Soil Classification from Piezocone Penetration Test Using Fuzzy Clustering and Neuro-Fuzzy Theory

The advantage of the piezocone penetration test is a guarantee of continuous data, which are a source of reliable interpretation of the target soil layer. Much research has been carried out for several decades, and several classification charts have been developed to classify in situ soil from the cone penetration test result. Even though most present classification charts or methods were developed on the basis of data which were compiled over many countries, they should be verified to be feasible for local country. However, unfortunately, revision of those charts is quite difficult or almost impossible even though a chart provides misclassified soil class. In this research, a new method for developing soil classification model is proposed by using soft computing theory—fuzzy C-mean clustering and neuro-fuzzy theory—as a function of 5173 piezocone penetration test (PCPT) results and soil boring logs compiled from 17 local sites around Korea. Feasibility of the proposed soil classification model was verified from the viewpoint of accuracy of the classification result by comparing the classification results not only for data which were used for developing the model but also new data, which were not included in developing the model with real boring logs, other fuzzy computing classification models, and Robertson’s charts. The biggest advantage of the proposed method is that it is easy to make the piezocone soil classification system more accurate by updating new data.