复合基神经网络在水声目标分类识别中的应用-声学技术.PDF

方世良　陆佶人 ( , 2 10018) , , , BP RBF : , , , The application of a combined basis neural network in the underwater acoustic target classification and recognition FANG ShiliangLU Jir en ( Depart ment f Radi Engineering, S utheast Universit y, Nanjing 2 10018) In this paper , a c mbined basis neural netw rk is pr p sed f r underwat er ac ustic target classificati n. - It is c mp sed f radial basis functi ns and multi lay er perceptr ns w h se characteristics n the classificat i n are different . It has been dem nstrated by the rec gniti n experiment t hat the netw rk has higher c rrect rec gnit i n rate and bett er adapt ability t untrained targets than BP r RBF netw rk. Key words : target classificati n, underw ater ac ustic, neural netw rk, basis functi n 1　引　言 , , , , , , : ( 1) , ( ) ; ( 2) BP ( ) - , , RBF ‘ 2 , , BP , BP , X : , , 1960 8, : 97-4-3; : 97-10-16 54 17 2 ( 1998) , Rj2 j , N 1 - - , , - 1,j x w RBF , , Euclidean , N , - - 2 x - w 1,j = ûx ( i) - w 1,j ( i) û ( 2)