Simulation research of exoskeleton suit based on Neural Network Sensitivity Amplification Control

Traditional exoskeleton suit need to install many complex sensors between the pilot and the exoskeleton system to measure the human-machine interactive information, which decrease the comfort of the pilot. Sensitivity amplification control can control the exoskeleton suit to trace the pilotpsilas movement as well as need no sensors between the pilot and the exoskeleton. However, sensitivity amplification control seriously relies on the systempsilas dynamic model and it is hard to build the exoskeleton suitpsilas dynamic model exactly because the exoskeleton suit is a multi-body, multi-degree and nonlinear system. So the dynamic model of the swing leg of exoskeleton suit was identified by BP neural networks, which simplified the procedure of building the system model. Neural network sensitivity amplification control was proposed and its feasibility was validated by simulation based on Simulink and SimMechanics toolbox in Matlab.