Parameter Identification And Control Of Lower Limb Exoskeleton Based On DeLaN

With the increasing aging of the country and society,rehabilitation-type exoskeleton robots such as elderly care and disability assistance show broad application prospects.In the field of exoskeleton robot research,there are still a series of problems such as low modeling accuracy,insufficient control performance,low human-machine coupling comfort and inaccurate human intention perception.Therefore,this thesis takes UEXO-Ⅰ lower limb exoskeleton prototype as the main research object,and carries out research on model parameter identification,human-machine coupling control and human intention perception.Ultimately,the exoskeleton human-machine can efficiently achieve control objectives.The main research and contributions of this thesis are as follows:1.Based on the “Euler-Lagrange” modeling method,the dynamic model of the lower limb exoskeleton prototype is established,and the inverse step controller for 2-DOF lower limb exoskeleton is derived.Based on the prototype dynamic model and canvas controller,MATLAB/SIMULINK is used to complete corresponding simulation experiments.2.Aiming at the balance problem between exoskeleton modeling accuracy and cost,a De La N model with friction force and residual compensation is proposed to identify parameters for exoskeletons.The network for exoskeleton prototype platform is improved to realize end-to-end training and parameter identification work.The torque error of the identification model for two joints is around 1.3Nm and 3.2Nm.3.Impedance and admittance controllers are designed based on parameter identification results to improve user experience of lower limb exoskeleton prototype while further verifying effectiveness of parameter identification network.The admittance control error is on the order of 1e-4 radians.4.For human intention perception,this thesis designs a lightweight gait sensing data acquisition system based on sensors attached to human body.Based on gait sensing data obtained by acquisition system a gait recognition network based on artificial features fusion with sensor time series is designed which can achieve more fine-grained gait recognition results.The overall recognition rate of the final gait reached 95.66%...