Enhancing the performance of serf atomic gyroscopes with novel dual-axis gradient compensation coils: Design, implementation, and in-situ validation

Spin-exchange relaxation-free (SERF) atomic gyroscopes (AGs) offer unprecedented theoretical precision for next-generation inertial navigation. However, complex boundary effects induced by high permeability magnetic shields and optical access holes significantly degrade the linearity of the compensated gradient field, becoming a key bottleneck restricting performance improvement. This study investigates the influence mechanism of magnetic field gradients on AGs performance, and designs, implements, and in-situ validates novel dual-axis gradient compensation coils adapted to complex boundaries. A dual-segment analytical model is constructed based on the target field method, combined with the image method to accurately characterize boundary effects. Furthermore, a hybrid optimization framework integrating trace-scaled regularization and dandelion optimizer is introduced to adaptively improve compensation gradient linearity. Results show that the maximum relative deviation between the measured and simulated values of magnetic field nonlinear error is less than 1%, and the coil gradient constant is highly consistent with theoretical value. Compared with traditional nested gradient-saddle compensation coils, the gradient nonlinear error is suppressed to below 1.22%, representing an improvement of nearly an order of magnitude. Crucially, in-situ validation demonstrates that the Allan deviation at 100 seconds is reduced by 23.8% to 0.0087 °/h, and the sensitivity is improved by 35.7% to 8.04×10 -6 °/(s∙Hz 1/2 ). This work resolves the long-standing challenge of generating high-fidelity gradient fields under complex boundary conditions, markedly enhancing the environmental adaptability and inertial measurement performance of AGs while establishing a robust magnetic compensation foundation for next-generation, high-precision AGs systems.