Study On Caliber Identification And Location Of Gas Pipeline Leakage Based On Improved LMD Method

Natural gas pipeline construction projects are developing rapidly at present,while leakage accidents caused by pipeline corrosion,aging,man-made damage and other factors are also frequent,which brings many adverse effects to human beings.How to deal with the problem of pipeline leakage and identify or locate the leakage of gas pipeline in time is of great significance for maintaining the safety of pipeline operation.In this paper,an improved local mean decomposition method is proposed and applied to the field of pipeline leakage detection.It is combined with SVM theory model,neural network model and cross-correlation location algorithm to realize pipeline leakage caliber recognition and location,and the effectiveness of the method is verified by experiments.Local Mean Decomposition(LMD)algorithm is a new adaptive signal processing method,which has good effect in dealing with non-stationary signals such as pipeline leakage acoustic signals.In this paper,the accuracy and effectiveness of the method in adaptive signal decomposition is verified by simulation signals.Meanwhile,aiming at the phenomenon of modal aliasing-the defect of the algorithm itself,this paper presents a method by adding Gauss white noise for decomposition several times,and finally the signal-to-noise ratio of the signal is improved,and an improved local mean decomposition(ELMD)algorithm is obtained.After the improvement of the algorithm,the collected pipeline leakage signals are decomposed by ELMD,and several PF components are obtained for subsequent caliber identification and leakage location research.In the stage of leak caliber recognition and classification,the envelope spectrum of each PF component is obtained by Hilbert transform,and the envelope spectrum entropy is used as the criterion for leak caliber learning and classification;Then the support vector machine(SVM)and neural network(NN)models are introduced to investigate the classification of leak caliber of pipelines by changing kernel function and adjusting kernel parameters to find the optimal solution of the model,and different models are compared to evaluate the classification effect of each model comprehensively.As for leak location,a kind of cross-correlation method is adopted in this paper,and three sensor architecture schemes are established.The situation of two-point synchronous leak are analyzed and the leak hole is located combining the time interval of two-point receiving signal with sound speed.On the basis of cross-correlation algorithm,this paper proposes a positioning method based on ELMD kurtosis value.By comparing the kurtosis of each PF component decomposed by ELMD,the PF components with high kurtosis values are taken as the main PF component,by which the signals are reconstructed.Experiments show that this method can effectively reduce the location error and obtain more accurate location information compared with the traditional cross-correlation algorithms.The studies indicate that the improved local mean decomposition algorithm has a good application effect in the field of pipeline leakage detection and location.By combining with other algorithms,it has certain advantages in the identification and location of leakage caliber,and can provide certain theoretical support for practical engineering application.