Research Of Acoustic Early Warning Detection Technology For The Pipeline Leak Based On Blind Source Separation

Early warning detection technology for the pipeline leak is a kind of very important technology in closely related to the national economy.Nowadays the main reason for most accidents of the pipeline leak is that the lawbreakers drill to steal petroleum, and also some are caused by the mistakes of machine operation. So early warning detection technology to protect the pipeline from destroy caused by drilling is the key technology that need to be done in pipeline transport in our country. The research of early warning technology for the pipeline leak is very important for the safety of pipeline.In this paper, blind source separation algorithm is used into the research of early warning technology for the pipeline leak. The damage signals are restored truly by separation of pipeline mixed acoustic signals. Early warning detection of pipeline can be realized by detecting characteristics of damage signal. Two input-two output timely blind source separation system is designed with infomax algorithm which based on natural gradient method. Two-route linearity instantaneous mix acoustic wave signals in the pipeline model are separated successfully. The researching scheme of a new kind of early warning technology for the pipeline leak is given and demonstrated.This paper mainly do the work following:(1) The early warning technology for the Pipeline Leak is researched based on blind source separation. The research focus on the problem that how to detach the two-route linearity instantaneous mix acoustic wave signals. Blind source separation algorithm is given because of infomax algorithm which is based on natural gradient method. The effectiveness of this method is proved in the simulated experiment of mixing-separation of square-wave and triangle wave. This algorithm has robustness under low-noise.(2) Two input-two output timely mix acoustic wave blind source separation system is designed and created. The blind source separation system has high sampling frequency and simple algorithm. It has the function of collection, separation, signal analysis and storage of two-route linearity instantaneous mix acoustic wave signals. The system real-time signal acquisition and the ability of analysis of data-processing is increased by optimization of learning rate. It provides practice test and analysis platform for application of theory.(3) The Semi-physical simulation for separated signals are realized resulting in the pipeline model and the linear transient vibration noise mixed blind source. The separation of destruction and vibration instantaneous linear mixed acoustic signals are realized successfully by experiment. Time-frequency domain analysis shows that separated damage signals have a clear identifiable characteristics and high value in early warning detection technology for the pipeline leak. The feasibility and effectiveness of early warning detection technology for the pipeline leak based on blind source separation are proved.Based on all of above, a new way of theory in the research of early warning technology for the pipeline leak is discussed in this paper, and its correctness and feasibility is proved by experiment and data analysis. It provides a solid theoretical foundation and practical experience for further research for early warning technology for the pipeline leak based on blind source separation.