改进型灰色神经网络模型在油田产量中的应用  

Application of Improved Grey Neural Network Model to Oil Yields

作　　者：张方舟[1] 严胡勇[1] 杨立全[1] 邱露露[1] 张媛媛[1] 高晓松[1] 

机构地区：[1]东北石油大学计算机与信息技术学院,黑龙江大庆163318

出　　处：《计算机技术与发展》2013年第6期241-244,248,共5页Computer Technology and Development

基　　金：国家科技重大专项(2008zx05010-001)

摘　　要：石油产量的精确预测,是石油企业制定合理的生产计划、避免盲目投资、实现可持续开发的重要条件。论文基于传统的灰色预测模型,根据大庆油田1992～2011年产量数据,推出了两种改进预测模型,分别对灰色模型进行参数优化和初值修正,并采用神经网络确定了组合模型中各单项模型权重,建立了改进灰色-神经网络组合模型,对大庆油田产量进行预测。实际数据分析结果表明:灰色-神经网络组合模型不仅可以有效解决BP网络训练样本不足的问题,还能有效运用各单项模型信息,从而明显提高了精度。通过进一步的分析、对比及讨论,文章认为,灰色-神经网络预测模型运用于国内外石油产量预测,方法可操作性强,结论科学性显著。Accurate forecasts of oil production are the important conditions that help the company make a reasonable plan, avoiding blind investment and achieving sustainable development. In this study, based on the traditional grey prediction model, according to the 1992- 2011 annual output data of the Daqing Oilfield, put forward two improved prediction model, optimize the grey model parameters and correct the initial value respectively, using neural network to determine the combination model of each single model weight, establish improved gray-neural network combination model to predict the production of Daqing Oilfield. The results of analysis showed that：the grey -neural network combined model can not only effectively solve the problem of insufficient BP network training samples,but also can effectively use each single model information, thus obviously improving the precision of forecast. After a comparison with works of other researchers and a subsequent analysis, concluded that the new method of gray-neural network combination model, which can be applied to predict the oil production at home and abroad, was both feasible and scientific.

关 键 词：大庆油田 灰色神经网络 石油产量 预测 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]