A Pipeline Leakage Aperture Identification Method Based on CNN-SVM Considering Sample Granularity

The accurate identification of pipeline leakage apertures is crucial for safeguarding the environment and conserving resources. This article proposes a novel approach for identifying pipeline leakage apertures through the fusion of convolutional neural network and support vector machine (CNN-SVM). By analyzing the continuous change state of pipeline leakage signals, a sample granularity calibration model for leakage signals is developed to obtain the optimal sample granularity. Using the leakage signal calibrated with the optimal sample granularity as an input, a pipeline leakage aperture identification based on the CNN-SVM is established. The proposed method was tested by conducting a pipeline leakage experimental setup, and the test results indicate that the CNN-SVM-based pipeline leakage aperture identification has the best performance, with an accuracy of 98.2%, which is 4.39% higher than when using CNN alone. This approach provides a theoretical foundation for the precise identification of pipeline leakage apertures and holds promising prospects for engineering applications.