Research On Precision Sensing Method For Human Lower Motion Intention Based On EEG And Emg

The exoskeleton is a kind of wearable mechanical structure together with smart and external mechanical energy.It offers additional power or ability which can protect human body from outside damage and strengthen the body function to improve the power and endurance of body.In recent years,the exoskeleton technology invested a lot of effort,has been reseached deeply to help in action,rehabilitation,manned space engineering,explosion protection and many other fields.At the same time,the precise recognition of human motion intention can grasp people's mental state,physical condition and work requirements,more important,the exoskeleton robot development for these areas brings the possibility of human-robot harmony,the realization of human-robot collaborative control,functional needs that serve people better.The relevant national research project is combined and the precision sensing method for human motion intention based on EEG and EMG that is used to the lower extremity exoskeleton for soldiers,astronauts or the elderly is studied in depth in this paper.Its main research work and the significance of the research results include the following aspects.1)Analysis of human lower motion patterns and the research of multi source sensing strategies.In view of the lower extremity exoskeleton robot may encounter complex terrain and changeable random task,it is difficult to guarantee the human-robot coupling action and high random switching only through specific known actions drived by the established typical gait planning procedures.In order to coordinate the lower extremity exoskeleton robot with human motion,sensing and recognizing the human motion in real time is needed to reach for the purpose of lower extremity exoskeleton robot real-time control.Therefore,it is necessary to simulate and extract the common six kinds of lower limb movements as following study based on the analysis of common movement patterns.Moreover,a multi information fusion method is proposed,which is based on the perception of EEG,the precise sensing of EMG and the deformation correction of optical fiber.It should be emphasized that human intelligence is involved in robot control in order to promote the practical application of the lower extremity robot.2)Research on the method of human lower motion intention by EEG.the sensing and recognizing of human motion intention EEG signal is studied for precision sensingof the exoskeleton robot.The EEG signals of six kinds of imagined motions are collected by experiment.The EEG signals of C3 and C4 channels have been selected and pre-processed.Then,the EEG signals have been decomposed by wavelet transform and wavelet packet transform and the eigenvalues including the wavelet packet decomposition coefficients and the energy coefficients have been extracted.According to the results of feature extraction,the LS-SVM recognition theory is used and the highest recognition rate of 66.11% is achieved for EEG prediction.3)Research on the method of human lower motion intention by sEMG.The lower limb surface EMG as the research object,the rectus femoris,medial vastus muscle,biceps flexor cruris and gastrocnemius muscles have been chosen as surface EMG signal acquisition points through in-depth analysis of the human lower limb motion mechanism.Aiming at the pattern recognition problem of human lower limb motion sEMG signals,the feature extraction method based on wavelet transform and dimensionality reduction method based on the feature of PCA have been researched.A multi class LS-SVM recognition model for the sEMG signal of the lower limb exoskeleton robot has been established,and the parameters have been optimized and discussed.The result has been proved that the classification accuracy of the classifier based on MOC is significantly higher than that of other classifiers.At the same time,it is proved that the selected feature representation method can be used to extract the feature vector which represents the specific action of the human body from the ever-changing,complex EMG signals.And the selected optimal multi class LS-SVM classifier has an ideal classification effect on the feature vectors representing different motions.4)Research on optical fiber deformation sensing method of human lower motion intention.According to the structure and motion mechanism of the lower limbs joints,the motion characteristics of the joints have been analyzed.Combined gait analysis,the knee joint angle is selected to perceive the lower limb movement of the human body.The fiber angle sensor is selected to measure the knee angle and the optical fiber real-time sensing of the human lower limb motion measurement system is constructed.The fractal dimension has been used to characterize the motion characteristics of the lower limbs based on the optical fiber sensing results of the knee joint motion.The fractal feature value vector has been brought into the designed support vector machine,and the data samples of lower limb have been identified by using support vector machine.Using grid search and cross validation algorithm to optimize the parameters of the classifier,the human lower limb motion is recognized as walking,running,ascent,descent,squatting and standing up.The recognition rate can reach 95%.5)Research on the construction of multi source precision sensing system and its information fusion method.EEG signal,in predicting the direction of human movement,can be perceived in advance and get rid of the shackles of the mathematical model better.It is the fastest of all the means of prediction;SEMG fine sensing system can recognize human motion patterns accurately and quickly,which has a more mature theoretical and experimental basis;Optical fiber deformation sensing system,which can obtain the position and posture of the human body joints and limbs with high accuracy and short reaction time,has great potential.It has a much smaller dependence on the mathematical model,which can get more reliable and complete data.Therefor,around for the exoskeleton robot collaborative control,a multi source sensing system of the human motion intention,which integrates EEG detection,sEMG detection and optical fiber detection,has been set up in the laboratory based on the study of the three kinds of human motion intention sensing methods of EEG,sEMG and optical fiber,and its method of multi source information fusion has been studied.Finally,the experimental results show that,the human motion intention multi source sensing system and the method of multi source precision sensing in view of information fusion based on EEG and sEMG,which can be used in the human-robot cooperative control of the lower extremity exoskeleton robot,are proved effective in this paper.