Evolution of Neurocontrollers in Changing Environments

One of the most challenging aspects of the control theory is the design and implementation of controllers that can deal with changing environments, i. e., non stationary systems. Quite good progress has been made on this area by using different kind of models: neural networks, fuzzy systems, evolutionary algorithms, etc. Our approach consists in the use of a memory based evolutionary algorithm, specially designed in such a way that can be used to evolve neurocontrollers to be applied in changing environments. In this paper, we describe our architecture, and present an example of its application to a typical control problem.