Análise e paralelização de algoritmos aplicados à identificação de sistemas dinâmicos não-lineares com modelo NCARMA Fracionário

Abstract

The main goal of this paper is to contribute to the performance of a Differential evolutionary algorithm (DE) used in the identification of nonlinear dynamical systems using the NCARMA Fractional model, by comparing algorithms, parallel or sequential, used by the same. The evaluation is made through an application developed specifically for this work with Qt environment using C ++ 14 language and parallel programming techniques. To achieve this goal we selected the main algorithms that are used by software developed in order to compare different methodologies and techniques used to solve them. Different types of algorithms have been tested in order to determine which method does best in each situation, as well as different testing methods, different libraries and sequential versus parallel computation. Finally, the final application has been validated and tested as to their ability to model pre-defined systems considering the total time spent.