基於RBFN模糊滑模自適應控制之髖膝外骨骼機器人在偏癱復健應用

Abstract

偏癱患者由於腦部損傷或中風等原因，常導致單側肢體運動功能障礙，嚴重影響日常生活活動能力。為解決此問題，本研究開發了一種能夠輔助偏癱患者進行步行訓練和日常活動的智能外骨骼系統，該系統針對偏癱患者所設計的髖膝外骨骼機器人，其控制基於徑向基底函數網路（Radial Basis Function Network, RBFN）模糊滑模控制(Fuzzy Sliding Mode Control,FSMC)方法設計。該系統結合了RBFN神經網路與模糊滑模控制的優勢，整合成一種自適應控制策略，能夠即時調整外骨骼機器人的輔助力度和運動軌跡。利用RBFN神經網路進行模式識別，然後通過模糊滑模控制器實現平滑的輔助控制。本研究進行了多組不同的實驗，從正常步行、慢速步行、變步距步行三種情境，觀察 PID、FSMC 與 RBFN-FSMC 三種控制器之表現。實驗結果表明，相比傳統的PID控制和單純的模糊滑模控制方法，所提出的RBFN模糊滑模控制系統在輔助效果、人機協同性和適應性方面均有顯著優勢。

This study proposes a hip and knee exoskeleton robot system designed for hemiplegic patients, based on the Radial Basis Function Network (RBFN) and Fuzzy Sliding Mode Control (FSMC) approach. Hemiplegia, often caused by brain injury or stroke, leads to motor function impairment on one side of the body, significantly affecting the ability to perform daily activities. To address this issue, an intelligent exoskeleton system was developed to assist hemiplegic patients with gait training and daily movement.The proposed system integrates the strengths of RBFN neural networks and fuzzy sliding mode control to form an adaptive control strategy capable of real-time adjustment of the assistive force and motion trajectory of the exoskeleton. The RBFN is used for pattern recognition, while the fuzzy sliding mode controller enables smooth and responsive assistance.Experiments were conducted under different walking conditions, including normal walking, slow walking, and variable step-length walking, to evaluate the performance of three controllers: PID, FSMC, and RBFN-enhanced FSMC. Experimental results demonstrate that the proposed RBFN-FSMC system outperforms traditional PID control and basic FSMC in terms of assistive effectiveness, human-robot coordination, and adaptability.