Zeroing Neural Network for Solving Hybrid Multilayered Time-Varying Linear System

Abstract

Hybrid multilayered time-varying linear system is a challenging problem, which has complex structure and time-varying characteristics. In order to solve this complex problem, we use the method of zeroing neural network to analyze the equivalence between different layers. According to the equivalent results, a continuous zeroing neural network model is proposed. In order to satisfy real-time computation and facilitate the hardware implementation, a five-instant time-discretization formula with high accuracy is proposed for the discretization of continuous zeroing neural network model. Then, corresponding discrete zeroing neural network model is proposed to solve hybrid multilayered time-varying linear system. It is worth noting that discrete zeroing neural network model can predict future-instant solution and satisfy the real-time calculation. Numerical experimental results show the effectiveness of proposed model.

Keywords

• Continuous zeroing neural network
• hybrid multilayered time-varying linear system
• time-discretization formula
• robot manipulator control