基于R/S分析的BP网络方法预测气井产量  

BP Network Method Based on R/S Analysis to Predict Gas Well Production

作　　者：孙俊义 徐正华 张颂颂 田浩年 侯杰 SUN Junyi;XU Zhenghua;ZHANG Songsong;TIAN Haonian;HOU Jie(Linfen Branch,Petrochina Coalbed Methane Co.,Ltd.,Taiyuan Shanxi 030032)

机构地区：[1]中石油煤层气有限责任公司临汾分公司,山西太原030032

出　　处：《中国科技纵横》2024年第14期114-116,共3页China Science & Technology Overview

摘　　要：本文利用R/S分析法对气井产量进行分析,提出了历史数据的时间序列和平稳时间序列两种不同表达方式,并利用BP网络方法预测出了气井产量。为验证这种方法的有效性,分别对BP网络与BP神经网络、贝叶斯算法三种方法进行了比较。结果表明,贝叶斯算法获得的预测结果比其他两种方法更准确、更符合实际。贝叶斯算法得到的预测结果与实际值的平均误差分别为0.67%和0.33%,而采用传统BP神经网络得到的预测结果与实际值的平均误差分别为1.71%和0.96%。R/S analysis method is used to analyze gas well production,and two different expressions of historical data time series and stationary time series are given.Then the output of gas well is predicted by using BP network method.In order to verify the effectiveness of this method,BP network is compared with BP neural network and Bayesian algorithm.The results show that the prediction result obtained by Bayes algorithm is more accurate and realistic than the other two methods.For the prediction results of the three methods,the average error between the prediction results and the actual value using the Bayesian algorithm is 0.67%and 0.33%respectively,while the average error between the prediction results and the actual value using the traditional BP neural network is 1.71%and 0.96%respectively.

关 键 词：R/S分析 BP网络方法 气井产量 贝叶斯算法 

分 类 号：TE19[石油与天然气工程—油气勘探]