Research On Controling Methods Of Lower Limb Rehabilitation Robot Based On SEMG

Surface electromyography signal (sEMG) is a kind of bio-electricity signalcollected from the surface of human skin by surface EMG electrodes, which canreflect the process of human neuromuscular activity. Not only at the area of clinicalmedicine, but also in the field of rehabilitation medicine, does it have importantapplied values. In this paper, a control strategy for the movement of knee joint oflower limb rehabilitation robot used for the patients with unilateral lower limb motorimpairments because of hemiplegic is designed based on sEMG, to make theindependent rehabilitation by patients themselves possible. Meanwhile, this paperfocuses on the key technologies in three parts which are the feature extraction ofsEMG, the formation of the model of "sEMG-the motion of knee joint" and theelectromyographic control of lower limb rehabilitation robot.First of all, the sEMG from RF, VM and VL is collected respectively by themeans of acquisition method combining posture and motion based on the formationmechanism, the characteristics, the influences and the composition of noise of sEMG.Meanwhile, the angle of knee joint is measurement in a direct way. Afterwards, RMS,iEMG and MPF three features value are extracted to analysis the motion of knee joint.At the end of this aspect, RMS of sEMG is used to analysis the characteristics of thesEMG from quadriceps.Based on the correlation between sEMG and the motion of knee joint, the modelof "sEMG-location" is established, by using the eigenvalues from sEMG as the inputof BP neural network, and the angle signal from knee joint as the output of neuralnetworks. The model of "sEMG-velocity (acceleration)" is also established, by usingsEMG eigenvalues and its first (first and second) order derivative together as the inputof neural network, and the final velocity (acceleration) of the knee joint as the outputof neural network, based on the model of "sEMG-location". The quantitativeidentification to the location/velocity/acceleration of knee joint using the eigenvaluesof sEMG is achieved.Finally, a passive control strategy of lower limb rehabilitation robot is designedin two ways of threshold control and pattern recognition take advantage of theplatform of lower extremity robot, to achieve the control to the movement of kneejoint of lower limb rehabilitation robot. First, the motion control of small angle jog ofknee joint is achieved by designing the threshold control strategy based on theresponse threshold and reset threshold. Besides that, based on three quantitativerelational models,"sEMG-location","sEMG-velocity" and "sEMG-acceleration", to finish the design for three forms of pattern recognition control strategies, achievingthe process control to a broader range movement of knee joint of lower limbrehabilitation robot.