基于支持向量回归机的油田生产预警模型研究

传统的油田开发动态生产预警采用独立性指标阈值判别方法,从而带来预警结果不准确、异常事件发生时报警而不是预警等问题。本课题提出一种油田生产预警模型,该方法将支持向量回归机(Support Vector Regression,SVR)用于油田生产预警中,通过分析历史生产动态数据,找到它们的变化规律,总结出生产异常警报形成模式,在油田异常事件的初期给出预警信号,提前分析处理潜在隐患,以便保证油田采收效率的稳定性。实验结果证明模型对于油田生产中发生的异常情况具有较高的预测准确性。

Research on Early Warning Model of Oilfield Production Based on Support Vector Regression

The discrimination method of independent index threshold is used in The traditional Oilfield development dynamic production warning,so it brings inaccurate warning results or it will alarm instead of early warning when an abnormal event occurs.This paper puts forward a early warning model of oilfield production,the method uses support vector Regression （SVR） for early warning of oilfield production.By analyzing dynamic data of historical production,finding their variation,summing up abnormal alarm formation mode,giving signal in the early warning,early analyzing and processing potential risks in order to ensure the stability of oil recovery efficiency is necessary.The experimental results show that the model for oilfield production abnormal condition has a high prediction accuracy.