动态预报方法在油田多峰产量预测中的应用

Application of dynamic forecasting method to oil field multi-peak production prediction

缪飞飞

MIAO Feifei

中海石油（中国）有限公司天津分公司，天津 300459

CNOOC China Limited, Tianjin Branch, Tianjin 300459, China

油田生产是一个随时间变化的动态系统,以往采用固定参数模型进行产量预测，忽略了模型参数的时变性，导致其预测误差随着预报期间（步长）的增大而增大。油田在开发生产过程中，由于增产措施及综合调整的实施，油田产量表现出多峰的运行状态。针对该情况，文中通过大量文献调研及研究，最终选择陈元千教授研究成果——多峰预测模型，并结合韩志刚教授提出的多层递阶预报方法，对油田动态系统多峰产量预测进行研究。实际油田应用结果表明，该方法预报精度高，研究成果实用、有效，对于预测油田产量有较强的推广应用价值。

Oilfield production is a dynamic system. The fixed parameters model of production forecast ignores the variability of system parameters, which leads to the prediction error getting bigger and bigger with the increasing prediction period. Because of the stimulation treatment and the oilfield comprehensive adjustment, the oil field production curve has multiple peaks. In this paper, through a large number of literature investigation and research, the multi-peak forecasting model by professor Chen Yuanqian and the multi-level recursive method by professor Han Zhigang were chosen to study the multi-peak forecasting. According to the practical application, the research result is practical and effective, which has a strong application value in oilfield multi-peak production prediction.

10.6056/dkyqt201803018