.510% respectively, 5.039% and 6.125%, 5.919%.