Identification method of long distance pipeline leakage based on auditory saliency map

Urban long distance pipe network plays a critical role in the urban lifeline. Most of the long distance pipelines are buried underground, so it is difficult to monitor, detect and locate the leakage. To solve this problem, this paper designs an identification method of pipeline leakage based on audible signal and the corresponding identification model through the introduction of the auditory saliency map into the pipeline leakage detection since transient vibration can cause the significant change of signal amplitude and frequency structure. This method identifies the leakage condition through the change of amplitude and frequency structure caused by the transient vibration of leakage sound. By the analysis of signal filter, feature extraction and the time domain sensing characteristic of the auditory neurons, and the integration across different scales, the obvious signal change of frequency domain and time domain can be obtained and used to identify the pipeline leakage. The results show that the minimum flow rate of pipeline leakage detected by this method is no more than 0.3 L/min. © 2020, Jilin University Press. All right reserved.