Evolutionary computation, an alternative solution to nonlinear optimization problems

Abstract

Analytical and numerical methods have been applied to solve problems in engineering. However, in some practical cases, they usually fail when there is a certain degree of complexity, for instance, when there is a certain lack of information about the elements of the system and when the unknowns are functions. These types of problems are often called nonlinear optimization problems. As an alternative to solving them, evolutionary computation methods are usually implemented, although they do not generate an exact solution, and provide a series of approximations that are generally feasible. In this context, the objective of this work is to briefly highlight the most typical characteristics of these type of algorithms, some advantages, and the importance of its use today. Due to the wide variety of existing methods, it would become complex to explain all of them in detail, so only a description of the differential evolution (DE) algorithm will be made because it is one of the most used and because there is current research that seeks to improve its performance.