Using genetic programming in nonlinear model identification

In this paper we summarize the use of genetic programming (GP) in nonlinear system identification: After giving a short introduction to evolutionary computation and genetic algorithms, we describe the basic principles of genetic programming and how it is used for data based identification of nonlinear mathematical models. Furthermore, we summarize projects in which we have successfully applied GP in R&D projects in the last years; we also give a summary of several algorithmic enhancements that have been successfully researched in the last years (including offspring selection, on-line and sliding window GP, operators for monitoring genetic process dynamics, and the design of cooperative evolutionary data mining agents). A short description of HeuristicLab (HL), the optimization framework developed by the HEAL research group, and the use of the GP implementations in HL are given in the appendix of this paper.