Design and Validation of New Discrete-Time Zeroing Neural Network for Dynamic Matrix Inversion

In the previous work, the zeroing neural network (ZNN) with continuous-time and discrete-time formulations has been studied for dynamic matrix inversion. In this paper, the discrete-time formulation of ZNN is further investigated for computing the inverse of dynamic matrix. Specifically, a special numerical difference rule is established on the basis of Taylor series expansion. By using such a difference rule to discretize the continuous-time ZNN model, the new discrete-time ZNN (DTZNN) model is thus proposed for dynamic matrix inversion. Comparative numerical results are provided to validate the effectiveness of the proposed DTZNN model compared with the existing DTZNN models.