Research On The Oil Pipe Leakage Detection And Positioning Technology Based On Intelligent Information Processing Theory

Pipeline transportation is an important way of oil transportation.It has great significance to the national economic construction.However,factors like the corrosion and aging and punching of stealing oil cause pipeline leakage accident frequently,which not only brings economic loss to enterprises,but also contaminates the environment and threatens people's life property safety.Therefore,timely detection and precise positioning of pipeline leakage are of great significance for reducing the loss and pipeline safety operation management.Owing to the complicated working environment and various modes and causes of leakage,there is no generality of oil pipeline leak detection and location system,and engineering testing site often requires several methods fusion and replenishment mutually.Applying new theory and method of signal processing and control engineering to solve the problem of oil pipeline leak detection and location is not only to supplement and verify the existing leakage detection and location methods,but also to improve the accuracy,applicability and timeliness of the existing methods.Based on the analysis of performance indexes of common pipeline leakage detection method,this paper explains the theoretical basis for the selection of infrasonic signal as the researching object,systematically studies the intelligent information processing methods and conducts leakage detection and positioning research on the acquisition signal from oil fields.The main research work is as follows:1.Analyzes and validates that the bilateral non-local mean filter is more stability and applicability for tube acoustic signal de-noising than those method such as mean filter,median filter,Gaussian filter,no-local mean filter and bilateral filter.Aiding at complicated pipeline leak monitoring signal with chaotic characteristics,a single leak signal de-noising method is difficult to effectively keep the local structure features of chaotic signal.Therefore,this paper firstly uses the the wavelet decomposition with one level and half hard-soft thresholding method to roughly remove the high frequency noise of pipeline monitoring signals,then the reconstructed signal is again decomposed into four levels by wave function of db9 and removed the middle-high frequency noise of the fourth wavelet decomposition signal.In the end,the reconstructed signal,that the fourth level wavelet decomposition is de-noised by half hard-soft thresholding method,is reduced the noise in spatial field by the proposed bilateral non-local mean filter.Experimental results show that the composite de-noising method of master-slave hierarchical structure can realize the effective noise reduction of pipeline leak monitoring signals.2.Puts forward the similarity measure method between pipeline monitoring signals by the information deviation being composed of information entropy and divergence of matrix singular value in the random uncertainty theory.Meanwhile,presents the series constructing method of EMD characteristic matrix,Hankel characteristic matrix,and Shannon information entropy of sampling signal based on the discrete probability distribution of characteristic matrix singular values.The concrete method of leakage locating is proposed by means of pattern matching,which firstly choose 50 sampling points signal without leakage,then calculate the information deviation between 50 points signal without leakage and actual sampling sub-sequence signal obtained by sliding window,finally draw the information deviation curve,and determine and locate the leakage by thresholding level method.3.Presents the leakage detection and position method based on fuzzy similarity and entropy in the Zadeh's classical fuzzy theory.Meanwhile,the intuitionistic fuzzy sets of sampling signals by combining power spectrum with Cauchy function are constructed,and pipeline leakage detection and location method utilizing intuitionistic fuzzy theory is given to enhance the universality of the fuzzy detection and location method.Studies the series of fuzzy c-means clustering algorithm including fuzzy clustering combined sample weighting with attribute weighting,adaptive weighting fuzzy clustering by information entropy regularization combined sample weighting and attribute weighting with class weighting,and adaptive three weighting combination fuzzy c-means clustering in Hilbert reproduction kernel space.In order to effectively clustering time sequence signals,the kernel space fuzzy c-means clustering algorithm with local information regularization constraint is proposed.A fuzzy leakage detection and location method based fuzzy partition entropy of the proposed adaptive fuzzy c-means clustering with local neighborhood information constraint is present.Experimental results indicates that the proposed fuzzy methods are suitable to detect and locate of oil pipeline leakage.4.Presents pipeline leak detection and location method based on data field theory.Meanwhile,the method of fast calculating potential value distribution of time sequence signals is put forward by combining truncated Gaussian window with data field potential function,and the leakage detection and location determines according to maximum non-zero potential value.Experimental results show that the proposed method based on data field theory can not only mine the singular point of time sequence signals,but also be applied to oil pipeline leakage signal detection and localization of oil pipeline transportation system.5.The acoustic wave and pressure wave signals collected in upstream and downstream sites of given petrochemical pipeline monitoring system are regarded as the researching objects.They are analyzed and processed by the proposed intelligence information method such as rand uncertainty entropy,fuzzy uncertainty entropy,and data field,and judged oil pipeline leakage to happen or not.If oil pipeline leakage happens,then the oil pipeline leakage positioning can determined by the above-mentioned location methods.Testing results show that the acoustic signal is more conducive to pipeline leak detection and accurate positioning than the negative pressure wave signal.