Time Delay Estimation In Distributed Optical Fiber Sensing And Positioning System

In recent years, with the development of "West-to-East Natural Gas Transmission Project", the oil and gas pipeline length is increasing continually. As a result, the oil pipeline safety issues are becoming increasingly prominent and oil stolen or leakage accidents continually occur. Therefore, studies on oil pipeline safety alarm and positioning system are gradually carried out.The research of oil pipeline safety alarm and positioning system belongs to traditional time delay estimation field, yet with some specialty. In the system, distributed fiber sensors collect vibration signals around the oil pipeline which are then sent to monitoring center. Then the monitoring center detects the signal and locates events by time delay estimation algorithm. Once there is an abnormal event, the system will alarm the pipeline inspection staff to check dangerous condition. The distributed sensing in the system will certainly cause the signal received is a mixture of all vibrations around the pipeline. Thus it is much more difficult to estimate the time delay for positioning the real abnormal event.Specific to that characteristic of the distributed sensing and positioning system, this paper provides a whole solution, including data collection, preprocessing and positioning. First, this paper analyzes distributed fiber system and data characteristics, constructs time delay system transmission model and design data detection scheme to separate abnormal event data with normal one. The data quality is also analyzed for assessing the time delay estimation results. Then, two algorithms are provided for data preprocessing (Principle Component Analysis (PCA) algorithm and cepstrum filtering algorithm). Then, it comes the time delay estimation process. This paper use two time delay estimation methods. A modified generalized correlation method is designed to eliminate correlated noises. The Modified Chirp Z-Transform(MCZT) method is designed for narrowband signal. Both of the methods achieve good results in practical tests. At last, in order to improve time delay estimation robustness, a data fusion algorithm based on probability density curve and Kalman filter is used.Finally, the solution designed in this paper can control the estimation error within 0.2 sampling interval (800 meters) in practical application with good and robust results.