Identification of Pipeline Leak Sizes Based on Chaos-Gray Wolf-Support Vector Machine

Accurately identifying pipeline leak size is of great significance for hazard assessment and timely rescue. This article proposes an identification method of nonmetallic pipeline leak size based on chaos-gray wolf-support vector machine (C-G-SVM). The acoustic signal features of different leak sizes are extracted from the perspectives of time domain, frequency domain, and shape. By using the gray relational analysis (GRA) method, the dimensionality of the above features is further reduced. Then, a nonmetallic pipeline leak size identification model based on C-G-SVM is established. The parameters of the SVM model are optimized by combining chaotic local search with gray wolf optimization (GWO) algorithm to improve the identification accuracy of pipeline leak sizes. Finally, the influences of different features, identification methods, and sampling duration on the identification accuracy of pipeline leak sizes are compared and analyzed. The analysis of nonmetallic pipeline leak test data based on acoustic methods verifies the effectiveness of this method. When the sampling duration is 20 s, the average identification accuracy reaches over 90%. The results show that this method can accurately identify the leak size of nonmetallic pipelines, providing a theoretical basis for engineering applications.