High-Precision Magnetic Field Control of Active Magnetic Compensation System Based on MFAC-RBFNN

Active magnetic compensation technology can effectively reduce magnetic field disturbances within a magnetic shielding room (MSR) and improve the signal-to-noise ratio of magnetoencephalography (MEG) measurement. But, for small-sized MSRs with external compensation coils, achieving high-precision magnetic field control is challenging, because it is difficult to establish an accurate mathematical model. In this article, an active magnetic compensation system is constructed based on model-free adaptive control with a radial basis function neural network (MFAC-RBFNN) method, which addresses the limitations of magnetic field control accuracy caused by the requirement for precise system model information. The nonlinear and coupling characteristics of the active magnetic compensation system were analyzed, and a model-free adaptive control (MFAC) controller is designed based on the input current and output magnetic field, and the utilization of radial basis function neural network (RBFNN) for estimating magnetic field disturbances. The experimental results are given to prove that the algorithm proposed can achieve high-precision control of magnetic field within the MSR without an accurate system model, and compared with proportional-integral-derivative (PID), the magnetic field disturbance reduction effect is improved by 2.4x. It contributes to generating a near-zero magnetic field environment with low magnetic field disturbance.