地震属性包含的地球物理信息十分丰富,但地震属性种类繁多,并且与储层特征对应关系复杂,单属性分析难以确保预测的准确性。人工神经网络方法具备较强的非线性映射能力,使用该方法可以综合利用多属性进行油气预测,提高预测精度。该文采用梯度下降神经网络算法,避免陷入局部极小值,有效加快网络收敛速度,使网络达到全局最优,提高了预测效果。 Seismic attributes contain very rich geophysical information, but there are a great variety of seismic attributes and complex relationships with reservoir characteristics. It is difficult to ensure the accuracy of prediction with single attribute analysis. Artificial neural network method possesses strong ability of non-linear mapping. Using this method, we can comprehensively use multiple attributes to predict oil and gas and improve prediction accuracy. In this paper, gradient descent neural network algorithm is used to avoid falling into local minima, to speed up the network convergence speed effectively, to make the network reach the global optimum and to improve the prediction effect.