CN115146684A - Water supply pipe network leakage detection method and device based on pipeline vibration signals - Google Patents
Water supply pipe network leakage detection method and device based on pipeline vibration signals Download PDFInfo
- Publication number
- CN115146684A CN115146684A CN202210904430.0A CN202210904430A CN115146684A CN 115146684 A CN115146684 A CN 115146684A CN 202210904430 A CN202210904430 A CN 202210904430A CN 115146684 A CN115146684 A CN 115146684A
- Authority
- CN
- China
- Prior art keywords
- pipeline
- water supply
- leakage
- vibration signal
- leakage detection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Examining Or Testing Airtightness (AREA)
Abstract
The invention discloses a water supply pipe network leakage detection method based on pipeline vibration signals, which comprises the following steps: step 1, acquiring a pipeline vibration signal of a water supply pipeline, and constructing time-vibration signal data; step 2, analyzing the time-vibration signals obtained in the step 1 to obtain a plurality of single scalar characteristic values; step 3, forming a feature vector of the vibration signal according to the multiple single scalar eigenvalues, labeling the feature vector, and forming a sample set by the label, the feature vector and corresponding pipeline information; step 4, performing supervised learning training on the pre-constructed model by using a sample set to obtain a water supply network leakage detection model; and 5, inputting the acquired field pipeline vibration signals into a water supply network leakage detection model, and outputting a judgment result of whether the pipeline has leakage or not. The invention also provides a water supply pipe network leakage detection device. The method provided by the invention can improve the accuracy and the generalization of the pipeline leakage detection.
Description
Technical Field
The invention relates to the field of municipal engineering urban water supply network leakage detection, in particular to a water supply network leakage detection method and device based on pipeline vibration signals.
Background
Water supply network leakage is the ubiquitous problem of water supply trade, and its influence can cause huge economic loss and wasting of resources, and it is up to standard to influence quality of water pressure, still can cause the road surface to collapse simultaneously and wait accident harm citizen's personal safety. Therefore, how to realize the leakage control of the pipe network is a problem to be solved urgently.
Collecting vibration signals generated by pipeline leakage is the most common leakage detection method at present, and manually using devices such as a sound bar, a ground detector and the like to perform leakage detection depending on manual experience is the most commonly used leakage detection method in water utilities. The manual listening method depends on the working experience of workers, wastes time and labor, belongs to passive leak detection, and is difficult to detect leakage loss in time, in order to reduce manual dependence, one method for improving the real-time performance of leak detection is to install noise recorders in a pipeline, the recorders set amplitude threshold values on different frequency bands according to background noise, and alarm is given when the signal amplitude exceeds the threshold values. In recent years, based on the rapid development of the internet of things technology and a micro-electromechanical system, a real-time pipe network leakage detection system depending on a distributed vibration sensor is popularized and applied, the system collects and uploads collected data to a processing center through the sensors distributed in a pipe network, and then the real-time leakage detection of the pipe network is more efficient through an algorithm, but the actual pipe network has a complex environment and more noise interference, and how to realize the leakage identification of collected pipe vibration signals becomes a difficult point.
Patent document CN110440148a discloses a method, an apparatus and a system for classifying and identifying a leakage acoustic signal, wherein the method comprises the following steps: acquiring a leakage acoustic signal of a pipeline, wherein the leakage acoustic signal is a time-amplitude signal; preprocessing the leakage signal to obtain the leakage acoustic signal and obtain a voiceprint image of the leakage acoustic signal; in the voiceprint image, the horizontal coordinate is time, the vertical coordinate is frequency, and the gray value is amplitude; inputting the voiceprint image into a convolutional neural network classification model; and carrying out classification recognition processing on the voiceprint image to obtain a classification recognition result of the leakage acoustic signal. The method can automatically identify the classification of the leakage signals, but needs a large number of data samples and has single characteristics, the model obtained by training has the problem of deviation, and meanwhile, the problem of information loss of the voiceprint image is directly extracted.
In the method, an original pipeline vibration signal sample is subjected to normalization and classification interception preprocessing to obtain a preprocessing sample set, and then the preprocessing sample is subjected to noise reduction processing and feature extraction to obtain a training sample; establishing an RNN (radio network) model, and classifying and identifying the pipeline vibration signals; establishing an RNN model, and predicting the position of a detector in the pipeline on the pipeline vibration signal; training the CNN model by adopting a training sample set, and training the RNN model by adopting a training sample set which is output by the CNN model and is classified and recognized; and sequentially processing the pipeline vibration signals acquired in real time by the CNN model and the RNN model which are trained to obtain the positioning data of the detector in the pipeline. The method realizes accurate positioning by combining a CNN model and an RNN model by improving a traditional model.
Disclosure of Invention
In order to solve the problems, the invention provides a water supply network leakage detection method based on pipeline vibration signals.
