Research On The Phase Recognition Technology Of Human Lower Limb Movement For Human-Machine Coordination

The human gait parameter wearable lower limb exoskeleton robot is an assistant walking device,which can help the human body to enhance the movement function and reduce the energy consumption.The exoskeleton device needs to achieve a high degree of human-computer coordination with the human body in the process of walking,so as to meet the comfort of wearing and increase the effect of assistance.The accurate recognition of gait phase is the premise and foundation of the precise control and humancomputer coordination of the lower limb exoskeleton robot.In order to solve the problems of low accuracy of gait recognition,which can not meet the motion effect of human-computer coordination.In this paper,a new hidden markov model gait phase recognition algorithm and a deep convolution long short term memory networks gait phase recognition algorithm are proposed based on the human-machine coordination of the lower extremity assisted exoskeleton robot.Through system construction,simulation analysis,theoretical research and experimental verification,the proposed gait phase recognition algorithm is effectively verified.The main research contents,methods and conclusions are as follows:1.The design of exoskeleton system of lower extremity.Based on human-machine bionics,a kind of light-weight bionic single leg assist exoskeleton system is designed,which is comfortable and natural to wear.According to the actual movement process,the mechanical structure of exoskeleton system is designed.The 3D model is built by Solid Works,and the stress bar is analyzed to ensure the reliability and safety of the structure design.The wearable human gait perception system and control scheme are designed.2.Gait phase recognition based on machine learning algorithm.A gait classifier model is built to recognize and analyze the preprocessed data.SVM,KNN and BP neural network are used to build a gait classifier.In addition,PSO optimization algorithm is used to optimize the parameters of support vector machine,eliminating the local optimization problem of parameter selection,and carrying out the identification and comparison experiments of KNN,SVM,PSO-SVM,BP and other different algorithms.3.A N-HMM gait phase recognition algorithm is proposed.In view of the complexity of traditional HMM model caused by the increase of data and the better solution of model parameter self-adaptive,an N-HMM gait phase recognition algorithm is proposed.By improving the forward backward algorithm and Baum Welch algorithm in the model,the parameters of the model are modified by the adaptive algorithm.Experiments show that the optimized N-HMM algorithm improves the accuracy of gait phase recognition and the adaptability of gait data.4.A DC-LSTM gait phase recognition algorithm is proposed.Aiming at the difficulty of feature extraction and over fitting of traditional LSTM data,the paper improves and optimizes the LSTM gait phase recognition algorithm,and proposes a DC-LSTM gait phase recognition algorithm.Through convolution and pooling of input data,the deep feature extraction of data can be satisfied.Combining the gradient of parameters and Min's distance,the convolution kernel is improved.The average recognition accuracy of the algorithm is up to 95.6%,which improves the recognition effect of human gait phase.5.Human machine coordinated wear test experiment.The subjective and objective parameters are evaluated through the human-machine coordinated wearability test.The average value of subjective index is greater than 85,reaching a good level;compared with SVM algorithm,the optimization algorithm DCLSTM,N-HMM and PSO-SVM proposed in this paper reduce the mean square error of knee angle by 56%,35% and 17% on average compared with SVM.The experimental results show that the optimization algorithm proposed in this paper is effective to achieve better human-machine coordination of motion,and to achieve the improvement of wearing comfort.This research can identify the human phase more accurately and improve the accuracy of gait phase recognition algorithm.Through the human-computer coordinated wear test experiment,it can achieve a more ideal goal of human-computer coordinated movement.The research of this paper will lay a solid theoretical and technical foundation for the integration research and application development of exoskeleton robot.