基于声学脉冲压缩的天然气管道安全监测关键技术研究

This project intends to use acoustic pulse compression technology in natural gas pipeline safety monitoring field, which will achieve an on-line and real-time monitoring of abnormal events in natural gas pipeline, such as hydrate congestion and pipeline leakage, so as to provide a guidance to workers to take actions in time to protect pipeline safety. (1) Focused on complex structures of acoustic sensor arrays and complex characteristics of incident signals, a series of research works will be taken, which include the emission and propagation characteristics of a wide-pulse-modulated signal in natural gas pipeline, the parameters optimization of acoustic sensor arrays/ incident signals, and the corresponding pulse compression and matched filtering technique, so as to solve the problems between spatial resolution and monitoring distance of the system and depress the noise. (2) A sound velocity-induced waveform drift compensation technology based on resampling algorithms will be proposed, as well as a quantitative analysis of hydrate/ leakage size, and a research on monitoring technology of the dynamic process of hydrate congestion, so as to locate and measure these abnormal events accurately. (3) Based on the characteristics of reflected signals, a target signal extraction and recovery technology research will be taken, and an event recognition and classification method based on deep learning algorithms will be proposed too, which both are focused on the classification and recognition between abnormal events and transient normal events. The research results of this project will produce significant economic and social benefits and escort China's energy security.

本项目拟开展基于声学脉冲压缩的天然气管道安全监测关键技术研究，旨在实现天然气管道内水合物凝聚和管道泄漏等异常事件的在线实时监测，从而指导工作人员及时采取有效措施，维护管道安全。（1）基于声传感器阵列的复杂结构及入射信号的复杂特征，开展宽脉冲调制信号在管道中的发射与传播特性研究、声传感器阵列和入射信号参数优化设计研究以及对应的脉冲压缩与匹配滤波技术研究，从而解决系统空间分辨率和监测距离的矛盾并抑制外界噪声的干扰。（2）开展基于重采样算法的声速诱导波形偏移补偿技术研究、水合物凝聚及泄漏尺寸定量分析研究、水合物凝聚过程动态监测技术研究，以实现异常事件精确定位与精准测量。（3）基于反射信号特征，开展目标信号分离、提取与恢复技术研究、基于深度学习的事件识别与分类方法研究，以实现不同类型异常事件以及瞬态正常事件的分类识别。该项目研究成果将产生显著的经济社会效益，为我国能源安全保驾护航。

天然气作为一种清洁、优质能源，对调整我国能源结构，推动低碳经济发展具有十分重要的作用。由于天然气本身的介质特殊性，管道承担了绝大部分的输送任务。然而，天然气在开采、加工和集输过程中，当温度和压力满足一定条件时易形成天然气水合物。一旦管道内发生水合物凝聚现象，轻则造成管道运输能力下降，干扰正常生产；重则可堵塞井筒、管道、阀门等设备造成停产，甚至可能导致管道局部压力增大引起输气管道爆裂，引发安全事故。.本项目开展了基于声学脉冲压缩的天然气管道安全监测关键技术研究，主要研究内容包括：（1）宽脉冲调制声信号在管道中的发射与传播特性研究；（2）天然气管道异常事件精确定位与精准测量技术研究；（3）天然气管道异常事件识别方法研究。结果表明，线性调频信号和相位编码信号经匹配滤波后，均可实现噪声抑制并实现微弱信号的检测。针对水合物凝聚和管道泄漏，定位精度可达0.001m，信噪比可达18.76dB。当原始反射信号淹没在冲激噪声与高斯白噪声中时，经匹配滤波信噪比提升至9.84dB和8.79dB。针对环境温度导致的声速变化对探测灵敏度和定位精度的影响，本项目采用重采样算法对声速变化导致的波形偏移进行补偿，将水合物凝聚和管道泄漏检测的信噪比分别提升至8.30dB和7.30dB，相对定位误差可达0.0303%和0.0830%。最后，本项目设计了长短时记忆（LSTM）网络和卷积神经网络（CNN）深度学习模型，可以正确判别水合物凝聚、管道泄漏和弯道、法兰等正常事件。其中，LSTM网络对于非混叠信号分类准确率可以达到100%，对于混叠信号分类准确率为82.78%；CNN网络对于混叠信号的分类准确率为100%。本项目在强背景噪声情况下实现了天然气管道内水合物凝聚及管道泄漏事件的动态监测及定位，并可正确分类管道事件，可及时指导运营方采取相应措施，有效保护管道安全，为我国能源安全保驾护航。