关键词: 抽油井, 动液面, 高斯过程回归, 预测, 石油, 动态建模

Abstract:

In practice, dynamic fluid level is traditionally measured onsite by using the acoustic method. This method, however, has its limitation in determining real-time dynamic liquid level. Determining real-time dynamic liquid level by analyzing the measured dynamometer card has poor precision. Model aging happens as time goes by with the data driven soft sensing modeling method. An incremental dynamic Gaussian process regression (IDGPR) was presented for the soft sensing modeling in order to realize real-time determination of dynamic liquid level. At the beginning a basic soft sensing model based on dynamic Gaussian process regression was established. After the model was put into application, it could be updated on-line through an incremental learning method. The model could be constantly adaptable to the change of operating condition and precisely predict dynamic liquid level. The application result in the oil field showed that the proposed soft sensing model achieved high prediction precision and good generalization ability, meeting engineering requirement.

Key words: pumping well, dynamic liquid level, Gaussian process regression, prediction, petroleum, dynamic modeling