���û��� K��means �����㷨�� ��BF �����編���ɺ���ض������ݻ�������Խ���ʶ���ۺ�������Ȼ٤���������ӡ�����ʱ��ܶ�������ʵ�ʵ�ʲ⾮���ϣ��������������ʶ��Ļ�����BF �����硣ѡȡ����о����м��¼�����ɾ���������֤���÷��������ʶ����������ҡ������ҵ� 6 �ֻ���ң�ʶ��׼ȷ��ƽ���ɴ� 70%���ϡ�

A new hydrid algorithm for training ��BF network based on moving K- -means clustering algorithm was adopted to identify the types of igneous rocks in eastern depression of Liaohe oil field�� By synthetically using natural gamma��neutron��acoustic��density and resistivity logging data��the basic ��BF neural network of igneous li- thology identification has been established�� Some wells with cores and cuttings are selected to test��the result shows that the method clearly identified the basalt and trachyte�� 6 kinds of igneous rocks�� The accuracy of recognition rate is more than 70%��