<p>In the process of finite element model updating, conducting sensitivity analysis on various input parameters can improve the accuracy of parametric sensitivity analysis, thereby enhancing the model’s precision. However, analyzing the sensitivity of multiple parameters poses a challenge. This paper addresses this issue through interpretable machine learning techniques. Utilizing interpretable machine learning for high-dimensional parametric sensitivity analysis helps improve both the efficiency and accuracy of such analyses. The first step of this method is to select update parameters. Three parameters were selected: geometric dimensions, material properties, and damping ratio. Then the performance of six machine learning algorithms was compared. Finally, the SVM-based interpretable machine learning technique was applied to the parameter sensitivity analysis process. This study’s findings demonstrate significant improvements over range analysis and ANOVA based on orthogonal experiments. The results indicate that interpretable machine learning techniques align with ANOVA and range analysis in identifying the top seven most sensitive parameters. However, variance analysis and range analysis may rank highly sensitive parameters lower due to their inability to account for parameter interactions, reducing accuracy. In contrast, interpretable machine learning techniques provide more diverse visualizations, enhancing the efficiency of parameter sensitivity analysis. They also effectively address the sensitivity analysis challenges in complex structures.</p>