MFCC-LSTM framework for leak detection and leak size identification in gas-liquid two-phase flow pipelines based on acoustic emission

Two-phase gas-liquid flows are crucial to the pipeline system. Due to their complicated flow state, existing leak detection techniques are unsuitable for two-phase flow pipelines. To avoid accidents caused by a leak of pipelines, we present a framework for combining the Mel-frequency cepstral coefficient and long short-term memory (MFCC-LSTM) based on acoustic emission (AE). A series of experiments are performed considering 1152 operating conditions, including flow pattern, leak size, direction, and location. In addition, the detection capability of the different features of AE signals combined with LSTM is discussed. The results show that the recognition accuracy of the MFCC-LSTM framework reaches 98.4%. Then, we further perform leak size identification under different flow patterns and found that the MFCC-LSTM framework still exhibits excellent performance. The proposed MFCC-LSTM framework provides a promising solution to identify the leak state and size in two-phase flow pipelines based on the AE technique.