Genetic programming approximation in cement paste experimental performance

Abstract eng:
This article introduces results of Genetic Programming used for creation of experimental data approximations together with the search for significant parameters of an affinity cement paste hydration model. Within Genetic Programming trees, the placements of constants still has not been satisfactorily solved. Therefore, the proposed contribution also presents a search for real-valued constants employing Ordinary Least Squares. Twenty trees as results of twenty independent runs of Genetic Programming are presented. From these results the best seven trees are chosen according to specific criteria and the approximations of experimental data are shown. Still, many aspects of Genetic Programming-based symbolic regression are uncovered and especially suppression of the overfitting issues remains unsolved.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket