Application of adaptive-network-based fuzzy inference systems to classification of swelling-shrinkage grade of expansive soils

The expansive potential of expansive soils is decided by many factors. It is a multi-factors evaluation problem for classification of swelling-shrinkage grade of expansive soil. The theory of adaptive-network-based fuzzy inference systems (ANFIS) was introduced and subtractive cluster method was applied to deciding fuzzy inference rules. After that, an ANFIS was proposed and applied to two engineerings. Good results were obtained for many test examples. It is concluded that ANFIS can function as BP neural network and fuzzy comprehensive evaluation on classification of swelling-shrinkage grade of expansive soil; and moreover, ANFIS is clearer than BP neural network in network structure and possesses of self-study ability compared with fuzzy comprehensive evaluation. The application of ANFIS on classification of swelling-shrinkage grade of expansive soil is feasible.