Neural Network Based Internal Model Decoupling Control of Three-motor Drive System

Abstract

The multi-motor drive system is a multi-input–multi-output, non-linear, and strong-coupling system. Its high-precision coordinated control performance can meet the requirements of many drive applications, such as urban rail transit, paper making, electric vehicle drive, and steel rolling. To decouple the velocity and the tension of the three-motor drive system, a new control strategy is proposed by incorporating two-degree-of-freedom internal model control with the back-propagation neural network generalized inverse. First, the composite pseudo-linear system is formed by a cascading connection for the neural network generalized inverse with the original system. Second, a two-degree-of-freedom internal model control method is introduced to this pseudo-linear system. Finally, both simulation and experimental results are given for verification. The proposed strategy not only effectively decouples the velocity and tension, in which this multi-input–multi-output non-linear system is transformed into a number of single-input–single-output linear subsystems with open-loop stability, but it also enhances the tracking performance of the system.

Keywords:

• decoupling control
• three-motor drive
• neural network
• generalized inverse
• internal model control