油井动液面深度测量中抗强随机噪声干扰的离散频谱校正算法研究

Accurate estimation of dynamic liquid level of oil well is of great significance to ensure the efficiency and reasonable production of oil field. Recently, a new method based on the column sound field model with white noise excitation was proposed and the problems in the traditional detection method, such as severe attenuation of weak echo signal and difficulty in identification of liquid surface wave can be avoided. However, further study reveals that high accuracy estimation is difficult to achieve by current spectrum correction algorithms because the received signal contains a lot of random noise for the adopted white noise excitation source. To overcome the defects, first, a new estimation algorithm based on time shifting was proposed. The constraint on time shifting in traditional phase difference method was removed and the problem of incorrect polarity estimation was avoided by importing multiscale analysis to enhance the robustness against random noise; Secondly, the influence of strong noise on correction algorithm was comprehensively analyzed, and also a noise performance evaluation system was established. At last, a spectrum parameter estimation model was proposed based on the characteristics of probability density of the proposed algorithm. The proposed spectrum correction method and parameters estimation model can provide a theory foundation as well as a new approach for accurately estimating resonance frequency in detection of oil well dynamic liquid level.

精确测量油井动液面深度对保障油田高效合理生产具有重要意义。基于白噪声激励下管柱声场模型的油井动液面深度测量方法可有效避免传统回波法中声波衰减严重、液面波难以辨识等问题，但进一步研究发现由于共振信号中包含大量的随机噪声，利用传统的频谱校正算法难以有效提高检测精度。因此，本项目拟首先提出一种基于长时移的比值校正算法，并通过引入多尺度的分析方法，克服传统频谱校正算法在强噪声下谱线定位错误的问题。其次，深入研究强噪声对离散频谱校正算法的影响，构建离散频谱校正算法的噪声性能评价体系。最后，通过研究各种离散频谱校正算法概率密度特点，建立强噪声下的频谱参数估计模型。本课题所提出的离散频谱校正算法和参数估计模型有望为强噪声下油井动液面测量中共振频率的精确估计提供新的理论基础和解决方案。

精确测量油井动液面深度对保障油田高效合理生产具有重要意义。基于管柱声场模型的油井动液面深度测量方法可有效避免传统回波法中声波衰减严重、液面波难以辨识等问题，但进一步研究发现，由于共振信号中包含大量的随机噪声，利用传统的频谱校正算法难以有效提高检测精度。.为此，本项目基于三个方面内容开展了研究：（1）建立系统输出的声学信号数学模型，提出新的共振谐波信号提取算法；（2）提出一种基于长时移的比值校正算法，构建离散频谱校正算法的噪声性能评价体系；（3）进行基于管柱声场模型测量原理的油井动液面测量仪器的设计与研究。.通过以上三个方面的研究，使得基于管柱声场模型的测量方法在理论与工程应用层面均取得了重要的进展：（1）建立了管柱共振信号的数学模型，并在该模型的基础之上提出自适应滤波器，完成了声学信号降噪算法的开发，将最大测量范围从不足200m提高至2200m，极大地提高了测量范围，降低了测量时间；（2）提出一种新的离散频谱校正算法和参数估计模型，对2200m长的实验管道进行研究发现，经过校正后的计算长度不受FFT点数的影响，计算结果更加稳定可靠；（3）设计了一套基于管柱声场原理的油井动液面测量样机，能够用于油田现场的进行测试。.本项目取得的研究成果，不仅提高了基于管柱声场模型的动液面测量方法的测量范围和测量精度，还进一步降低了测量时间，提高了该方法在油田现场大规模应用的可能性，同时还具有重要的科学意义：.（1）本项目建立了空气柱共振信号的数学模型，研究了自适应滤波器，提出了多传感器信息融合处理的方法，为提取油套环空气柱共振信号提供了新的思路；.（2）本项目提出基于长时移的比值校正算法，极大地保证了共振谐波波动次数的精度，降低了基于该方法测量动液面时的误差，体现了离散频谱校正算法在工程应用中具有的重要价值。