A water supply pipe network leakage detection method based on pipeline vibration signals comprises the following steps:
step 2, analyzing the time-vibration signals obtained in the step 1 to obtain a plurality of single scalar characteristic values;
step 3, forming a feature vector of the vibration signal according to the plurality of single scalar eigenvalues obtained by analysis in the step 2, labeling the feature vector according to whether the pipeline from which the vibration signal is sourced is lost, and forming a sample set by the label, the feature vector and corresponding pipeline information;
step 4, performing supervised learning training on the pre-constructed classification model by using the sample set obtained in the step 3, and after adjusting model parameters, obtaining a water supply network leakage detection model for judging whether the pipeline has leakage;
and 5, inputting the acquired field pipeline vibration signals into a water supply network leakage detection model for identification and classification, and outputting a judgment result of whether the pipeline has leakage or not.
According to the method, the time-vibration signal is analyzed to obtain a plurality of single scalar characteristic values, the plurality of single scalar characteristic values obtained through analysis form the characteristic vector of the vibration signal, the characteristic information during model training is increased, meanwhile, the pipeline information is added into the sample set, the generalization capability of the model is enhanced, and therefore the accuracy of pipeline leakage judgment is improved.
Specifically, the pipeline vibration signal is acquired by a sensor arranged on the surface of the pipeline or the surface of the valve.
Preferably, the step 2 further includes a preprocessing process, in which the time-vibration signal data is filtered by a band-pass filter, and a frequency domain range related to pipeline leakage is intercepted, so that some interfering signals are removed, and the effect of subsequent model training is ensured.
Preferably, the preferable interval of the frequency domain range is 100-1500HZ, and the leakage vibration signal ranges under different pipeline working conditions and different pipeline self conditions are different, so that the application range of the final model is widened.
Preferably, the single scalar eigenvalue obtained by analysis in the step 2 includes an energy root mean square of the vibration signal, an average Teager energy operator, an information entropy, a sample entropy and a fractal dimension, and the eigenvalue is less affected by external noise.
Specifically, the expression of the Root Mean Square (RMS) of the energy is:
wherein x (N) represents a vibration signal, and N represents the total length of the vibration signal;
the expression of the average Teager energy operator (mean TEO):
wherein x (N) represents a vibration signal, and N represents the total length of the vibration signal;
the expression of the entropy of information (ApEn):
ApEn(e,r,N)＝C e (r)-C e+1 (r)
where N denotes the total length of the vibration signal, e denotes a given embedding dimension, and L denotes the reconstructed vibration signal (L = L) 1 ,l 2 ,…,l N-e+1 ，l i ＝x i ,x i+1 ,…,x i+e-1 ，i＝1,2,…,N-e+1)，θ[d(l i ,l j )]To representStep function, d (l) i ,l j ) Represents the distance between the vectors, and r represents a given tolerance value, which is a positive value;
the expression of the sample entropy (SampEn):
wherein e represents a given embedding dimension, r represents a given tolerance value, and is a positive value;
the expression of the fractal dimension:
construction of the temporal sequence X (q, p) is:
where p =1,2, … k, p represents an initial time point, and q represents a signal length of a reconstructed time series. The mean of the curve lengths L (q) is obtained by averaging L (p, q) for all p for each q, resulting in an array of mean values L (q). Then establishing a graph of log L (q) and log L (1/q), and obtaining a Fractal Dimension (FD) through slope least square linear fitting:
preferably, the pipeline information in step 3 includes pipeline material, pipe diameter and leakage position, and generalization when improving the model training.
Specifically, the pipeline material comprises a metal material and a plastic material.
Preferably, the classification model in step 4 is constructed based on a K-nearest neighbor classification algorithm with majority voting as a classification decision.
Specifically, the training process of the classification model is as follows:
step 4.1, dividing the sample set into a training set and a testing set, wherein the ratio of the number of samples in each set is 8:2;
step 4.2, inputting the feature vectors and the corresponding labels of the training set into a K nearest neighbor classification model for supervised learning, performing 10-time cross validation on the training set, and selecting an optimal hyper-parameter K of the model according to a classification recognition result of the test set;
and 4.3, repeating the step 4.1 and the step 4.2 by trying the combination and the dimension of different feature vectors according to the required classification effect to obtain the feature vector combination and the dimension which show the best performance.
Specifically, the water supply pipe network leakage detection model in step 5 is to evaluate a plurality of dimensional characteristic values according to an input vibration signal, perform two-classification processing on an evaluation result, and output 0 or 1, where 0 represents no leakage, and 1 represents the presence of leakage, and its specific expression:
where m represents the number of samples in the training set, I () represents an indicator function,
is a feature vector of dimension n, y i ∈Y＝{c 1 ,c 2 ,…c N Denotes the class of the vector, E x Representing the feature vector to be identified, N k (E x ) Representing a set of k nearest neighbor feature vectors contained in the training set found by a distance metric when y = c j When the output result is 1, when y is not equal to c j The output result is 0.
