CN-116702074-B - Water supply network pipe explosion detection method for extracting pressure high-frequency component based on wavelet decomposition

CN116702074BCN 116702074 BCN116702074 BCN 116702074BCN-116702074-B

Abstract

The invention provides a water supply network pipe explosion detection method based on wavelet decomposition and pressure high-frequency component extraction, which comprises the following steps of S1, generating an original pressure monitoring value matrix, S2, performing high-low frequency separation on the original pressure monitoring value matrix by adopting discrete wavelet transformation to generate a high-frequency disturbance value detection column vector at the current moment, S3, detecting an outlier in the high-frequency disturbance value detection column vector by adopting a COF algorithm, S4, screening the detected outlier, marking the outlier meeting the characteristic as an abnormal point, correcting the pressure value corresponding to the abnormal point after the detection of the day is finished, S5, repeating the steps, and sending out a pipe explosion alarm when the continuous abnormal moment number of the abnormal point meeting the pipe explosion characteristic exceeds a time threshold value. The invention is beneficial to more accurately reflecting the instantaneous change condition of the pressure in the water supply network, ensures the real-time property of detection, and is suitable for various water supply networks.

Inventors

TAO TAO
HUANG HUANCHUN
YU NA
XIN KUNLUN
LI SHUPING
YAN HEXIANG
WANG JIAYING

Assignees

同济大学

Dates

Publication Date: 20260508
Application Date: 20230524

Claims (6)

