The magnetometer is a multi-parameter system, parameters optimization is always the problem to improve the sensitivity of the atomic magnetometer. Multiple parameters hardly decide the optimized conditions in the experiments because of the influence of several parameters and parameters may interact with each other. Besides, the influence of parameters on the signal-to-noise ratio can hardly be analyzed theoretically due to the complex physical configuration. We used a generalized polynomial chaos expansion to construct an agent model for the SERF atomic magnetometer to replace its complex physical model, and used a variance-based sobol method to perform a global sensitivity analysis of the parameters in this paper. The mathematical method of uncertainty quantification is used to comprehensively analyze six parameters’ effect on SERF atomic magnetometer performance. The results show that the probe laser power has the greatest impact on the atomic magnetometer, and it is the parameter with the highest degree of linearity with the atomic magnetometer. In accordance with the results of the uncertainty quantification analysis, the parameters of the probe laser power of the SERF atomic magnetometer were experimentally optimized, and the performance of the magnetometer was significantly improved.

中文翻译：

---

SERF 原子磁力计参数优化的不确定性量化


磁力计是一个多参数系统，参数优化一直是提高原子磁力计灵敏度的问题。多个参数在实验中很难决定最佳条件，因为多个参数之间存在影响，而且参数之间可能会相互作用。此外，由于物理结构复杂，参数对信噪比的影响很难从理论上分析。本文使用广义多项式混沌展开构建了SERF原子磁力计的代理模型来代替其复杂的物理模型，并使用基于方差的sobol方法对本文的参数进行了全局敏感性分析。采用不确定度量化的数学方法，综合分析了6个参数对SERF原子磁力计性能的影响。结果表明，探针激光功率对原子磁力计影响最大，是与原子磁力计线性度最高的参数。根据不确定度量化分析的结果，对SERF原子磁力计的探测激光功率参数进行了实验优化，磁力计的性能显着提高。