Research Of The Pipeline Leakage Features Diagnosis Method

Pipe is used in electric power and petrochemical widely. In the course of pipeline,it may leak due to corrosion, welding quality, vandalism, etc. Pipeline leak will causemajor economic losses and bad social effects. When the fluid of pipeline contains anumber of toxic and hazardous substances, it will cause serious environmentalpollution. Over the years, domestic and foreign scholars have been studying thedetection methods of pipeline leak. However, the research of pipeline leakage faultdiagnosis is not a lot. This paper attempts to find pipeline leakage characteristicsdiagnostic solution.In the process of pipeline leaking, the fluid erupted outward through leak holesand formed sound wave, and this process can be seen as the phenomenon of acousticemission. Acoustic emission sensors can be used to pick up the acoustic emissionsignals at this moment to detect and diagnose the process of pipeline leaking. Thispaper uses wavelet packet method to decompose acoustic emission signals caused byleakage. Then, take the percentages of each node energy and total energy aseigenvectors of this signal. Apply neural network technology, taking the eigenvectorsof acoustic emission signals as input and taking the type of pipeline leakage as output,to achieve the diagnosis process. This paper obtained the regular pattern of acousticemission signals variation with the parameters through experiments, and apply thismethod to distinguish non-leakage, wear leakage and welding quality leakssuccessfully. This method has certain feasibility in the aspect of pipeline leakdetection and diagnosis.Through theoretical study and experimental validation, pipeline leak diagnosismethod proposed in the paper is feasible and reliable and can be applied to onlinemonitoring of all kinds of pipes. It provides a new method for pipeline leakagedetection.