A noise-suppressing Newton-Raphson iteration algorithm for solving the time-varying Lyapunov equation and robotic tracking problems

Abstract

The Newton-Raphson iteration (NRI) algorithm is a prevalent computational methodology in many fields and can be used to solve the time-varying Lyapunov equation. However, the performance of the traditional NRI (TNRI) algorithm is severely degraded under noisy conditions. To overcome this weakness, a noise-suppressing NRI (NSNRI) algorithm is proposed in this paper. By utilizing the Kronecker product theorem, the time-varying Lyapunov equation can be transformed into a linear matrix equation. Based on that linear matrix equation, the related theoretical analyses of the convergence and the noise-suppressing property of the NSNRI algorithm under various noise conditions are provided. To verify the theoretical analyses, numerical simulations for solving the time-varying Lyapunov equation and an application to manipulator motion tracking are presented. For comparison purpose, the TNRI and two zeroing neural network (ZNN) algorithms are also introduced in these simulations. As indicated by the simulation results, the NSNRI algorithm is superior in terms of the convergence accuracy and the robustness to noise.