Research On Real-time Detecting Method For Pipeline Weak Leakages Based On Chaos Theory And Generalized Fuzzy Hyperbolic Model

With the fast development of national economy and establishment of the national pipeline network,the safety of pipelines' running can never be too important.For timely detection of oil pipeline can effectively reduce the incidence rate of oil spill and reduce the destruction.In all of the fluid pipeline leak detection methods,the negative pressure wave(NPW)detection is simple to realize and its cost is low,so it has been widely used in real pipeline detection.But the traditional negative pressure wave method is poor in detecting weak leakage.And it can be strongly influenced by the condition adjustment of the pipeline system.To solve these problems,this paper creatively puts forward a new method of on-line detection of pipeline weak leakage using chaos theory and generalized fuzzy hyperbolic model(GFHM).This thesis firstly discusses the feasibility of using chaos theory to deal with the pipe pressure issues.Relevant principles of the nonlinear dynamics were used to study the actual pipeline measurement data.The pipeline pressure time series' stationarity and nonlinearity were both proved.Then the pressure time series' chaotic characteristics were calculated such as correlation dimension and Lyapunov exponents.Finally we got the conclusion that pipeline pressure time series are chaotic time series,which was the theoretical foundation of the follow-up chaotic analysis for pressure time series.This study introduces the generalized fuzzy hyperbolic model(GFHM),and set up the pressure time series prediction model with GFHM based on the time series prediction principle.Then the BP neural network method is adopted to identify the model,and real measurement data simulation is used to measure the model's approximation ability,it's proved that the method is accurate and effective.A new kind of pipeline on-line detection method is proposed by applying the forecasting model to the actual pipeline detecting process.To solve the problems of weak leakages' accurate identifying,this thesis builds a new generalized fuzzy hyperbolic model to analyze the pipeline operation condition by using the chaotic dynamic characteristics of pipe pressure.This new method is named as the pipeline pressure's chaotic analysis method based on GFHM.And this model achieves the purpose of correctly identifying negative pressure wave signal generated by the weak leakage of pipeline.Then pipeline condition adjustment information is used to train a generalized fuzzy hyperbolic classification model offline.The negative pressure wave signals caused by normal condition adjustment are classified,which will effectively remove condition adjustment's influence on the leak detection and reduce the false alarm.Combining the above analysis and location technique of negative pressure wave method,a new pipeline on-line detection method for weak leakage is established.With comparison of the simulation results of this method and the previous detection method based on pressure time series prediction,the detection method based on chaotic analysis of weak leakage is better in both accuracy and the false alarm rate.After a simulation from the mass of real data,the results prove that the proposed pipeline on-line detection method for weak leakage can effectively improve the accuracy and sensitivity of the fluid pipeline leakage detection.