油田开发指标系统的概率模拟预测

Probability Simulation Prediction on the System of Oilfield Development Indices

基于油田开发指标系统的复杂性和其影响因素带有明显的随机特征,从理论和实用上研究了油田开发指标系统的不确定性预测方法.方法提出采用神经网络建立开发指标和影响因素的关联关系;利用微分模拟研究开发指标的影响因素的概率分布;使用蒙特卡罗方法对影响因素进行随机抽样;依据得到的关联关系和影响因素的随机样本建立油田开发指标的概率预测模型.此外,设计了概率模拟器.实例研究表明开发指标的预测值以某概率落在某个区间内,这可为决策者做出符合实际的决策和准备应急决策方案提供依据.

The uncertainty prediction method of oilfield development index system was studied theoretically and practically in this paper based on its complexity and obviously random characteristics of its influencing factors.This method proposes that using neural network to build the incidence relation between a development index and its influencing factors, applying differential simulation to study probability distribution of influencing factors of the development index,utilizing Monte Carlo method to sample randomly for influencing factors, and according to obtained incidence relation and random sample of influencing factors to construct the model of probability prediction.Besides,a probability simulator was designed. The case study shows that the predicted value of development index falls in some interval in some probability,and which may provide the reference for decision makers to make decisionmaking to match the real world and to prepare contingency plan of decision-making.