A CNN Model for Gas Pipeline Leakage Detection Based on MFCC Feature Extraction

The safe operation of gas pipelines in urban, industrial, agricultural, and other areas still faces risks and challenges. The pipeline leak detection method is aimed at improving the safety of gas pipeline operations in various fields and reducing losses caused by gas leaks. This method utilizes highly sensitive microphones to collect real-time sound signals in pipeline environments and uses the Mel-scale Frequency Cepstral Coefficients (MFCC) algorithm to extract features from the collected sound signals. Then, a convolutional neural network algorithm is used to identify whether there is a gas leak. This method can accurately identify pipeline gas leaks with up to 98.958% accuracy.