Research On System Design And Human-robot Interaction Strategy Of Lower Limb Rehabilitation Robot

The lower limb rehabilitation robot,as a new type of intelligent medical device,can effectively promote the leg movement function rehabilitation of stroke patients.It is widely concerned by scholars all over the world.At present,the research on multi-dimensional human-robot information interaction,such as active strength training,movement intention recognition and abnormal physiological information detection,is not in-depth enough.In order to solve the above problems,this thesis proposes a sitting and lying lower limb rehabilitation robot.On this basis,researches are carried out around force information interactive control strategy in active force training,movement intention recognition strategy based on annealing chaotic neural network,and real-time monitoring of patient’s physiological information in training.Firstly,based on the physiological structure of human lower limbs and the needs of human-robot interaction in cooperative training,the sitting and lying lower limb rehabilitation robot system is designed.It mainly includes the design of joint transmission mechanism,auxiliary adjustment mechanism and sensor control system.Then the robot is modeled and analyzed,including kinematics/dynamics analysis and simulation.According to the structural characteristics of the lower limb rehabilitation robot,the target controlled model is established.The safe workspace under different training postures is analyzed,and the variable speed training trajectory is planned.Secondly,based on the enthusiasm and safety of active training,the interactive strategies of virtual reality active strength training and lower limb flexion and extension strength training are formulated respectively.The virtual reality active training scene that can provide positive feedback for patients is built.In order to improve the virtual body feeling and control accuracy in training,the variable stiffness admittance control strategy and interactive adaptive tracking control strategy are developed;By analyzing the biomechanical characteristics of the knee joint,the relationship between knee bending angle,soft tissue compression,and muscle group stimulation is clarified.Thus,a lower limb flexion and extension strength training that can both reduce the damage of the knee joint soft tissue and ensure the training effect of the target muscle group is designed.At the same time,a dual input robust control tracking strategy is proposed to improve the stability and tracking accuracy of the system.Thirdly,in order to accurately obtain the patient’s lower limb movement intention,the characteristics of lower limb musculoskeletal movement are analyzed.The target muscle group is determined,and the sample data of electroencephalogram signals under four joint movements are collected.The sample data is preprocessed,and the candidate eigenvalues are optimized through clustering evaluation indexes.In order to analyze the movement intention of joints,a chaotic neural network model with annealing attenuation hidden layer neurons is proposed.The structural parameters of the model are determined,and the training data set is constructed.The model is trained and preliminarily tested,and the corresponding autonomous training control strategy is established.Then,in order to analyze the fatigue degree of patients in active rehabilitation training,the data set is built on collected multiple training characteristic parameters and fatigue level questionnaire answers.The regression prediction model of fatigue degree is established by using the extreme gradient boosting machine learning algorithm,and the fatigue degree can be judged according to the change of characteristic information.In order to judge the abnormal muscular tension in passive training,a recognition strategy based on human-robot relative micro motion is proposed.Comparing patients’ limb movements with mechanical leg joint movements,the muscular tension of the lower limbs in passive training can be judged.Meanwhile,the corresponding emergency plan is formulated to prevent secondary injury.Finally,the virtual reality active strength training and lower limb flexion and extension strength training are tested,including the performance test of adaptive admittance controller and robust admittance controller,joint tracking performance test and training effectiveness test.At the same time,additional volunteers without sample data collection are selected to carry out adaptability experimental test on the movement intention recognition strategy and the training fatigue prediction scheme.