Signal Processing Researches For Oil Well Dynamic Fluid Level Detection Based On The Column Sound Field Model

The accurate detection of the oil well's depth is of great significance to improve the oil recovery efficiency.At present,due to the sound device's insecurity and difficult identification of liquid surface reflection wave,the echo method widely used in oil field to detect dynamic fluid level depth for oil well can't better meet the oil field mining requirements.Jia et al.proposed a new method for detecting the liquid level of the oil well based on the sound field model of the column,which provides a new way for the complex oil wells.In this paper,a further study based on this new method has been done to improve the detection accuracy.Firstly,the model is improved by establishing the relationship between the two adjacent resonant frequencies' interval(35)f and the depth dynamic liquid,which effectively avoid the identification of order n of thenf.Secondly,although the instantaneous pulsed sound waves in the existing oilfield production can excite the air column resonance in the well casing,the resonance sound waves is very unfavorable to the detection due to its energy decaying seriously.In this paper,the continuous white noise is used to generate the resonance of the multi-stage frequency of the underground air column,and the resonant sound waves can be supplemented by continuous sound in the wellhead.Thirdly,for the random white noise,the noise generated by the various obstacles and the mechanical noise generated by the downhole motor operation,a spectral FFT method based on Welch power spectrum estimation is proposed,which can greatly eliminate the interference of noise and extract the frequency difference of resonant harmonics accurately and conveniently.In addition,for the problem of inherent error of FFT due to spectrum leakage and fence effects,this paper have studied the classic interpolation FFT algorithm.It is found that the accuracy of the existing algorithms are not stable and have a great relationship with the frequency deviation through the simulation analysis.In this paper,an improved iterative interpolation algorithm is proposed,and the correction accuracy is not changed with the frequency deviation.By using this method,the FFT results of Welch power spectrum estimation can be corrected to further improve the detection accuracy of moving surface.Finally,experiment results have verified the feasibility of the signal processing method above,and the absolute error in pipeline experiment are less than 0.5 meters for the length within 150 meters.