“…The measure node may be described by forward NN or dynamic recursive NN. Above all approximately range of the nucleus steelyard's measure lag is confirmed example for 6s 10s so lag time is 6s(equals tree control periods) as its principle is choosing the least value in range [7,8] , And then the forward NN is trained again and again according to the sample data structured by formula [y m ( k) y m (k-4), y m (k-5),y m (k-6),y m (k-7),u(k-4), u(k-5),u(k-6),u(k-7)] quoted from (1.1), so the high precision is obtained. For recursive NN, iteratively training the sample structured by [ y m ( k),y m (k-4),u(k-4)] also meets the need.…”