Pipeline small leak detection based on virtual sample generation and unified feature extraction. (November 2021)

• Record Type: Journal Article
• Authors: Zang, Dong; Liu, Jinhai; Qu, Fuming
• Is Part Of: Measurement. Volume 184(2021)
• Publication Date: 2021-11
• Subjects: Oil pipeline -- Small leak detection -- Virtual sample generation (VSG) -- Unified feature extraction (UFE); Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2021.109960
• Languages: English
• ISSNs: 0263-2241

Abstract

Abstract: Due to the lack of samples and concealed features, petroleum pipeline small leak detection is still a great challenge. In this paper, a method based on virtual sample generation (VSG) and unified feature extraction (UFE) techniques is proposed to detect small leak. First, an effective sample generation algorithm based on limited raw sample and prior knowledge is designed. We verify the effectiveness of the generated sample from sample diversity and statistical similarity. Then, seven statistical features and a group of symbol transformation features are extracted to deeply mine sample information. And then, the extracted features are combined to unified features (UFs). Finally, small leak detection models are obtained by training UFs using four machine learning methods. In the experimental process, the proposed method is compared with other state-of-the-art small leak detection methods. Experimental results show that the proposed method has a better ability in pipeline small leak detection. Graphical abstract: Highlights: An effective data generation method is designed to enrich limited pipeline data. A UFE method is designed to mine the local and global features for pipeline data. The proposed method performs well in detecting the oil pipeline small leak.