基于BP神经网络的饱和砂土液化判别方法_李方明.pdfVIP

2 : BP # 109# tmi e; ( 3 two a lgorithm s of the BP neural newt ork m odel both have the higher accuracy fo r liquefaction po tent ia l estmi a tion of sand and LM a lgorithm is faster than the BP algorithm, but needs m ore EMS m em ory in the process o f calcu la tion, and therefore, LM a lgorithm is recomm ended in this paper; ( 4 the liquefaction potential est mi ation for other sam ples under u sing the w e ights and thresholds o f the BP neura l netw ork m odel presented in th is text, the eva luated result m aybe conservative; ( 5 considering the m a in factors to affect sand liquefaction po tentia,l the di-f ficulty of acqu iring the indexes, and under the condit ion that the indexes selected are consistent w ith those in Code for se ism ic design of bu ildings in China, M 4 or M 5 a neura l netw ork m ode l is recomm ended in th is paper. These m ode ls are very easy in applicat ion, m oreover, the accuracy of the m odels for liquefaction potentia l estmi at ion o f sand is higher than that in Code for seism ic design of bu ildings in Ch ina. K ey words: sand liquefaction; BP neural netw ork; LM a lgorithm; faster BP algorithm; liquefaction po tentia l , , , , , [ 1~ 5] , , , ; , , , , , , , , , , , , , , , , BP , ( I ) (D ) ( a ) c (d ) ( d )(H ) ( R ) w s 0 0 (N )( d ) (C ) (N ) 63. 5 50 u 1 3 11, BP LM BP , , BP , , 1 BP BP , , [ 6][ 7] BP , , , , BP , , : T abusearch