人工神经网络在油藏数据训练中的应用

Artificial Neural Network in the Training of Reservoir Data

由于人工神经网络(ANN)较传统的模式识别方法更具优越性,可以应用到油藏数据的训练中。采用3种神经网络训练函数,即,网络的权值和阈值训练函数、利用Levenberg-Marquardt规则训练函数和阶梯变化梯度算法,分别对油藏数据进行训练,得到成功的训练结果,从而把预测的油藏参数和三维(3D)地震属性联系起来。实验表明,把ANN用于训练油藏数据是可行的。

Because neural network is much more advanced, compared with the traditional pattern recognition, so we introduce the method to the training of reservoir data. Training the reservoir data with three training functions, trainwb, trianlm and trainscg, gain the successful training results, and then correlate the reservoir parameter with 3D seismic attributes. The results show its feasibility in using ANN to train reservoir data.