Design And Research Of Lower Limb Rehabilitation Exoskeleton Based On Motor Imagery EEG Signals

Many people experience muscle weakness or paralysis after a stroke,which can affect their balance and mobility,usually in just one arm or leg.The appearance of exoskeleton can help patients recover health through rehabilitation training,but the patient's active participation is low.A lower limb rehabilitation exoskeleton based on motor imagery electroencephalogram(EEG)is designed,and uses the patient's EEG to control the brain-computer interface to provide a communication system between the human brain and exoskeleton equipment.The mechanical structure of the lower limb rehabilitation exoskeleton is designed.The EEG feature extraction,feature recognition and optimization classification algorithm of the motor imagery are studied,and combines EPOC+ to complete the brain-controlled lower limb rehabilitation exoskeleton experiment.Firstly,The motion mechanism of the joints of the lower limbs of the human body is designed to design the structure of the lower limb exoskeleton model.Then,the driving method and an improved power output scheme are proposed.A static analysis of the lower limb rehabilitation exoskeleton was performed to verify the strength and stiffness of the exoskeleton structure.The D-H model was established for kinematic analysis,and the relationship between joint rotation angle and terminal posture was obtained.The simulation of the skeletal movement space and acceleration experiments verified the rationality of the structure.The corresponding dynamic model is established for the supporting state of the legs during walking.Through the virtual prototype test,the angle and torque curves of each joint are obtained.Secondly,in the research of EEG feature extraction,the brain-computer interface experiment examples were analyzed,including the relevant information such as the source of the data set,the collection device,the collection process,the placement of the electrode,etc.,and the phenomenon of ERD/ERS generated during motor imagination.A classification scheme for motor imagery EEG signals was designed and the evaluation criteria of the classification system were given.Filtering and independent component analysis are used to remove EEG high-frequency interference and artifact signals and the pre-processed signal is obtained.Using the common space mode and the discrete wavelet transform method to extract the features of the EEG,the ? in descending order and the decomposition and reconstruction signals are obtained.In the research of EEG feature recognition problem,for the problem of low accuracy of EEG two-class and multi-class classification,the linear discriminant analysis and support vector machine algorithm with high classification accuracy are analyzed.Using the particle swarm optimization method to optimize the key classification parameters of the support vector machine,the classification optimization scheme is designed,and the support vector machine classification model based on the particle swarm is established.Optimize the penalty parameters and kernel function parameters of the design model,get the fitness function value of the iterative process,and experiment with MATLAB to get the best parameter combination.A classification experiment of EEG signals was carried out.The experimental results showed that the average accuracy rate of SVM classification and recognition based on particle swarm optimization reached 90.3%,which can effectively improve the classification accuracy of electromagnetism EEG signals.Finally,a brain-controlled lower limb rehabilitation exoskeleton system based on EPOC+ was built.A PC-side exoskeleton control program was written,and combined with EPOC+ EEG,a brain-controlled exoskeleton classification exercise experiment was carried out.The experimental results verified the effectiveness of the brain-controlled exoskeleton and the feasibility of the control scheme,and initially realized the EEG-controlled movement of the brain-controlled exoskeleton.