激光诱导击穿光谱（laser-induced breakdown spectroscopy，LIBS）是一种新兴的原子发射光谱技术，由于其具有无需样品制备、可远程实时分析等优势，在分析领域具有广阔的应用前景。然而，较差的定量性能阻碍了LIBS的广泛应用，较高的信号不确定性和严重的基体效应是影响其定量性能的瓶颈问题。尽管学者针对基体效应和不确定性对LIBS强度和定量的影响机理做了很多研究，但仍然不充分，存在不确定性和基体效应存在耦合，二者作用机制不一、基体效应研究集中于物化性质对强度的影响以及现有修正不确定性和基体效应的方法仍有待丰富等问题。针对以上问题，论文建立了等离子体辐射模型，从等离子体参数角度对基体效应与不确定性进行初步解耦，从而得以研究两种效应单独存在时对定量产生的影响，并设计了温度控制和多光谱平均方法以提高LIBS的定量性能。论文通过考虑电子跃迁过程中的自发辐射、受激吸收、受激辐射（可忽略）过程的，通过能量守恒方程建立了辐射模型。之后通过以样品间参数平均值表征不确定性、样品内参数波动表征不确定性，从而实现两种效应的解耦合。在探究多光谱平均对LIBS定量的影响时，发现仅存在不确定性时多光谱平均效果较显著，但两种效应同时存在时多光谱平均对定量的改善作用非常有效。论文随后通过研究表征基体效应的参数对定量的敏感性分析，得出样品间温度差异对定量准确性影响较大的结论，提出通过对光谱进行数据筛选来实现等离子体的温度控制，从而可以较显著的降低基体效应。在铜锌合金上的定量实验显示，先通过温度控制降低基体效应，再通过多光谱平均方法降低不确定性可有效提高LIBS的定量性能。

Laser-induced breakdown spectroscopy (LIBS) is an emerging atomic emission spectroscopy technique with promising applications in the analytical field due to its advantages such as no sample preparation and remote real-time analysis. However, the poor quantitative performance hinders the widespread application of LIBS, and the high signal uncertainty and severe matrix effect are the bottlenecks that affect its quantitative performance. Although scholars have done a lot of research on the mechanism of matrix effect and uncertainty on the spectral intensity and quantification of LIBS, it is still inadequate, and there are problems such as coupling of uncertainty and matrix effect, different mechanisms of their effects, the research on matrix effect focuses on the effect of physical and chemical properties on strength, and the existing methods to correct uncertainty and matrix effect still need to be enriched. To address these problems, the paper establishes a plasma radiation model to initially decouple the matrix effect and uncertainty from the perspective of plasma parameters, thus enabling the study of the quantitative impact of the two effects when they are present separately, and designs a temperature-controlled and multispectral averaging method to improve the quantitative performance of LIBS.The paper establishes a radiation model through the energy conservation equation by considering the spontaneous radiation, excited absorption, and excited radiation (negligible) processes of the electron leap process. Afterwards, the two effects are decoupled by characterizing the uncertainty in terms of inter-sample parameter averages and intra-sample parameter fluctuations. When investigating the effect of multispectral averaging on LIBS quantification, it was found that the effect of multispectral averaging was more significant when only uncertainty was present, but the improvement of quantification by multispectral averaging was very effective when both effects were present. The paper then investigates the sensitivity analysis of the parameters characterizing the matrix effect on quantification and concludes that the temperature difference between samples has a large impact on the quantification accuracy, and proposes that the temperature control of the plasma can be achieved by data screening of the spectra, which can reduce the matrix effect more significantly. The quantitative experiments on Cu-Zn alloys show that the reduction of matrix effects by temperature control followed by the reduction of uncertainty by multi-spectral averaging method can effectively improve the quantitative performance of LIBS.