Meta-heuristics applied to structure selection and parameter estimation of polynomial nonlinear state space models
Meta-heuristics, Nonlinear systems, Systems identification, Selection of structures, Parameter estimation, State space models
The system identification aims to determine mathematical models capable of descri- bing the dynamic characteristics of the system from their observations. Generally, the identification process is divided into the stages of experimental data collection, model structure determination, parameter estimation and model validation. In this work, we in- vestigate the most complex stages of the identification process, the problems of structure determination and parameter estimation of the model, specifically, we study these pro- blems for polynomial nonlinear state space models. For this, an algorithm was proposed for determining the structure and estimation of model parameters based on optimization techniques known as meta-heuristics. Unlike traditional methods, meta-heuristics use a set of possible solutions and strategies, usually based on nature, to find the solution of the evaluated problem. Among the techniques studied are the genetic algorithm, particle swarm optimization and the bat algorithm. Thus, the application of the identification al- gorithm aims to find a model or set of models in polynomial nonlinear state space capable of representing the global dynamics of the evaluated system.