中文摘要：

验证、确认与不确定度量化（V＆V＆UQ）是评估物理模型可信度和量化复杂工程数值模拟结果置信度的系统方法．验证是要回答数值模拟程序是否正确求解了物理模型和程序是否正确实施或给出求解模型的误差、不确定性大小及使用范围，确认是要通过数值结果回答物理模型是否反映了真实客观世界或反映真实客观世界的可信程度．文章围绕爆轰流体力学模型，剖析了模型中不确定性因素，给出了影响模拟结果不确定性的关键因素清单，并对其开展了敏感度分析，确认了模型的适应性．

英文摘要：

Verification, validation and uncertainty quantification （V＆V＆UQ） is a method of assessing the credibility of physical model and quantifying the confidence level of numerical simulation result in complex engineering. Verification is used to answer the question whether the physical model is well solved or the program is implemented correctly, and it will give the ranges of error and uncertainty. Validation is used to answer the question whether the physical model reflects the real world or the confidence level of the physical model. This article deals with the detonation computational fluid dynamics model, and analyses the uncertainty factor in modeling, then presents the key factor which affects the accuracy of the simulation result. Due to the complexity of the explosive detonation phenomenon, there are a huge number of uncertainty factors in the detonation modeling. The sensitivity analyses of these uncertainty factors are utilized to distinguish the main factors which influence the output of the system. Then uncertainty quantification is conducted in these uncertain factors. After comparing the simulation result with the experiment data, the adaptation of the model is validated. This procedure is applied to the cylindrical test with TNT explosive. From the result, we can see that the parameters in the JWL EOS are calibrated and the accuracy of the model is validated. By the way, through conducting the uncertainty quantification of this system, we obtain that the expectation and standard deviation of detonation pressure for TNT are 1.6 and 2.2 GPa respectively. Detonation velocity and position of the cylindrical wall accord well with the experiment data. That means that the model is suited in this case. This technique is also extended to the detonation diffraction phenomenon. We can conclude that simulation result is greatly affected by the scale of the cell. From these examples, we can infer that this method also has a wide application scope.