| The loss of lower limbs will bring obstacles to patients,especially the above-knee amputation will seriously affect their ability to walk,and bring a heavy burden to the family and society.Wearing prostheses with good performance can greatly improve the mobility of amputees and their participation in social activities.At present,due to the low cost of passive prostheses,passive prostheses are the most widely used.However,patients wearing passive prostheses often walk with unnatural gait,high energy consumption,and it is difficult to walk smoothly in different terrains.Intelligent lower limb prosthesis can automatically adapt to the external environment by detecting the external environment and recognizing the human motion intention,making the wearer's gait more natural and more labor-saving.Therefore,intelligent lower limb prostheses can bring more functions to patients.Powered(active)lower limb prosthesis is a research hotspot in the intelligent lower limb prosthesis field because it has power output and can better compensate for the functions of amputees.The lower limb power prosthetic system existing now usually obtain the motion intention of amputees through the amputees' movements,such as slowing down,stopping,pressing electronic switches or making limb movements unrelated to walking(exaggerated hip extension or forward/backward swing of the prosthesis,etc.),so as to realize the conversion of different gait control strategies.This kind of non-intuitive control mode can not achieve stable,smooth and natural walking for the prosthesis under different road conditions.Therefore,the lower limb prosthesis must understand the intention of human movement while perceiving the external environment,and control the prosthesis to walk in a proper gait.It is necessary to solve the key problem of motion intention recognition and realize the information interaction between prosthesis and human.Aiming at the problems of motion intention recognition of intelligent lower limb prosthesis,the following research work has been carried out from the aspects of new sensors,high-quality portable signal acquisition systems,limb motion recognition,environment and gait recognition.Firstly,according to the phenomenon that muscle shape changes(MSC)in the process of movement,this dissertation proposed to use a nano gold flexible and stretchable material to fabricate sensors that could detect the MSC signal by attaching to the surface of the skin.Then a multi-channel MSC signal acquisition system was developed for the sensors,and used to detect the limb movements.Finally,the influences of the size,placement and sampling rate of the MSC sensor on the accuracy of limb motion intention recognition were studied.The experimental results show that as a new information source for motion intention recognition,the accuracy of four-channel MSC signal for limb motion intention recognition can reach over 95%,and proper selection of sampling rate and features for motion intention recognition can greatly reduce computation.Secondly,aiming at the complexities caused by multi-mode and multi-channel signals in some motion intention recognition systems,and according to the phenomenon that the MSC signal and EMG signal are emitted in the process of muscle movement,this dissertation proposed a method of fabricating nano gold flexible and stretchable material into EMG/MSC hybrid signal sensors to detect the EMG and MSC signals on the same layer of the sensor simultaneously.Then,using the frequency division multiplexing technology,a multi-channel EMG/MSC hybrid signal acquisition system was developed.Using the above sensors and system to measure EMG signals,the signal-to-noise ratio was about 3dB higher than that of the targeted commercial equipment device,which proves that the signal quality has reached the same level as the commercial equipment.The experiments of testing the system proved that the system had the same signal quality level as the commercial equipment.At the same time,the redundancy of multi-mode and multi-channel signals were studied,and the feasibility of using hybrid signals for motion intention recognition with fewer sensors and fewer channels was validated.Thirdly,given a large amount of computation in the method of using image and video for the environment and gait recognition,this study proposed to use a single line light detection and ranging(LiDAR)module combined with an Inertial measurement unit(IMU)to scan environment automatically with the swing of the lower limb when walking.Therefore,a portable and wearable multi-mode signal acquisition system was developed.Accordingly,a new algorithm was used to complete the environment reconstruction.The verification results of environment reconstruction show that the average accuracy of the five types of terrain is more than 90%.To solve the problem of ground reaction force measurement in gait recognition,this study used the flexible pressure sensor array to detect the gait phase and then estimated the ground reaction force with an artificial neural network(ANN).This research provides a method for gait recognition and control of the intelligent prosthesis.Fourthly,based on the previous researches,aiming at the problems of time delay,transmission instability and increasing of computation when using the wireless transmission in motion intention recognition,this study proposed to use the distributed data collecting and computing method to distribute the data preprocessing and feature extraction to sub-systems for processing.This not only reduced the amount of data transmitted by 80%,which ensured the stability of wireless communication,but also reduced the requirements of the host's computing performance,making the motion intention recognition system more practical.Finally,according to the general control structure of intelligent lower limb prosthesis,the experimental verification platform for intelligent lower limb prosthesis was completed,and some performances were validated. |
PDF Full Text Request