Underground pipe leak detection system and method

Patent metadata

Field	Value
Publication number	US-12487146-B2
Application number	US-202217678918-A
Country	US
Kind code	B2
Filing date	Feb 23, 2022
Priority date	Dec 9, 2021
Publication date	Dec 2, 2025
Grant date	Dec 2, 2025

Abstract

Provided is an underground pipe leak detection system, including a sensing device, a storage device, and a processing device. The sensing device is used for collecting a voice signal from an underground pipe during a time period. The storage device is used for storing a voice dataset, and storing the voice signal transmitted by the sensing device. The processing device may access the storage device. The processing device is configured to execute the following operations: training a classification model using the voice dataset; extracting features of the voice signal; inputting the features of the voice signal into the classification model that has been trained to determine if there is a leak in the underground pipe.

First claim

What is claimed is:

1 . An underground pipe leak detection system, comprising:
a sensing device, used for collecting a voice signal from an underground pipe during a time period; a storage device used for storing a voice dataset, and storing the voice signal transmitted by the sensing device; and a processing device, accessible to the storage device, and configured to execute the following operations: training a classification model using the voice dataset, wherein the voice dataset comprises a plurality of voice-data-with-leak and a plurality of voice-data-without-leak, and each of the voice data includes features of a historical voice signal; extracting features of the voice signal; and inputting the features of the voice signal into the classification model that has been trained to determine if there is a leak in the underground pipe; wherein each of the voice-data-with-leak in the voice dataset that is used for training the classification model includes a cleft shape parameter; and wherein the processing device is further configured to input the features of the voice signal into the classification model that has been trained, so as to determine a cleft shape of the leak of the underground pipe.

2 . The system as claimed in claim 1 , wherein the features comprise time domain features and frequency domain features; and
wherein the processing device is further configured to perform a Fast Fourier Transform on the voice signal to extract the time domain features and the frequency domain features of the voice signal.

3 . The system as claimed in claim 1 , wherein the processing device is further configured to train a regression model using the voice dataset, wherein each of the voice-data-with-leak in the voice dataset includes a cleft size parameter that corresponds to the cleft shape parameter; and
wherein the processing device is further configured to input the features of the voice signal into the regression model that has been trained, so as to calculate the cleft size of the leak of the underground pipe.

4 . The system as claimed in claim 1 , wherein each of the voice-data-with-leak in the voice dataset that is used for training the classification model includes a pipe material structure parameter; and
wherein the processing device is further configured to input the features of the voice signal into the classification model, so as to determine material and structure of the underground pipe.

5 . The system as claimed in claim 4 , wherein the processing device is further configured to train a regression model using the voice dataset, wherein each of the voice-data-with-leak in the voice dataset includes a leak distance parameter; and
wherein the processing device is further configured to input the features of the voice signal into the regression model that has been trained, so as to calculate a distance between the leak of the underground pipe and the sensing device.

6 . The system as claimed in claim 1 , wherein the processing device is further configured to calculate a location of the leak based on the distances between the leak of the underground pipe and the sensing devices at different locations during the time period, as well as the coordinates transmitted from the sensing devices.

7 . The system as claimed in claim 1 , wherein each of the voice data in the voice dataset includes an ambient interference parameter; and
wherein the processing device is further configured to execute an ambient interference removal operation, so as to remove elements of ambient interference from the voice signal.

8 . The system as claimed in claim 1 , wherein the processor device is further configured to execute an abnormal event removal operation, so as to remove elements of abnormal events from the voice signal.

9 . The system as claimed in claim 1 , wherein the sensing device further comprises:
a hydrophone, used for collecting an original signal from the underground pipe during the time period; a charge amplifier, used for amplifying output power of the original signal; a filter, used for filtering the original signal whose output power has been amplified, so as to remove noises whose frequency is not in a specific range from the original signal; an A/D converter, used for converting the original signal that has been filtered from an analog signal to a digital signal, wherein the original signal that that has been converted is the voice signal; and a processor, used for transmitting the voice signal to the storage device.

10 . An underground pipe leak detection method, comprising:
using a processing device to train a classification model using the voice dataset, wherein the voice dataset comprises a plurality of voice-data-with-leak and a plurality of voice-data-without-leak, and each of the voice data includes features of a historical voice signal; using a sensing device to collect a voice signal from an underground pipe during a time period; using the processing device to extract features of the voice signal; using the processing device to input the features of the voice signal into the classification model that has been trained to determine if there is a leak in the underground pipe; and using the processing device to input the features of the voice signal into the classification model that has been trained, so as to determine a cleft shape of the leak of the underground pipe; wherein each of the voice-data-with-leak in the voice dataset that is used for training the classification model includes a cleft shape parameter.

11 . The method as claimed in claim 10 , wherein the features comprise time domain features and frequency domain features; and
wherein the method further comprises: using the processing device to perform a Fast Fourier Transform on the voice signal to extract the time domain features and the frequency domain features of the voice signal.

12 . The method as claimed in claim 10 , further comprising:
using the processing device to train a regression model using the voice dataset, wherein each of the voice-data-with-leak in the voice dataset includes a cleft size parameter that corresponds to the cleft shape parameter; and using the processing device to input the features of the voice signal into the regression model that has been trained, so as to calculate the cleft size of the leak of the underground pipe.

13 . The method as claimed in claim 10 , further comprising:
using the processing device to input the features of the voice signal into the classification model, so as to determine the material and structure of the underground pipe; wherein each of the voice-data-with-leak in the voice dataset that is used for training the classification model includes a pipe material structure parameter.

14 . The method as claimed in claim 13 , further comprising:
using the processing device to train a regression model using the voice dataset, wherein each of the voice-data-with-leak in the voice dataset includes a leak distance parameter; and using the processing device to input the features of the voice signal into the regression model that has been trained, so as to calculate the distance between the leak of the underground pipe and the sensing device.

15 . The method as claimed in claim 14 , further comprising:
using the processing device to calculate the location of the leak based on the distances between the leak of the underground pipe and the sensing devices at different locations during the time period, as well as the coordinates transmitted from the sensing devices.

16 . The method as claimed in claim 10 , furt

Assignees

• Ind Tech Res Inst

Inventors

• Fan Sheng-Cyuan
• Yeh Che-Wei
• Huang Wun-Sheng
• He Long
• Chu Chien-Cheng
• Chiu Hsin-Yu
• Tsai Feng-Jie
• Hsu Fu-Te

Classifications

• G06F18/214

  Generating training patterns; Bootstrap methods, e.g. bagging or boosting · CPC title

• G01M3/007

  Leak detector calibration, standard leaks (G01M3/207 takes precedence) · CPC title

• G06F17/142

  Fast Fourier transforms, e.g. using a Cooley-Tukey type algorithm · CPC title

• G06N20/00

  Machine learning · CPC title

• G01M3/243 Primary

  for pipes · CPC title