Liquefaction assessment by artificial neural networks

Abstract

The liquefaction potential of a soil mass during an earthquake is dependent on both seismic and soil parameters. The impact of these soil and seismic variables on the liquefaction potential of soil is investigated through computational and knowledge based tools called neural networks. A back-propagation neural network model is utilized. The back-propagation learning algorithm is a developing computational technique that assists in the evaluation of experimental and field data. The artificial neural network is trained using actual field soil records. The performance of the network models is investigated by changing the soil and seismic variables including earthquake magnitude, initial confining pressure, seismic coefficient, relative density, shear modulus, friction angle, shear wave velocity and electrical characteristics of the soil. The most efficient and global model for assessing liquefaction potential and the most significant input parameters affecting liquefaction are summarized. A forecast study is performed for the city of Izmir, Turkey. Comparisons between the artificial neural network results and conventional dynamic stress methods are made.