High-Precision Noise Suppression Method Based on IACAC-LSSVM for MEG Measurement

Magnetoencephalography (MEG) measurement, which detects extremely weak magnetic field signals in the brain for studying brain function, is susceptible to external magnetic field noise. Therefore, creating a low-noise and weak magnetic environment is crucial for enhancing MEG measurement accuracy. The active magnetic compensation (AMC) system used for the magnetic shielding room (MSR) is an effective approach, yet issues like inaccurate MSR model and system time lag affect compensation precision. To address these issues, this article proposes a high-precision magnetic field noise suppression method based on an improved all-coefficient adaptive control combined with least-squares support vector machine (IACAC-LSSVM). It utilizes all-coefficient adaptive control (ACAC) to control the inaccurate MSR plant, improves the linear addition of ACAC output through nonlinear state error feedback (NLSEF) to enhance disturbance suppression capability, and combines LSSVM to predict control error and increase bandwidth. Experimental results show that this method improves control accuracy by 36.8% when the model changes, has strong disturbance suppression ability, and increases the noise suppression bandwidth by 1.8 and 1.3 times compared with $\mu $ -synthesis control and ACAC. When applied to MEG measurement in a compact MSR, it effectively suppresses low-frequency noise and highlights the alpha rhythm, strongly supporting the clinical diagnosis of brain function.