US 12,467,820 B2
Methods and systems for leakage analysis of urban pipelines and storage media
Yongmei Hao, Changzhou (CN); Min Li, Changzhou (CN); Juncheng Jiang, Changzhou (CN); Zhixiang Xing, Changzhou (CN); Qiang Yao, Changzhou (CN); Lihua Wang, Changzhou (CN); Fan Wu, Changzhou (CN); and Zhengqi Wu, Changzhou (CN)
Assigned to CHANGZHOU UNIVERSITY, Changzhou (CN)
Filed by CHANGZHOU UNIVERSITY, Jiangsu (CN)
Filed on May 12, 2023, as Appl. No. 18/317,053.
Application 18/317,053 is a continuation in part of application No. PCT/CN2022/143171, filed on Dec. 29, 2022.
Claims priority of application No. 202210649566.1 (CN), filed on Jun. 10, 2022.
Prior Publication US 2023/0417622 A1, Dec. 28, 2023
Int. Cl. G01M 3/24 (2006.01); G06N 3/08 (2023.01); F17D 5/06 (2006.01); G01M 3/00 (2006.01); G06N 3/09 (2023.01)
CPC G01M 3/243 (2013.01) [G06N 3/08 (2013.01); F17D 5/06 (2013.01); G01M 3/00 (2013.01); G06N 3/09 (2023.01)] 11 Claims
OG exemplary drawing
 
1. A method for leakage analysis of an urban pipeline based on local characteristic scale decomposition (LCD), comprising:
obtaining sample data and a corresponding time-domain waveform diagram of an infrasonic original signal of the urban pipeline;
obtaining several intrinsic scale components (ISCs) and a residual component by adaptively decomposing the infrasonic original signal using the LCD;
calculating a mutual information entropy between adjacent ISCs, and obtaining a combination of high-frequency parts by determining a high-frequency part of the original infrasonic signal using the mutual information entropy;
calculating a similarity coefficient between the combination of high-frequency parts and the original infrasound signal, and obtaining an effective characteristic signal and an effective time-domain waveform diagram by extracting an effective characteristic component from the combination of high-frequency parts according to the similarity coefficient;
determining whether the urban pipeline leaks using the effective time-domain waveform diagram, and if the urban pipeline leaks, extracting an average peak value and a mean square amplitude of the effective characteristic signal, and analyzing a leakage aperture of the urban pipeline according to the average peak value, the mean square amplitude, and a preset aperture function, wherein the leakage aperture refers to a size and a shape of a leakage location in the urban pipeline; wherein
the determining whether the urban pipeline leaks using the effective time-domain waveform diagram includes:
determining a leakage probability and the leakage aperture of the urban pipeline by processing pipeline characteristics, transportation characteristics, and the effective time-domain waveform diagram of the urban pipeline based on a judgment model, wherein the judgment model is a machine learning model, the judgment model is obtained by training based on a plurality of second training samples with second labels, the second training samples include a plurality of sets of sample pipeline characteristics, sample transportation characteristics, and sample effective time-domain waveform diagrams of sample urban pipelines, and the second labels include leakage apertures of the sample urban pipelines;
a training process of the judgement model includes:
inputting the plurality of second training samples with the second labels into an initial judgment model;
constructing a second loss function through the second labels and results of the initial judgment model;
iteratively updating parameters of the initial judgment model by gradient descent based on the second loss function; and
in response to a second preset condition being met, completing the training process, and obtaining the judgment model, wherein the second preset condition refers to the second loss function converging or a count of iteration rounds reaching a second threshold;
generating an emergency repair plan based on the leakage aperture; and
repairing the urban pipeline at the leakage location based on the emergency repair plan.