Application of Genetic Programming for Estimation of Soil Compaction Parameters

The aim of this study is to propose two numerical models by a well-known soft computing method, Genetic Programming (GP), for the estimation of soils compaction parameters. Genetic Programming is a pattern recognition approach that has the ability of modeling the non-linear behavior of complex engineering problems. The input variables were the soil classification properties, and the outputs were the Optimum Moisture Content (OMC) and Maximum Dry Density (MDD). To provide model, a database including properties of different soils classified as CH, CI, CL, GC, GM, MH, MI, ML and SC was used. In addition, a new Multiple Linear Regression (MLR) based formula using the database, compared with the GP based model. Study results revealed that the proposed formula by GP can predict the compaction parameters of soils in a highly precise manner, and its outputs were in satisfactory conformity with real test results. Performances of the proposed models evaluated using the regression statistical analyses. The proposed formulae can be useful for the preliminary design of engineering projects and are more useful for cases with time and financial limitations.