"Identification of Nonlinear Model Structures by Genetic Programming"
Identification of Nonlinear Model Structures by Genetic Programming
Sprache des Titels:
Identifying nonlinear model structures as a part of analyzing a physical system means generating an algebraic expression (as a part of an equation) that describes the dynamic behaviour of a physical system. In this thesis we present a method based on genetic programming to evolve an algebraic representation of a system's measured input-output response data.
The research described in this thesis was done for the project "Specification, Design and Implementation of a Genetic Programming Approach for Identifying Nonlinear Models of Mechatronic Systems". The goal of this project is to find models for technical, especially mechatronic systems. Our task was to examine, whether the methods of genetic programming are suitable for this challenge or not. The methods of genetic programming, based on the theory of genetic algorithms, introduce new ways of encoding a problem, so that a solution candidate (as an individual of a genetic algorithm's population) can be interpreted as a structure, more generally as a program or even as a formula.
The following tasks had to be carried out and are described in this thesis:
• The problem had to be specified exactly as a genetic programming problem, and we had to find a way of encoding it. Furthermore we had to design appropriate crossover and mutation operators.
• The designed genetic programming model has been implemented as a part of an already existing framework for proto-typing and analyzing optimization techniques.
• Concrete test data was used for evaluating the quality of the results achieved by the implemented genetic programming model. In this connection we also tested new algorithms that are based on evolutionary algorithms and designed to improve the quality of the results. These new generic concepts, too, were evaluated and compared to the classical genetic programming approach.