论文部分内容阅读
本文全面论述了神经网络专家系统及其开发步骤，并利用所开发出的神经网络专家系统，以胜利樊家油田１０口已知油井的储层参数和原油参数及单井日产量为学习样本，成功地预测了２口预测井的单井日产量。说明利用神经网络专家系统可以克服常规的统计方法和模糊方法无法正确确定各参数隶属度及权重分配的缺点，并指出通过神经网络专家系统进行油田预测的准确、快速及有效性。
This paper comprehensively discusses the neural network expert system and its development steps. Based on the developed neural network expert system, taking reservoir parameters, crude oil parameters and single well daily production of 10 wells in Fanjia Oilfield of Shengli as learning samples, We have successfully predicted single-well daily production for the two predicted wells. It indicates that the neural network expert system can overcome the shortcomings of the conventional statistical methods and fuzzy methods in determining the membership degree and weight distribution of each parameter and points out the accuracy, speediness and effectiveness of predicting oilfield by neural network expert system.