长时井下压力监测数据流动过程识别方法研究

利用长时井下压力计监测井下生产在提高油藏和油井管理方面正发挥着越来越重要的作用。在长时井下压
力监测数据的解释过程中，准确地确定流动变化过程起始时间至关重要，但由于其庞大的数据量而使得手工划分和处
理这些数据不切实际。基于小波模极值理论和噪声鲁棒微分算法，利用模拟的井下压力数据对比研究了长时井下压
力监测数据中流动变化过程（突变点）的识别方法。结果表明：小波模极值方法的误识别率相对较高，对噪声比较敏
感，在采用该方法之前必须对数据进行降噪处理；而基于噪声鲁棒微分算法的二阶导数识别方法可以准确、有效地识
别出流动变化过程，并且对噪声具有稳健性，可以在不对信号进行降噪处理的情况下识别流动过程。研究结果为自动
处理长时井下监测数据提供了一种新的手段。

Study of Identification Method of Transient Flow from Permanent Downhole Pressure Data

The real-time monitoring technology of the downhole conditions with permanent downhole pressure gauge is playing
an important role in improving reservoir and well management. During the interpretation of permanent downhole pressure data，
it is vital to acquire a good well-test result that accurately identifies the beginning time of new transient flow. Due to the large
volume of the collected data by permanent downhole gauge，it is impractical to partition and process these data manually. Based
on wavelet transform module maximum theory and noise-robust differentiator，the identification methods of transient flow from
permanent downhole pressure data are investigated with the synthetic data. The results show that the wavelet transform module
maximum method may not identify some key transient flows，and it is sensitive to the noise. The data must be de-noised before
using this method. But with the noise-robust differentiator，the transient flow can be identified accurately and effectively using
its 2nd derivative. It is robust to noise and can be used to identify the transient flow without data de-noising. The study provides
a new automatic method to process the permanent downhole data.