A neural circuit model of tracking control for continuous-time nonlinear dynamic systems

Abstract

A neural circuit model of tracking control for unknown nonlinear dynamic systems is proposed. A first-order differential equation with variable structure and an output equation are used to describe the model. The model gives a possibility to reach a finite convergence time to working modes and limited tracking error. It does not need learning phase in offline mode. The model uses only outputs of the system and object to minimize tracking error of object trajectory. It has simple structure and can be used if internal dynamics and parameters of control system are unknown. Results of computer simulations of the model applications for optimal tracking control of rotation angle of two-link planar elbow manipulator confirming theoretical statements and illustrating its high performance are provided.