Development of Stability Formulas for Rock Armor and Tetrapods Using Multigene Genetic Programming

Abstract

Stability formulas for rock armor and Tetrapods were developed using multigene genetic programming, which is one of the machine learning methods. The developed formulas are shown to be more accurate than previous empirical formulas and neural network models. They are also compared against a few independent data that were not used in the development of the formulas. To give an idea of the confidence of the formulas, the equations for 90% prediction error band are presented. The formula for Tetrapods is somewhat less accurate than that for rock armor, probably because the experimental data for rock armor were produced by a single author, whereas those for Tetrapods were produced by several authors, so the data were inhomogeneous. The formula for rock armor includes the crest freeboard of the structure, which was not considered in previous studies, so it can be used for low-crested structures as well.