Application Of FES And Attitude Sensors In Modeling Of Ankle Joint Angle Adjustment

In recent years,Functional Electrical Stimulation(FES)has been widely used in limb rehabilitation and has become a hot research topic in the field of intelligent rehabilitation.FES referred to the cutting-edge technology that restores the function of specific parts of the body by stimulating muscles,the muscle has lost nerve control through electrical stimulation.FES causes muscles to contract,and its effectiveness depends on the precise control of stimulation time and stimulation intensity.With the development of the FES system from the open-loop to the closed-loop,establishing the model of joint angle under electrical stimulation is essential for studying the FES closed-loop control system.Therefore,FES and attitude sensors are employed to build a human experiment platform and to explore the response and characteristics of the ankle under FES.From a phenomenological perspective,a neural network-based H(Hammerstein)model was established and trained using genetic algorithms.From a physiological perspective,the mechanism model of ankle joint angle under FES was established,and the Extended Kalman Filter(EKF)algorithm was used to parameter estimation.Two models were compared and discussed.The specific research contents are as follows:Firstly,an experiment platform was built with an attitude sensor and an electrical stimulator,and FES adjusted the angle of the ankle joint in experiment based on different experimental subjects.The stimulus frequency of the experiment was selected as 25 Hz,the stimulation current intensity was fixed at 25 mA,and the pulse width was adjusted to explore the relationship between ankle joint angle and the stimulation intensity.The results show that the response of joint angle to the stimulus presented time-delay,nonlinear,and time-varying characteristics,which provided theoretical and data basis for the subsequent establishment of ankle joint models of different individuals.Secondly,according to the experimental results,from the point of view of phenomenology,the H model of neural network structure was established.The parameters of the model were identified by the genetic algorithm.The ten-fold cross method was used to validate the model in the body experimental data.The results show that the H model can effectively predict the angle changes of the ankle joint under electrical stimulation.Again,ankle joint kinematics,FES muscle contraction dynamics and Lagrangian inverse dynamics were analyzed based on the mechanistic method,and an ankle musculoskeletal model was deduced under electrical stimulation.Although the mechanism model has better interpretability,the structure is complex and the parameters are identified more difficultly.In this thesis the model was simplified and the unknown parameters of model was identified by using EKF algorithm.The results showed that the model can effectively predict the response of the changes of the ankle joint angle under electrical stimulation.Finally,the model errors of the two models in different individuals and different electrical stimulation input sequences were compared and analyzed.The results show that the normalized root mean square error(NRMSE)of the ankle joint musculoskeletal model based on the physiological methods was lower than 20%,NRMSE of the H model based on phenomenological methods is lower than 30%on average.In this thesis,two models can predict the change of the ankle joint angle and the mechanism model is slightly better by comparing the accuracy and the robustness of the model.This study lays a foundation for the optimization of the configuration of electrical stimulation parameters,in order to explore the characteristics of FES during the movement of the ankle joint angle,and improve the accuracy of the adjustment of the electrical stimulation closed loop treatment system.