基于气泡声学的水下气体泄漏检测方法研究

发布时间：2018-03-05 13:23

本文选题：气泡声学　切入点：水下气体泄漏　出处：《天津大学》2016年博士论文　论文类型：学位论文

【摘要】：水下气体泄漏广泛存在于自然界以及工业环境中,尤以海洋环境下最为常见。海洋开发过程中,由于油气管道破损、海底冷泉、可燃冰开采和碳封存失效所导致的气体泄漏不仅会导致直接经济损失,还将引发安全和环境问题,产生恶劣影响。水下气体泄漏检测技术的应用能够有效发现泄漏的存在,在泄漏早期及时采取修复和补救措施,避免造成更加严重的后果。因此,研究水下气体泄漏检测技术具有重要意义。基于水下气体泄漏时将形成气泡这一特有现象,本文提出了以气体泄漏时所产生的气泡声学信号作为识别目标的泄漏检测方法,具有适用性强、灵敏度高等优点,有望成为现有技术的重要补充。由于气体的可压缩性,气泡将具有一定的弹性,因此在受到激励时将发生脉动并在液体中产生声学信号。本文分析了气泡脉动过程中的动力学模型,研究了气泡声学信号的传播和衰减特性,并在实验室环境下搭建了模拟水下气体泄漏平台,基于高速成像技术研究了不同流量下气泡的形成过程及声学信号特点。同时,引入了时频分析方法对声学信号的产生原因进行研究,证实气泡生成时的“夹断”现象以及气泡的融合和分裂均将提供激发气泡产生声学信号的初始能量。微量泄漏条件下,气泡的生成具有周期性和重复性。针对这一特点,本文提出了一种基于时频能量分析的微量泄漏信号检测方法以及相应的自适应谐波检测算法,可识别出噪声环境下的微量泄漏信号和气泡产生的频率,并基于置信度分析研究了窗函数长度、噪声强度和信号衰减系数对算法性能的影响。与传统频域分析方法相比,该方法具有更好的抗噪声干扰能力。针对含噪信号中气泡声学信号的位置识别问题,本文提出了基于双稳系统及量子粒子群寻优随机共振的气泡信号位置识别方法。本文选取脉冲形式的输出信号作为衰减振荡信号的最优随机共振结果,基于双稳系统模型,利用量子粒子群寻优实现了噪声环境下的气泡信号位置识别方法,进一步提出了反向随机共振方法,可有效改善气泡信号的位置识别精度,有助于泄漏源的定位。随着泄漏流量的增加,泄漏所产生的气泡声学信号将具有持续性特征,可采取基于阵列的信源数估计算法对其进行识别。本文提出了一种分析基于最小描述长度准则的信源数估计算法性能的新方法,相比现有分析方法在小快拍数下具有更好的预测精度,多泄漏源条件下可对不同信源数估计结果的概率进行计算,解决了算法欠估计程度的分析问题,对水下泄漏检测阵列的参数选择和设计具有指导意义。
[Abstract]:Underwater gas leakage exists widely in the natural and industrial environments, especially in the marine environment. The gas leakage caused by combustible ice mining and carbon sequestration failure will not only lead to direct economic loss, but also lead to safety and environmental problems, which will have adverse effects. The application of underwater gas leakage detection technology can effectively detect the existence of leakage. Timely repair and remedial measures should be taken at the early stage of the leakage to avoid more serious consequences. Therefore, it is of great significance to study the underwater gas leakage detection technology. In this paper, a leak detection method based on bubble acoustic signals produced during gas leakage is proposed, which has the advantages of strong applicability and high sensitivity, and is expected to be an important supplement to the existing technology. Bubbles will be elastic, so they will pulsate and produce acoustic signals in liquid when excited. In this paper, the dynamic model of bubble pulsation is analyzed, and the propagation and attenuation characteristics of bubble acoustic signals are studied. A simulated underwater gas leakage platform is built in laboratory environment. The bubble formation process and acoustic signal characteristics under different flow rates are studied based on high speed imaging technology. In this paper, time-frequency analysis is introduced to study the causes of acoustic signals. It is proved that the phenomenon of "clamping" during bubble formation and the fusion and splitting of bubbles will provide the initial energy to stimulate bubbles to produce acoustic signals. The generation of bubbles has periodicity and repeatability. In this paper, a micro-leakage signal detection method based on time-frequency energy analysis and the corresponding adaptive harmonic detection algorithm are proposed. The microleakage signal and the frequency produced by bubbles in noisy environment can be identified. Based on confidence analysis, the effects of window function length, noise intensity and signal attenuation coefficient on the performance of the algorithm are studied. This method has better ability to resist noise interference, aiming at the position recognition of bubble acoustic signal in noisy signal. In this paper, a method of bubble signal location identification based on bistable system and quantum particle swarm optimization stochastic resonance is proposed. The output signal in pulse form is selected as the optimal stochastic resonance result of the attenuation oscillation signal, and based on the bistable system model. The position recognition method of bubble signal in noisy environment is realized by quantum particle swarm optimization, and the inverse stochastic resonance method is proposed, which can effectively improve the position recognition accuracy of bubble signal. Contribute to the location of the leak source. As the leakage flow increases, the bubble acoustic signals generated by the leakage will have persistent characteristics, This paper presents a new method to analyze the performance of the source number estimation algorithm based on the minimum description length criterion. Compared with the existing analysis methods, the prediction accuracy is better under the condition of small beat number. Under the condition of multiple leakage sources, the probability of different source number estimation results can be calculated, and the analysis problem of the degree of underestimation of the algorithm can be solved. It is of guiding significance for parameter selection and design of underwater leak detection array.
【学位授予单位】：天津大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：P754

【相似文献】