Active vibration control of a modular robot combining a back-propagation neural network with a genetic algorithm

In this paper, a genetic algorithm based back-propagation neural network suboptimal controller is developed to control the vibration of a nine-degrees-of-freedom modular robot. A finite-element method is used to model the modules of the robot, and the entire system dynamic equation is established using the substructure synthesis method. Then the joint stiffness parameters are identified based on the experimental modal analysis experiment. After modeling the whole structure with the models of the robotic modules and the joint parameters, simulations of the vibration control for the modular robot in several configurations are carried out. It is shown that the control method presented in this paper is effective at suppressing the residual vibrations of the modular robot.