Research On Neural Networks Sliding Mode Control Of Powered Transfemoral Prosthesis Knee Joint

Active prosthetic knee joint is developed to provide walking assistance for lower limb amputees.It can help lower limb amputees to walk freely like normal people and thus improve quality of life,which has important research and application values in the area of rehabilitation aid.Due to the characteristics of nonlinearity and uncertainty of the prosthetic knee system as well as the requirements of comfort and safety for the wearer,robust control methods are considered to guarantee an excellent control performance.The research focus of this paper is the theoretical study of the control method,and the effectiveness of the methods are verified through simulation experiments.This paper proposes neural network based sliding mode control strategies for the prosthetic knee joint.The main contents are as follows:Firstly,in order to improve the accuracy of the active prosthetic knee model and improve the control effectiveness of the controller,the lower limbs of the human body are simplified to a rigid rod model based on the traditional lower limb dynamics modeling method.The movement of the lower limbs in the stance phase and the swing phase are analyzed,and the dynamic model of the lower limbs of the human body are re-derived for the design of the controller.VICON motion capture system is used to collect the motion information of the lower limbs,and the angle changes of the lower limbs joints during the gait cycle are obtained.Secondly,a robust controller based on combining RBF neural network and sliding mode control is designed for the nonlinear uncertainty in the prosthetic knee model system.Considering the chattering caused by traditional sliding mode control method,the RBF neural network is used to estimate the nonlinear term to reduce the controller gains and chattering.The weights of the RBF neural network are adjusted online through an adaptive method to ensure the stability of the system and avoid complicated offline training.Finally,the comprehensive interference of the radial basis neural network interference observer is introduced to estimate the system interference,and the terminal sliding mode control method is used to achieve the control accuracy problem of the prosthetic knee joint system under external interference.The RBF neural network disturbance observer can observe the uncertainty and the comprehensive external disturbances,and the observed value can be employed to compensate the sliding mode controller to effectively improving the tracking accuracy.The controller is designed based on the terminal sliding mode method can improve the convergence speed of the system error and ensure that the system tracking error can converge to zero in a limited time.