Research on An Method of Muscle Force Prediction Based on Dynamic Fuzzy Neural Network

Muscle force is the energy source of human body motion. It is not only the basic data basis for the rehabilitation evaluation of human motor function, but also can provide a theoretical basis for the control of rehabilitation robots and the design of bionic joints. However, muscle force cannot be measured directly. Therefore, this paper combines the surface electromyogram(sEMG) with the joint angle, and propose a method of muscle force prediction based on dynamic fuzzy neural network (D-FNN). Firstly, subjects were required to perform flexion and extension of elbow movement in uniform speed and the time domain characteristics were extracted from the sEMG of subjects' biceps brachii. The elbow angle was collected at the same time. Next, the contraction force of the biceps brachii was obtained based on the OpenSim platform which can used to simulate human motion. Then, based on D-FNN, the mapping from sEMG and elbow angle to muscle force were obtained. Finally, the effectiveness of the method was verified by experiments.