Data-Driven Stochastic Control of Atomic Number Density in Spin-Exchange Relaxation-Free Co-Magnetometers

This paper presents an innovative application of stochastic control methodologies for optimizing the performance of spin-exchange relaxation-free co-magnetometers under conditions of internal heat source instability-induced atomic number density fluctuations. A data-driven modeling framework is adopted to develop an ARMAX model for the non-magnetoelectric heating system, followed by the implementation of a minimum variance control strategy to stabilize atomic number density within the gas cell. Comparative analysis against open-loop and ADRC approaches, based on simulation and experimental results, reveals a substantial reduction in atomic number density variance, showcasing superior control efficacy. This study emphasizes the integration of stochastic control techniques into advanced instrumentation, thereby offering avenues for enhancing the operational capabilities of thermal atomic instruments.