Pipeline Leak Detection Based On Information Entropy

Pipeline transportation is one of the most common transportation methods and plays an important role in oil and gas transportation.With the increasing construction of pipelines,leakage occasionally occurs due to natural or human factors,which brings huge economic losses and environmental pollution.Additionally,it cause serious production accidents and even endangers people's lives.Therefore,how to establish an efficient pipeline leak detection system and detect pipeline leakage timely is one of the hot topics today.The main research work of this thesis is as follows:A complete pipeline leak detection system is designed.It is consist of hardware and software.Collect pressure signal of four conditions(normal,leak start,continuous leakage,end of leak)of pipeline through experiment,as the analysis of algorithm.The characteristics of the time series are obtained by using permutation entropy method for pipeline pressure.Through observing the abrupt change of these characteristics,pipeline leak can be detected.Considering the drawback of single-scale permutation entropy in reflecting the complexity of pressure time series,a novel pipeline leak detection method based on multi-scale weighted permutation entropy is proposed,the dynamic characteristic of the pressure time series on different scales is obtained.The results show that multi-scale weighted permutation entropy has obvious superiority.A novel pipeline leak detection method based on Hellinger distance is proposed in this paper for pipeline pressure time sequences.Firstly,the normal dynamic pressure wave signal is collected and the time sequences are selected by adopting maximum overlapping moving window method.Then the Hellinger distance statistic is constructed to monitor the state of the pipeline.In order to monitor the pipeline leak accurately,the control limit is established through the Shewhart control chart.The results show that the method is simple and has a good real-time.