Abstract: Most of the existing bearing capacity calculation methods of foundation soils are empirical formulas in the design specifications or manuals. Each of them has its own applicable range and it has regional in characteristics. It is therefore necessary to select different empirical formulas to reliably determine the foundation bearing capacity values according to different insitu test methods and considering local experience. In order to determine the characteristic value of the bearing capacity of foundation soils more efficiently, this paper presents an artificial intelligencebased method for determining the bearing capacity value based on the BP neural network. Throughout the characteristic values of the bearing capacity of foundation soils obtained by using different test methods in the project, a training set and a test set are established. Test results show that the bearing capacity of the foundation soil determination method based on BP neural network established in this paper has advantages of high efficiency and reliability, and it can quickly and effectively give the characteristic value of bearing capacity value based on test data.

Key words: Bearing Capacity of Foundation Soils, Artificial Intelligence, Insitu Testing, Determination Method