一种可用于储层参数预测的基于神经网络的遗传算法

A Generetic Algorithm Based on Neural Network for Reservor Parameter Predication

储层物性参数的展布一直是油藏描述的关键问题和难点。在遗传算法与神经网络结构模型相结合的基础上,设计了一种用遗传算法训练神经网络权值的方法,并把这种方法用于储层参数的预测。实验表明该神经网络具有较强的预测能力和实用性。

Distribution of the reservoir physical parameter always is the bottleneck problem and difficulty of reservoir description. Based on the combination of Genetic Algorithm and Neural Network, an approach to learn the weights of Neural Network is put forward, which applies to predict reservoir parameter. Experiments show the neural network occupies preferable predication performance and practicability.