Research On Interactive Control Method Of Biomimetic Upper Limb Rehabilitation Robot Based On Multi-source Signal Fusion

With the increasing of aging population and the number of physical disabilities in our country,the pressure in the field of rehabilitation therapy will continue to increase.As a kind of convenient and efficient rehabilitation training equipment,upper limb rehabilitation robots are more and more widely utilized in the field of rehabilitation therapy.In order to provide a safe,active and comfortable rehabilitation training environment for patients,the dynamic model of upper limb rehabilitation robots,active movement intention recognition of human upper limbs and human-machine interaction control methods will be investigated in this paper.The main contributions are shown as follows:(1)To solve the problem of dynamic modeling of upper limb rehabilitation robots,this paper analyzes the upper limb bone and muscle structure from the perspective of anatomy and biomechanics.In addition,considering the interference factors such as joint damping,muscle contraction force and external friction during rehabilitation training,a dynamic model of human-machine coupling with interference items is established,which lays the foundation for the research of human-machine interaction control methods.(2)To solve time-varying problems,two high-precision discrete-time zero-type models are proposed and analysed for online solving dynamic system of linear equations(DSLEs).Theoretical analyses demonstrate that the proposed models possess zero stability,consistency and convergence,which lays the algorithm foundation for the recognition of human active motion intention.Furthermore,two discrete-time noise-tolerant zeroing-type models are investigated and verified to solve nonlinear time-varying equation problems(NTVEPs)with different types of noise.Theoretical analyses and experimental results prove that the developed models are feasible and superior when utilized to solve NTVEPs,thus laying an algorithm framework for human-machine interaction.(3)In order to recognize the active movement intention of the upper limbs,a multipleinput and multiple-output Elman neural network(ENN)prediction model was established,which based on the upper limb dynamics model and surface electromyogram(s EMG)signals of the human upper limbs.The joint torque is effectively predicted and recognized,which lays the theoretical foundation for the human-machine interaction control of the upper limb rehabilitation robot.(4)Aiming at the research on the human-machine interaction control method of the upper limb rehabilitation robot,in an ideal environment,a zeroing neural network(ZNN)human-machine interaction controller based on the active movement intention of the upper limb is established.In the environment of noise interference such as joint damping and muscle contraction force,a noise-tolerant zeroing neural network(NTZNN)human-machine interaction controller based on the active movement intention of the upper limb is established to avoid secondary injury of the injured upper limb,which provides a safe,active and comfortable rehabilitation training environment for patients.