Research On Lower-limb Muscle Force Prediction Based On Surface Electromyography

In the social life, the limbs may lose some or all of the athletic ability because of the internal or external factors such as stroke and traffic accident, which have a strong impact on individual, family and society. More attention was paid to the intelligent action auxiliary robot to help the people with damaged athletic ability to regain independent life. The rehabilitation assistant system combined with surface electromyography signal, which enhanced the athletic ability of human body and retained their subjectivity and flexibility, can improve or even solve the problem effectively, which been widely studied and applied.This paper took the skeletal muscle as research objects, which is the power section of human joint. On the basis of analyzing the relationship between the surface electromyography signal and the muscle force, the focus was mainly on the force prediction methods and muscular fatigue compensation strategy. What's more, we verifying the accuracy and practicability of the system by combined with six degrees of freedom lower limb rehabilitation robot. The following presents the prime work:(1) The relationship between electromyography signal and muscle force was studied, also the contraction of skeletal muscle. The joint angle and muscle fatigue which affect the relationship between electromyography signal and muscle force were studied especially. Furthermore, the experiment of signal collection was designed, and the time delay between electromyography signal and muscle force was compensated using the extremum locating method.(2) The force prediction method was studied adopting the support vector regression. Here the muscle action was divided into static contraction and dynamic contraction based on the force characteristics of skeletal muscle, the method of different muscle action was analyzed separately. And the parameters of the models were optimized by genetic algorithm.(3) The muscular fatigue compensation strategy was considered for muscle force prediction. Based on the muscular fatigue experiments, the relationship between muscular fatigue and predicted force was discussed, and the muscular fatigue compensation strategy was employed to extend the force prediction method to muscular fatigue condition. The practicability of the force prediction was further improved.(4) The application of the force prediction was studied, especially for the rehabilitation robot. Combining with the designed force prediction software, the predicted force was used for force-position control. And the stability of the system were verified.In this paper, we designed the signal acquisition protocol for force prediction based on the complex relationship between sEMG signals and muscle force. Then the signals are been synchronization processed. Support vector regression was used for human lower extremity muscle force prediction on the strength of muscle acticity. A muscle force prediction system was established and then we completed the force-speed displacement control experiments on the basis of lower limb rehabilitation robot.