Modeling Method for Robot Servo System Based on IGSA-RBFNN

摘要

To solve the modeling problem for robot servo system, an ensemble approach based on Radial Basis Function Neural Network (RBFNN) and Improved Gravitational Search Algorithm (IGSA) called RBFNN-IGSA, is presented in this paper. Firstly, RBFNN is used to build the servo system model and identify parameters of the system, and the recursive least squares is used in learning process, then during training time, IGSA is used to optimize the thresholds of RBFNN to simplify the complexity of counting process, and reduce the time of servo system modeling.