Leak Detection For Gas Pipeline Based On Acoustic Data-driven

With the development of national economy and the adjustment of the energy structure strategy,the proportion of natural gas is increasing for the advantages of efficiency and cleanness.Pipeline is the main way to transport natural gas,and China is building large number of pipelines for natural gas transportation.Because of the defects of pipeline,natural environment and artificial destruction,leak occurs occasionally,which has destroyed the surrounding environment and even caused economic losses and personal death of people.Thus,the technology on the leak detection and location of pipelines is an important means to ensure the safe operation of the pipeline.Acoustic method has the potential to be a good technology for its high sensitivity and accurate location.In this paper,by using the acoustic leak detection system in the lab,signals are collected at the cases of different leakage sizes and stable case.Firstly,features are extracted from the different acoustic signals,and then their effects on the classification are analyzed qualitatively.Secondly,the high dimensional features are reduced by using principal components analysis(PCA)method.Based on that,the acoustic signals are classified.Then,linear discriminant analysis(LDA)is used to reduce the dimension of extracted features and analyzed its dimension reduction result.At last,based on the low dimensional features,K-means clustering method is adopted for the discriminant analysis and classification processing.By studying the acoustic data,some conclusions can be drawn as follows:(1)The stable acoustic data can be distinguished from the leak acoustic data,the different leakage data under different aperture sizes can be distinguished.(2)The principal component analysis method can be used to reduce the dimension of the multidimensional features.However,the classification result is not obvious by using the low dimensional features.(3)The linear discriminant analysis method can be used to reduce the dimension of the multidimensional features.In the low dimensional space,the acoustic signal features within the same class can be gathered closely,while those within the different class have the large spatial difference relatively.