A noise-tolerant fast convergence ZNN for dynamic matrix inversion

Abstract

In this paper, a noise-tolerant fast convergence zeroing neural network (NTFCZNN) adopting a new power-versatile activation function (PVAF) is proposed and analyzed for solving dynamic matrix inversion (DMI). The proposed NTFCZNN has not only the fixed-time convergence ability but also the strong noise-tolerant ability when it is used to solve DMI problems. The new NTFCZNN and the original ZNN (OZNN) activated by the recently reported universal AF (versatile AF) are simultaneously used in the matrix inverse problem under the context of all kinds of distractions. Then, through a comprehensive comparative analysis of the simulation results, the powerful anti-disturbance ability of NTFCZNN is better highlighted. Both the theoretical verification under various circumstances and the sharp contrast simulation experiments are sufficient to show that the NTFCZNN model has high reliability and noise resistance in the process of solving the DMI problems.

KEYWORDS:

• Recurrent neural network (RNN)
• OZNN
• NTFCZNN
• DMI

AMS SUBJECT CLASSIFICATION:

• 60-08 65C20

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Research Foundation of Education Bureau of Hunan Province, China: [Grant Number 20B216]; the Natural Science Foundation of Hunan Province, China: [Grant Number 2020JJ4315].