Artificial Neural Network For Estimation Of Parameters And Solving Dynamic System Of Maize Foliar Disease

Abstract

In this thesis, we applied Artificial neural network (ANN) for approximating the solution and estimating the parameters of first order systems of differential equations. First, the network trained to determine the learning parameters, the weights and biases. Once the learning parameters are known, it is possible to determine the ANN output and write the solution of the system in terms of the ANN output. We impose to this solution a minimization condition to obtain the unknown parameters of the system. A vectorised algorithm for this method was provide and test on non-linear systems of ODEs. Finally, we tested the accuracy of the ANN approach to approximate the solution and estimating the parameters by comparing it with theoretical values of the systems.