[Adaptive weighted fuzzy iterative learning control strategy for functional electrical stimulation integrating...

To address the limitations of conventional iterative learning control in functional electrical stimulation-based wrist extension rehabilitation, including the lack of active user intent, poor convergence, and the difficulty in balancing fatigue mitigation with comfort, this paper proposes an adaptive weighted fuzzy iterative learning control strategy enhanced by guiding-side surface electromyogram (sEMG) threshold gating and dynamic frequency adjustment (AWFILC-GF). Using pre-recorded joint angles from healthy subjects as reference trajectories, this method employs contralateral sEMG gating for intent-driven activation and adopts a dynamic frequency modulation strategy to delay muscle fatigue. Furthermore, the AWFILC algorithm integrates fuzzy gain tuning with subcutaneous fat compensation to achieve model-free closed-loop trajectory tracking solely through pulse width adjustment. Experimental results demonstrate that, compared to baseline adaptive weighted fuzzy iterative learning control (AWFILC), the proposed method significantly reduced the relative decay rate of peak wrist extension angle tracking ratio by 6.8 percentage points in stroke patients (16.3% vs. 23.1%, P<0.05), significantly lowered the pulse duty cycle to 3.34% ( P<0.01), and maintained the visual analog scale score at 3.2 ( P>0.05), indicating effective fatigue alleviation without compromising comfort. Additionally, the system requires only a single electrode pair and an inertial measurement unit, achieving stable convergence within five to seven iterations. In summary, the proposed strategy achieves an effective integration of active intent, dynamic frequency modulation, and adaptive tracking of target movement trajectories, providing a clinically translatable technical solution for the rehabilitation of mild-to-moderate stroke with Brunnstrom recovery stage (BRS) of IV-V.