基于神经网络的油管传输射孔（TCP）信号分析
发布时间：2018-09-19 06:33
【摘要】：目前在油田测井作业中,工人采用在井口听爆破声音和用手触摸油管壁感觉振动判断射孔弹是否起爆,通过将射孔器提出地面数弹孔计算射孔数,这种方法是十分落后的。经常会发生因作业油井过深导致振动不明显,人员在井口感觉不到振动,从而无法判断射孔弹是否起爆。如果对起爆信号判断失误,就会给接下来的作业带来很大影响,甚至造成人员伤亡。所以需要一种有效的方法,对射孔弹起爆信号进行分析,判断射孔弹是否起爆并计算有多少弹起爆。通过对大量油管传输射孔信号作傅里叶分析,确定射孔信号的有效频率范围,根据这些参数设计巴特沃斯带通滤波器滤除有效频率之外的噪声部分。小波变换具有良好的时频局部化特性,利用小波变换对有效信号进行降噪处理,去除夹杂在有效信号中的高频噪声,重构出射孔信号,此时的信号就十分接近原始信号。同时利用信号经过小波变换后的分量确定射孔弹的起爆开始和结束点,利用这段时间差计算射孔数,结合每次作业中设定的射孔总数可以计算射孔率。通过不断调整BP网络参数得到神经网络模型,此模型经过大量样本数据的训练就具备识别射孔信号的能力,经过实验证明,此网络可以识别一级起爆的油管传输射孔信号。本文将数字滤波器、小波变换和神经网络方法结合,提出基于神经网络的油管传输射孔信号分析,经实验证明此方法有效解决了油田测井作业中对射孔弹起爆判断问题和射孔弹数的计算问题,同时可以利用BP神经网络识别油管传输射孔信号,基于神经网络的油管传输射孔信号分析方法为油田工作人员判断油井射孔质量优劣提供了较为准确的依据。
[Abstract]:At present, in oil field logging, workers use the sound of blasting at the well head and the sensory vibration of touching the oil pipe wall to judge whether the perforator is initiating or not. This method is very backward by putting forward the surface number of perforators to calculate the number of perforations. It often happens that the vibration is not obvious because the working well is too deep, and the personnel can not feel the vibration at the well head, so it is impossible to judge whether the perforating projectile is detonated or not. If the initial signal is misjudged, it will have a great impact on the next operation, and even cause casualties. Therefore, an effective method is needed to analyze the ejection signal of perforation, to judge whether the projectile is primed and to calculate the number of ejection. The effective frequency range of perforation signal is determined by Fourier analysis of a large number of perforation signals transmitted by tubing. According to these parameters, Butterworth bandpass filter is designed to filter the noise part beyond the effective frequency. Wavelet transform has good time-frequency localization property. Wavelet transform is used to reduce the noise of the effective signal, remove the high frequency noise in the effective signal, and reconstruct the perforated signal. The signal is very close to the original signal. At the same time, the signal components after wavelet transform are used to determine the initiation point and the end point of the projectile, the perforation number is calculated by using this time difference, and the perforation rate can be calculated by combining the total number of perforations set in each operation. The neural network model is obtained by constantly adjusting the parameters of BP network. The model is trained by a large number of sample data and has the ability to recognize perforation signal. The experiment shows that the neural network can recognize the perforation signal transmitted by the tubing that is primed by one stage. In this paper, the digital filter, wavelet transform and neural network are combined to analyze the perforation signal of tubing transmission based on neural network. The experiments show that this method can effectively solve the problem of judging the ejection of perforation and calculating the number of perforated projectiles in oil field logging operations. At the same time, the BP neural network can be used to identify the perforating signals transmitted by tubing. The neural network based perforation signal analysis method provides a more accurate basis for oil field workers to judge the perforation quality of oil wells.
【学位授予单位】：沈阳理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P631.81;TP183
本文编号：2249327
[Abstract]:At present, in oil field logging, workers use the sound of blasting at the well head and the sensory vibration of touching the oil pipe wall to judge whether the perforator is initiating or not. This method is very backward by putting forward the surface number of perforators to calculate the number of perforations. It often happens that the vibration is not obvious because the working well is too deep, and the personnel can not feel the vibration at the well head, so it is impossible to judge whether the perforating projectile is detonated or not. If the initial signal is misjudged, it will have a great impact on the next operation, and even cause casualties. Therefore, an effective method is needed to analyze the ejection signal of perforation, to judge whether the projectile is primed and to calculate the number of ejection. The effective frequency range of perforation signal is determined by Fourier analysis of a large number of perforation signals transmitted by tubing. According to these parameters, Butterworth bandpass filter is designed to filter the noise part beyond the effective frequency. Wavelet transform has good time-frequency localization property. Wavelet transform is used to reduce the noise of the effective signal, remove the high frequency noise in the effective signal, and reconstruct the perforated signal. The signal is very close to the original signal. At the same time, the signal components after wavelet transform are used to determine the initiation point and the end point of the projectile, the perforation number is calculated by using this time difference, and the perforation rate can be calculated by combining the total number of perforations set in each operation. The neural network model is obtained by constantly adjusting the parameters of BP network. The model is trained by a large number of sample data and has the ability to recognize perforation signal. The experiment shows that the neural network can recognize the perforation signal transmitted by the tubing that is primed by one stage. In this paper, the digital filter, wavelet transform and neural network are combined to analyze the perforation signal of tubing transmission based on neural network. The experiments show that this method can effectively solve the problem of judging the ejection of perforation and calculating the number of perforated projectiles in oil field logging operations. At the same time, the BP neural network can be used to identify the perforating signals transmitted by tubing. The neural network based perforation signal analysis method provides a more accurate basis for oil field workers to judge the perforation quality of oil wells.
【学位授予单位】：沈阳理工大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P631.81;TP183
【参考文献】
相关期刊论文 前3条
1 Saeed Chehreh Chelgani;Behzad Shahbazi;Bahram Rezai;;Estimation of froth flotation recovery and collision probability based on operational parameters using an artificial neural network[J];International Journal of Minerals Metallurgy and Materials;2010年05期
2 杨建广,吕绍林;地球物理信号处理技术的研究及进展[J];地球物理学进展;2002年01期
3 胡丽莹;肖蓬;;快速傅里叶变换在频谱分析中的应用[J];福建师范大学学报(自然科学版);2011年04期
，本文编号：2249327
本文链接：https://www.wllwen.com/kejilunwen/diqiudizhi/2249327.html
