Abstract: In this paper, the oilfield injection-production system is regarded as a "black box" system. In view of systems engineering, the conception of an overall information predection model is presented. The POISSON CYCLE model and the LOGISTIC CYCLE model, preseented by professor Weng Wen Bo, are used to describe the process of the production and the water cut changes in the oilfield injection-production system, and are taken as an essential mathematic model. The parameters of models are given by using the generalized recurrence gradient algorithm for system identification used in the modern theory of control which was presented by Professor Han Zhigang, and then a complete mathematical model for predicting the oil production and water cut are obtained. As verified by results calculated using oil field data from the oil field both at home and abroad satisfactory results are obtained both in the data matching and preformance predition. The precision of the performance prediction by using dynamic parameters is even higher than that using the classical ststic parameters.

摘要： 本文提出了整体信息预测模型的概念,以Poisson旋回、Logistic旋回公式为模型基本结构,采用系统辨识推广的递推梯度算法确定模型中的参数,得到了油田产油量,综合含水率预测的数学模型。国内外油田实际资料计算结果表明,模型的拟合精度和预报精度都是令人满意的。