NEURAL-NETWORK - AN ALTERNATIVE TO PILE DRIVING FORMULAS

Artificial neural networks are capable of learning complex nonlinear r elationships from a large amount of accumulated data, and similar to h uman brains, are noise and fault tolerant. This unique capacity sugges ts that neural networks would be very useful in certain geotechnical e ngineering applications. A back-propagation network is set up and trai ned to predict the pile bearing capacity from dynamic testing data. Th e trained network produces better results than a pile driving formula approach. The effects of various network parameters on the network res ults are examined in detail. The general understanding developed is po tentially useful for the application of neural networks in other geote chnical engineering problems.