Pipeline Leakage Detection Based On Ultrasound Wave Velocity And Support Vector Machine

Pipeline transportation is widely used in petrochemical industry,but various unstable human and natural factors often cause pipeline leakage.Because petrochemical products are usually flammable,explosive and highly polluting,it is inevitable to study pipeline leakage detection and location in the study of pipeline transportation.Because long-distance oil pipeline has the characteristics of diversity of regional environment,long distance transportation and large transportation volume,various uncertainties along long-distance oil pipeline bring great challenges to effective leakage detection and location of long-distance oil pipeline.When the pipeline leaks suddenly,if we can quickly find the leak and locate the location of the leak point accurately,we can not only reduce the economic losses of the pipeline operators,but also reduce the pollution to the environment and the threat to the safety of human life and property to a certain extent.Therefore,in order to make the pipeline run safely and efficiently,it is of great practical significance to study the pipeline leakage detection and location.The main research work of this paper is as follows:According to the problems of inaccurate negative pressure wave when the location of the pipeline leak location,Firstly,wavelet transform,variational mode decomposition(VMD)and empirical mode decomposition(EMD)are used to denoise the pressure signal,and their processing effects are compared.A signal processing method suitable for pipeline leakage detection and location is selected.Secondly,the pressure signal processed by this method is used to extract the pressure inflection point under the leakage condition.Finally,the pipeline leakage location is carried out by the time difference of the pressure inflection point extracted from the pressure signal.On the basis of the original method of locating the pressure transmitters at the first and last stations,a new method of locating the leakage of long-distance oil pipelines is proposed,which relies on the pressure transmitters installed at the first,middle and last stations of pipelines.Through the analysis of experimental results,EMD can be effectively used to eliminate the noise of pressure signal,and the location accuracy of the new leak location method is obviously better than that of the original location method.To solve the problem of small leakage detection in long-distance oil pipelines,a method of small leakage detection based on particle swarm optimization support vector machine(PSO-SVM)is proposed.This method utilizes the different characteristics of the ultrasonic wave velocity signal under normal and leakage conditions,extracts the time domain feature and waveform feature of the ultrasonic wave velocity signal,and takes the extracted eigenvalue as the input of support vector machine(SVM).Because different types of SVM and different types of kernel function,normalized interval,penalty factor c and nuclear parameter g will affect the detection and identification of pipeline leakage,particle swarm optimization algorithm is used to find the optimal penalty factor c and nuclear parameter g after obtaining the optimal SVM type,kernel function type and normalized interval.The field experiment results show that the proposed method can effectively improve the recognition accuracy of small leakage detection in oil pipeline compared with the classical SVM.