关键词: 自组织特征映射神经网络, 储层, 延吉盆地

Abstract: Beginning with analyzing advantage and disadvantage of the existing methods used in the reservoir quality synthetic evaluation, the authors put forward to applying SOFM (self-organizing feature map) neural network method to be used in the oil reservoir quality synthetic evaluation on spatial database and have appraised the Dalazi group reservoir of the Yanji basin using the method proposed. The results indicate that the first category reservoirs were mainly developed in the Chaoyangchuan central-depression and at the western margin of Yan-D4 well with oval-shaped distribution pattern and at the western margin of the Chaoyangchuan depression-that is , in between Yan-D6 and Yan-3 with crescent-shaped distribution pattern; The blocks of the second category reservoirs were quite big and they are collectively distributed at the peripheral of the Chaoyangchuan-depression and the Maoershan bulge, are irregularly distributed at the eastern and southern margins of the Qingchaguan depression and at the northern margin of the Dexin depression; The third category reservoirs were mainly developed in the Chaoyangchuan centraldepression and to the south of the Chaoyangchuan Town, and they are also distributed at the eastern margin of the Qingchaguan depression as small, sporadic strips and patches; and also irregularly distributed in the Dexin depression; The fourth category reservoirs are mainly developed in the western uplift and the Lianhuadong ramp area, also are sporadically distributed in the center of the Qingchaguan depression; Areas belong to the fifth category of poor reservoirs.

Key words: self-organizing feature map neural network, reservoir, Yanji basin