由自组织神经网络和BP网络组成的双重神经网络，能够克服各自单独使用的局限性，可在复杂地区进行储层及油气预测。在BP网络训练时，采用同伦学习算法，可得到全局最优解，且收敛速度很快。实际应用表明，在用自组织神经网络或BP网络不能进行储层及油气预测的地区，采用双重神经网络能取得很好的效果。 The dual neural network composed of self-organizing neural network and BP network can overcome the limitation of their own use and can predict reservoirs and hydrocarbons in complex areas. In BP network training, using the homotopy learning algorithm, the global optimal solution can be obtained and the convergence speed is fast. The practical application shows that the dual neural network can achieve good results in regions where self-organizing neural networks or BP networks can not predict reservoirs and hydrocarbons.