1. A water supply network pipe explosion detection method based on wavelet decomposition and pressure high-frequency component extraction is characterized by comprising the following steps: Step S1, sampling and preprocessing real-time high-frequency pressure data of a single monitoring point to generate an original pressure monitoring value matrix; s2, performing high-low frequency separation on the original pressure monitoring value matrix by adopting discrete wavelet transformation, reserving a high-frequency part, and generating a high-frequency disturbance value detection column vector at the current moment; s3, detecting outliers in the high-frequency disturbance value detection column vector by adopting a COF algorithm; Step S4, based on transient pipe network pressure change characteristics of the pipe explosion working condition, screening the detected outliers, marking the outliers conforming to the characteristics as abnormal points, and correcting the pressure values corresponding to the abnormal points after the detection on the same day is finished; step S5, repeating the steps S1-S4, when the number of continuous abnormal moments of the abnormal point which accords with the tube explosion characteristics exceeds a time threshold value, giving out a tube explosion alarm, Wherein, the step S1 comprises the following substeps: Step S1-1, for high-frequency pressure monitoring data of 64Hz, sampling every 10 seconds, wherein a sampling value is an average value of the pressure monitoring data in 10 seconds, and the corresponding time is t 1 = (i, j), which represents the ith and jth time in the original pressure monitoring data set; Step S1-2, extracting the high-frequency pressure monitoring data m days before the time to be detected, wherein the number of sampling data of the monitoring points every day is n, and the current time to be detected t 2 = (m, n) represents the nth time of the m th day in the original pressure monitoring data set; s1-3, preprocessing a sampled 10-second original pressure monitoring data set; step S1-4, dividing the preprocessed original pressure monitoring data set into m row vectors with the length of n, and storing the m row vectors into each row in the original pressure monitoring value matrix P in time sequence, wherein for a single monitoring point, the corresponding original pressure monitoring value matrix P is as follows: , Wherein, the element P i,j in the original pressure monitoring value matrix P represents the pressure sampling value at the j-th moment of the i-th day, the m-th row of P represents the pressure detection value of the monitoring point on the day before the time t 2 = (m, n) to be detected, Said step S2 comprises the sub-steps of: s2-1, performing multistage discrete wavelet transformation on the original pressure monitoring value matrix P, wherein the decomposition level number k is 1/2 of the maximum decomposition level number of the matrix, and 1 approximation coefficient f and k detail coefficients wi are respectively obtained through k-level wavelet decomposition; s2-2, reserving an approximation coefficient f, setting each level of detail coefficient wi as 0, and carrying out k-level wavelet reconstruction to obtain a reconstructed low-frequency pressure value matrix P 1 ; step S2-3, the difference between the original pressure monitoring value matrix P and the reconstructed low-frequency pressure value matrix P 1 is the high-frequency disturbance value matrix H of the monitoring point, , Wherein, the element H i, j in H represents the high-frequency pressure disturbance value at the j-th moment on the i-th day, and the m-th row of H represents the high-frequency pressure disturbance value of the monitoring point on the day before the time t 2 = (m, n) to be detected; step S2-4, wherein the last column of the high-frequency disturbance value matrix H is the high-frequency disturbance value detection column vector of the current time to be detected , , Wherein the high frequency disturbance value detects a column vector The element h i in (a) represents the high-frequency disturbance component extracted from the original pressure monitoring value at the same time of the i th day of the history, and h m is the high-frequency disturbance pressure value at the time to be detected.
2. The water supply network pipe explosion detection method based on wavelet decomposition and pressure high-frequency component extraction according to claim 1, wherein the method is characterized in that: wherein, the step S1-3 comprises the following substeps: S1-3-1, if the historical pressure monitoring data of the monitoring point is in the absence of less than 5min, performing linear interpolation on the pressure monitoring data in the time period, filling the missing value, and if the historical pressure monitoring data of the monitoring point is in the absence of more than 5min, replacing the historical pressure monitoring data by adopting the pressure data at the same moment in the previous day; And S1-3-2, setting a high-low threshold value by taking the pressure magnitude order under the normal working condition as a reference, and removing the pressure monitoring data which are obviously abnormal in the original pressure monitoring data set.
3. The water supply network pipe explosion detection method based on wavelet decomposition and pressure high-frequency component extraction according to claim 1, wherein the method is characterized in that: Wherein, the wavelet basis function in the step S2-1 multistage discrete wavelet transformation adopts db 4.
4. The water supply network pipe explosion detection method based on wavelet decomposition and pressure high-frequency component extraction according to claim 1, wherein the method is characterized in that: wherein, the step S3 comprises the following substeps: step S3-1, detecting the high-frequency disturbance value into a column vector Inputting a COF algorithm; Step S3-2, calculating the high-frequency disturbance value detection column vector Chain distance of each detection sample h i : , Wherein m is I.e. the high-frequency disturbance pressure value at the same time m days before the time to be detected, To detect the Euclidean distance between the (k-1) th neighboring sample and the kth neighboring sample of sample h i ; step S3-3, calculating the high-frequency disturbance value detection column vector COF value of each detection sample h i : , Step S3-4, arranging COF values of each detection sample h i from large to small according to the absolute value, wherein the maximum absolute value is the outlier in the detection sample h i ; Step S3-5, if the high-frequency disturbance value detects a column vector If the COF value of the last detection sample h i belongs to the selected outlier, the step S4 is entered for identifying the abnormal characteristics, otherwise, the step S1 is returned for detecting the tube explosion at the next moment.
5. The water supply network pipe explosion detection method based on wavelet decomposition and pressure high-frequency component according to claim 4, wherein the method is characterized in that: wherein, the step S4 comprises the following substeps: Step S4-1, calculating the high-frequency disturbance value detection column vector Average value of first (m-1) elements except current time Standard deviation sigma: , , S4-2, judging whether the outlier is abnormal according to the 3 sigma principle, and performing according to the pipe bursting pressure drop characteristic and other abnormal pressure characteristics Marking: , Wherein, the Indicating that no abnormality occurs in the pressure at the current time to be detected, Indicating that there is an abnormality in the drop in pressure, Indicating that an abnormality in pressure rise occurred; step S4-3, according to the pressure drop characteristic of the detonation tube, when When the detection method is=1, the step S5 is carried out to judge the number of continuous abnormal moments, otherwise, the step S1 is returned to carry out the tube explosion detection at the next moment; step S4-4, after the detection is finished in the same day, the method is carried out The original pressure monitoring value corresponding to the detection time of=1 or 2 is corrected, and the average value of the pressure data at the same time of the previous m days is adopted as the correction value.
6. The water supply network pipe explosion detection method based on wavelet decomposition and pressure high-frequency component according to claim 5, wherein the method is characterized in that: wherein, the step S5 comprises the following substeps: step S5-1, the time to be detected is determined The successive numbers of occurrences are noted as n, once they occur If the number is equal to 0 or 2, the tube explosion is not generated, n is set to 0, and the step S1 is performed for tube explosion detection at the next moment; And S5-2, setting a continuous abnormality judgment time threshold, judging that the pipe is burst at the moment if n exceeds the time number corresponding to the time threshold, and sending out a pipe burst alarm.

