Neurofuzzy mapping of CPT values into soil dynamic properties

Ever since the early days of soil mechanics there has been a strong tendency to develop correlations between easy-to-obtain parameters and mechanical properties of geotechnical materials. This interest has not decreased as more elaborated methods have been proposed. On the contrary, once a method under development reaches certain degree of sophistication, researchers usually devote a great deal of time to establish simplified procedures. The study presented in this paper follows a similar path in the sense that a data-based knowledge procedure instead of mathematically based methods is used to develop a manner to estimate soil dynamic properties directly from cone-tip penetration resistances. This method is based on a hybrid system integrated by merging artificial neural networks and fuzzy logic. The procedure is evaluated comparing its results with actual measurements of cone-tip penetration resistances and shear wave velocities done in twin borings.