Abstract:In order to cope with three special problems in the hydrocarbon prediction using usual pattern recognition,namely,big quantitaty of samples,nonlinear parameters and known sample quantity which varies greatly in different areas,we change usual BP algorithm into the BP algorithm whose r} may be adaptively adjustable, and combine the new BP algorithm with SOM network to form a perfect mufti-parameter hydrocarbon prediction system. This prediction system can avoid the above problems. The prediction system causes fast convergence and can be used to predict hydrocarbon without known sample. Its land and offshore applications show that the method results in good hydrocarbon prediction effect in different areas where exploration degrees are different respectively.
施继承, 聂勋碧. 基于神经网络的油气预测系统[J]. 石油地球物理勘探, 1996, 31(5): 685-692.
Shi Jicheng, Nie Xunbi. Hydrocarbon prediction system based on neural network. OGP, 1996, 31(5): 685-692.