Description

Water supply network pipe explosion detection method for extracting pressure high-frequency component based on wavelet decomposition Technical Field The invention relates to a water supply network pipe explosion detection method, in particular to a water supply network pipe explosion detection method based on wavelet decomposition and pressure high-frequency component extraction. Background Pressurized pipes can age gradually over time due to corrosion, structural fatigue, cumulative effects of external environmental factors related to ground subsidence movement or third party effects (surface loading, etc.), and leakage can occur. The pipe network leakage has various expression forms, and the pipe explosion is one of the most destructive, so that a large amount of tap water can be lost in a short time, and meanwhile, the urban environment can be damaged, the quality of drinking water can be polluted, and adverse social influence can be caused. The notice about enhancing the leakage control of the public water supply network published in 2022 proposes the goal of constructing a precise, efficient, safe and long-acting water supply network leakage control mode, and the informatization and intelligent management level of leakage is required to be improved. Therefore, the rapid detection research of the water supply network pipe explosion event is developed, and the rapid detection research has important significance for repairing the pipe explosion in time, reducing the pipe network leakage and guaranteeing the water supply safety. With the development of computer technology and communication technology, data acquisition and monitoring systems (SCADA systems) have been widely used in the urban water supply industry. The SCADA system can collect data such as flow and pressure of the pipe network, realizes real-time comprehensive monitoring of the running state of the pipe network, provides strong data support for pipe explosion detection of the pipe network, and can realize pipe explosion detection through means such as data mining, machine learning, hydraulic simulation and the like. At present, a great deal of research on a water supply network pipe explosion detection method based on SCADA data drive has been developed at home and abroad, but the prior art still has the defects that for pressure data, although pressure changes caused by slow water leakage, valve operation, other maintenance operations and the like can be detected in a few minutes or a few hours, pressure transient caused by pipe explosion events often occurs in a few seconds, the existing research mostly adopts 5-15 min pressure acquisition frequency which possibly causes lagging pipe explosion detection results and lower sensitivity, the prior art is generally based on a pipe network which carries out partition metering, only pipe explosion detection at metering partition level can be realized, but due to high transformation cost, most areas in China do not carry out partition metering, a mode of centralized monitoring in a full pipe network range is still adopted, the pipe network is large in scale and has complex topological structure, and partition metering in the prior art is difficult to apply. Disclosure of Invention The present invention has been made to solve the above problems, and an object of the present invention is to provide a method for detecting a pipe explosion of a water supply network, which extracts a pressure high frequency component based on wavelet decomposition, and to provide the following technical solutions. The invention provides a water supply network pipe explosion detection method for extracting pressure high-frequency components based on wavelet decomposition, which is characterized by comprising the following steps of S1, sampling and preprocessing real-time high-frequency pressure data of a single monitoring point to generate an original pressure monitoring value matrix; s2, performing high-low frequency separation on an original pressure monitoring value matrix by adopting discrete wavelet transformation, reserving a high-frequency part, and generating a high-frequency disturbance value detection column vector at the current moment; Step S3, detecting outliers in the column vector by adopting a COF algorithm to detect the high-frequency disturbance value; Step S4, based on transient pipe network pressure change characteristics of the pipe explosion working condition, screening detected outliers, marking the outliers conforming to the characteristics as abnormal points, and correcting pressure values corresponding to the abnormal points after the detection on the same day is finished; and S5, repeating the steps S1-S4, and sending out a pipe explosion alarm when the number of continuous abnormal moments of the abnormal point which accords with the pipe explosion characteristic exceeds a time threshold. The water supply network pipe explosion detection method based on wavelet decomposition and pressure high