Research On Compliant Series Elastic Actuator Structure Design And Interactive Control Method

The number of people with physical disabilities caused by stroke is rapidly increasing with the arrival of an aging society.Traditional rehabilitation training has drawbacks such as a single training mode,long rehabilitation period,and poor effectiveness.Therefore,rehabilitation robots are of great significance in promoting patient neuroplasticity,improving rehabilitation training effects,and enhancing patient quality of life.Currently,most rehabilitation robots use rigid drives,which have advantages such as high position control accuracy.However,position control alone is insufficient to guarantee patient safety.Rehabilitation robots with characteristics such as safety,compliance,and force perception are considered a hot topic in the field of intelligent medical rehabilitation.This paper focuses on the study of compliant series elastic actuator structures and interactive control algorithms.The main research content of this paper is as follows:(1)A novel compliant series elastic actuator structure is designed based on torsion spring devices and linear spring devices.Considering the safety issues of rigid actuators for subjects,combined with the biomechanics of human gait,a novel compliant series elastic actuator is proposed to output flexible power for rehabilitation robots.The output torque of the actuator is increased in the limited space of the actuator by using the worm gear reducer,and the gear rack is used as the transmission device to improve the load-bearing capacity of the actuator.The torsion spring devices and compression spring devices are utilized to provide safe and flexible power to the actuator.Based on the mass-spring-damper system model,the interaction force between the worm gear reducer,the gear rack device and the interference caused by the various components of the actuator are considered to establish a dynamic model of the actuator.(2)A linear quadratic(LQ)optimal control algorithm is developed to improve system convergence and stability.An optimal control algorithm is developed to address the problem of parameter tuning in proportional integral differential(PID)control methods.The optimal parameters are obtained by selecting appropriate weight matrices,enabling precise force control for the compliant series elastic actuator.The effectiveness of the optimal controller is verified through numerical simulations and platform experiments.(3)A noise-tolerant zeroing neural network(NTZNN)control algorithm is presented to suppress noise generated by the actuator.An NTZNN controller is presented to address the disturbance caused by mechanical structural errors,mechanical vibration,and signal interference during the operation of the compliant series elastic actuator.Compared with the traditional original zeroing neural network(OZNN)controller,the NTZNN controller is shown to suppress various types of mixed noise,and improve the convergence,stability,and robustness of the actuator system.(4)A human-machine interaction control algorithm is proposed to keep the subjects safe.A dual-mode interaction controller is suggested based on the dynamic model of the compliant series elastic actuator,and a nonlinear disturbance observer is designed to improve the robustness of the controller.Through numerical simulations and platform experiments,the designed controller is shown to have low output impedance characteristics and can achieve precise force output control,improving the safety and compatibility of human-machine interaction.