Study Of Fes Cycling Rehabilitation Training System Based On Motor Assistance

Functional Electrical Stimulation (FES) is a cross-frontier rehabilitation technology that using electrical impulse via electrodes to stimulate the muscles that are loss of nerve control to contract, thus rebuild their function of related limbs and organs of the body. FES technology has been applied in a variety of areas with various forms. FES cycling system is based on FES technology and integrated with machinery, control and rehabilitation techniques, which can make patients realize the voluntary exercise and assistant walk. Patients may benefit from the exercise for the improvement of cardiorespiratory function, the increase of muscular strength and thickness, and the rebuilding of partial function. Furthermore, it may relieve patients from the family burden and strengthen their confidence to overcome disease. The advantages have been proved by numerous experiments. In the present study, bases on the reality of FES cycling technology, the FES cycling system was constructed and modeled, and the related control strategy was studied. The simulated and experimental results indicate that the system functioned well in the rehabilitation training of patients and achieved the proposed goal.Firstly, this paper reviews the various techniques of FES cycling system. Based on the present advantages of the cycling system, a FES cycling system based on motor assistance was studied. The characteristics of the system were summarized: it consists of two close loops for control and five structural parts. Constitutions and working principles of every part were introduced in detail with the viewpoint of hardware and software. A dynamic model was built to describe the interaction between the object and the system during the training process. In the model, the kinematical model, electrical stimulation - muscle force model and dynamics model were integrated. Then simulation was conducted under the circumstance of overcoming leg weight. The comparison between the simulated and experimental results verified the accuracy of the model. Based on this analysis, a dynamic model and a simulation model with velocity and pulse width of assistant motor as input data were established to get the muscles output torque. The velocity and pulse width in the identification test and step response test were used in the simulation. The simulation results are in accordance with experimental ones which indicate that the model is with content validity. Secondly, the system cycling kinematics model was established in this study to get the limit of the angle of related joints in terms of crank angles, which are essential to determine the joint states. Synthesized the effect of muscles contraction to joint state, a zero limit stimulation pattern was obtained to optimize the maximum muscle cadence torque and avoid joints contradictory movement caused by muscle. To testify the experimental optimization method with the aim to obtain the maximum output torque and the minimum muscle force, the optimal objective function was constructed to get the optimization results of initial and terminated angles for three groups of muscles under various crank velocities. With consideration of the delay of the response of simulated muscles, a delay factor was introduced in the model to modify the stimulating patterns. Moreover, the hacking movement was analyzed and its influence factors were determined.Finally, in order to realize the constant power output of FES cycling system, a systematic identification-based model was developed in the respect of two closed loops, velocity and power. The identifications related to PWM - speed model and the pulse width-power model were tested and the model parameters and the structure in accordance with the selected modal patterns and rules were obtained. The comparison of model output and the actual output data testifies the accuracy of identification model. The RST control strategy of the closed-loop control was used in the study based on the analysis of the open-loop performance of two models. Then, a controller was designed based on the RST algorithm after the determination of the performance specification of the closed loop controller. The simulation and experiments results show that the parameters of controller can satisfy the design requirements, and its stability, anti-disturbance and tracking performance are good. With consideration of the effect of the muscles fatigue on the pulse width - power model, a fuzzy PID auto-adjusted controller was designed with integration of advantages of fussy system and PID. The tracking and anti-disturbance performance of this controller was testified by experimental and simulated results. Besides, a LQR controller and a modified RST controller were designed using fuzzy T-S controlling strategy, based on the identification of the models for the fatigue and non-fatigue states of patients respectively. Whereat, these controllers for different state were integrated through T-S techniques to satisfy the requirements of system so that it can generate output automatically and proportionally when it works at non-fatigue, fatigue states and in-between. The simulated results indicate that the design of the controller is in good use.