非对称下肢减重外骨骼刚柔膝关节设计与步态自学习研究

For the one-side knee Osteoarthritis (OA) patients, using the lower-extremity-exoskeleton (LEE) to relieve the compressive load and the painful feeling in the knee has become a new method of gait rehabilitation. However, studies in structure matching and movement coordination between mechanical LEE and biological body require further investigation.. In previous studies, the applicant discovers a rigid hinge mechanism cannot couple with the nonlinear structure of human biological knee joint, and fixed-angle controlled LEE cannot adapt to the patient individual gait. Preliminarily, the applicant has designed an adaptive knee joint exoskeleton with cam mechanism, and verified the passive hip learns to be in sync with the healthy hip, and the passive LEE supports the partial human bodyweight.. This project plans to decouple the kinematics basis of gait phase, analyze the relationship between gait phase and LEE locomotion, and study the weight-support method of knee joint exoskeleton based on gait phase. This project plans to model human knee biology, analyze the human-machine compliant fusion structure, and study the design method of rigid-compliant knee joint with optimal topology. This project plans to utilize the principle of machine learning, analyze the autonomous rule of gait parameters in different rehabilitation stages, and study the method for the disabled-side LEE learns the gait of healthy-side progressively. Finally, this project will reveal the scientific principle of compliant coupling and movement coordination between human and LEE.. This project has important theoretical significance and application value for the asymmetric LEE rehabilitates the one-side knee OA patients.

利用下肢外骨骼减小膝关节的受力和疼痛已成为单侧膝骨关节炎患者步态康复的新方法，但对机械外骨骼和生物学人体的“结构匹配和运动协调”的研究有待深入。.申请者在前期研究中发现刚性铰链式机械关节不能耦合人体生物膝关节的非线性结构，角度固定控制式外骨骼步态不能匹配患者个性化行走。申请者已初步设计凸轮式自适应膝关节外骨骼，并验证左右侧髋关节步态自学习的外骨骼对体重的部分支撑。.本课题拟通过解耦步态相位的运动学基础，分析步态相位和外骨骼的控制关系，研究基于步态相位的膝关节外骨骼支撑方法；通过人体膝关节生物学建模，分析人机闭环柔顺融合规律，研究刚柔耦合膝关节最优拓扑结构设计方法；通过建立机器学习机制，分析不同康复阶段参数自治规律，研究残障侧外骨骼递进式学习健康侧个性化的步态。最终，揭示“人机结构与运动共融”的科学机理。.本课题利用非对称下肢外骨骼康复单侧膝骨关节炎患者步态，具有重要的理论意义与应用价值。

下肢关节损伤是常见疾病，单侧下肢关节疾病具有普遍性，髋关节位居下肢“关节-肢体”串联型挂式结构的最上方，在控制步态上具有优势。我国下肢关节骨关节炎发病率在75岁以上的老年人中高达到70%；我国下肢关节的年损伤发病率达3%。因此，改善单侧下肢关节患者的行走对于提高国民生活质量具有重要研究意义。本项目主要研究内容有：步态特征运动分析与无线运动感知网络构建、下肢两侧协同运动规律表达和外骨骼减重支撑的设计、仿生变刚度阻尼柔性关节膝关节的仿生机械设计、人机耦合柔顺驱动控制策略、递进式康复策略制定、基于机器学习的步态康复智能进化。本项目的重要结果和关键数据有：人工神经网络能够准确分类步态相位，小步数（10步）就能快速完成参数训练，在实际应用中有很大的时间优势；外骨骼能在步态支撑相支撑人体体重，在步态摆动相适应人体关节运动；仿生变刚度阻尼柔性关节具有比较大的变结构运动学兼容性优势；支撑相和摆动相分别采用两种不同的控制模式，两者之间的切换由当时刻的步态相位决定；下肢康复可以分为软瘫期、痉挛期和恢复期，可分别进行被动运动、助力运动和阻抗运动；人体穿戴同步外骨骼后，同步式外骨骼髋关节在周期性上可以很好地拟合健康侧髋关节，两者数据变化趋势基本一致，但是精确的变化范围有一定的差异，但从整体趋势上来讲，同步式髋关节外骨骼对健康腿步态的复现效果还是准确的。本课题的科学意义是：人机结构共融，人体生物学非线性关节与机械柔性关节形成人机闭环结构链的融合匹配；人机运动共融，人体步态和外骨骼支撑、驱动时的柔顺协同运动规律，残障侧同步式外骨骼对健康侧个性化步态的智能自学习算法。