油井增产措施效果预测方法研究

通过分析不同影响因素与措施效果的内在联系,采用经验类比法与BP神经网络法建立了措施效果的预测模型。模型克服了解析法、数值模拟法、统计回归法进行措施效果预测的固有缺陷,综合考虑了影响措施效果的油藏地质、开发工艺等因素,可以方便地预测措施后的产量递减规律。实例应用表明,两种预测方法单井次最大相对误差25%,平均相对误差15%,取均值后单井次最大相对误差18%,平均相对误差8%,基本能够满足矿场进行措施效果预测的工程需要。

Research on prediction method of stimulation effect

Analyzing the internal links between different influence factors and stimulation effect, this paper establishes the prediction model of stimulation effect with empirical analogy method and BP neural network method. The model overcomes the inherent defect of analytic method, numerical simulation method, statistical regression method, considers the impact factors of reservoir geology, development process and so on, and can conveniently predict the production decline law after measures.The instance application shows that single well maximum relative error is 25% while the average relative error is 15%. After taking the average of the two methods, single well maximum relative error is 18% and the average relative error is 8%.The method can basically meet engineering needs of predicting the measures effect.