Uncertain nonlinear system control with fuzzy differential equations and Z-numbers

In this paper, the solutions of fuzzy differential equations (FDEs) are estimated by using two types of Bernstein neural networks. Here, the uncertainties are in the form of Z numbers. Firstly, we transform the FDE to four ordinary differential equations (ODEs) at par with Hukuhara differentiability. After that we develop neural models having the structure of ODEs. By using modified backpropagation technique for Z number variables, the training of neural networks are carried out. The results of the simulation illustrate that these innovative models, Bernstein neural networks, are efficient to approximate the solutions of FDEs which are on the basis of Z-numbers.