The invention also provides a water supply network leakage detection device, which comprises a computer memory, a computer processor and a computer program which is stored in the computer memory and can be executed on the computer processor, wherein the computer memory adopts the water supply network leakage detection model; the computer processor, when executing the computer program, performs the steps of: and inputting the acquired field pipeline vibration signals into a water supply network leakage detection model, and outputting a result of whether the corresponding pipeline has leakage or not through analysis and calculation.
Compared with the prior art, the invention has the following beneficial effects:
(1) The vibration signal of the pipeline is analyzed, the characteristic information quantity of the pipeline vibration signal is increased, and the training effect of the model can be improved.
(2) The sample set composed of a plurality of single scalar characteristic values can reduce the influence of external noise on pipeline leakage detection and improve the prediction accuracy of the classification model.
(3) Additionally adding pipeline information in the sample set can improve the accuracy and the generalization of pipeline leakage detection.
Drawings
FIG. 1 is a schematic flow chart of a method for detecting leakage of a water supply pipe network based on a pipeline vibration signal according to the present invention;
fig. 2 is a confusion matrix diagram of the identification results of the water supply network leakage detection device provided in the present embodiment.
Detailed Description
As shown in fig. 1, a method for detecting water supply pipe network leakage based on pipeline vibration signals includes:
In the water supply network region in which this embodiment is located, the concentration frequency of the leakage noise is measured: the metal tube is about 1kHz or so, and the plastic tube is about 200Hz or so; however, in actual leakage detection, the frequency distribution of the collected acoustic leakage signal fluctuates due to differences in the pipe structure, the pipe material, the pipe size, the leakage opening, the fluid flow rate, and the pressure in the pipe, and therefore, the collection frequency range of the bandpass filter in this embodiment is 100 HZ to 1500HZ.
Step 2, analyzing the time-vibration signal obtained in the step 1:
the pipeline vibration signal is expressed as X = (X) 1 ,x 2 ,…,x n ) Wherein x is amplitude, n is signal length, the eigenvector of the vibration signal is E = (E) 1 ,e 2 ,…e k ) Where e is the eigenvalue and k is the eigenvector dimension, the composed pipeline feature database D can be represented as:
where m is the total number of signal samples in the database.
The single scalar characteristic value influencing leakage judgment provided in the embodiment comprises an energy root mean square of a vibration signal, an average Teager energy operator, an information entropy, a sample entropy and a fractal dimension.
Wherein the expression of Root Mean Square (RMS) of energy:
wherein x (N) represents a vibration signal, and N represents a total length of the vibration signal;
expression of the average Teager energy operator (mean TEO):
wherein x (N) represents a vibration signal, and N represents the total length of the vibration signal;
expression of information entropy (ApEn):
ApEn(e,r,N)＝C e (r)-C e+1 (r)
where N denotes the total length of the vibration signal, e denotes a given embedding dimension, and L denotes the reconstructed vibration signal (L = L) 1 ,l 2 ,…,l N-e+1 ，l i ＝x i ,x i+1 ,…,x i+e-1 ，i＝1,2,…,N-e+1)，θ[d(l i ,l j )]Representing a step function, d (l) i ,l j ) Represents the distance between the vectors, and r represents a given tolerance value, which is a positive value;
expression of sample entropy (SampEn):
wherein e represents a given embedding dimension, r represents a given tolerance, and is a positive value;
the expression for fractal dimension:
construction of the temporal sequence X (q, p) is:
where p =1,2, … k, p represents the initial time point, and q represents the signal length of the reconstructed time series. The mean of the curve lengths L (q) is obtained by averaging L (p, q) for all p for each q, resulting in an array of mean values L (q). Then establishing a graph of logL (q) and logL (1/q), and obtaining a Fractal Dimension (FD) through slope least square linear fitting:
and 3, forming a feature vector of the vibration signal according to the single scalar eigenvalue obtained by analyzing in the step 2, labeling the feature vector according to whether the pipeline from which the vibration signal is sourced has leakage or not, and forming a sample set by the label, the feature vector and corresponding pipeline information, wherein the pipeline information comprises a pipe material, a pipe diameter and a leakage position.
Step 4, performing supervised learning training on the pre-constructed model by using the sample set obtained in the step 3:
step 4.1, dividing the sample set into a training set and a testing set, wherein the ratio of the number of samples in each set is 8:2;
step 4.2, inputting the feature vectors and the corresponding labels of the training set into a K nearest neighbor classification model for supervised learning, performing 10-time cross validation on the training set, and selecting an optimal hyper-parameter K of the model according to a classification recognition result of the test set;
and 4.3, repeating the step 4.1 and the step 4.2 by trying the combination and the dimension of different feature vectors according to the required classification effect to obtain the feature vector combination and the dimension which show the best performance.
Wherein the function of the K nearest neighbor classification model is as follows:
when a new unknown pipeline signal feature vector is input into a trained model, the algorithm finds K signal feature vectors which are most adjacent to the signal in training set data, most of the K signals belong to a certain class, and then the input unknown signals are classified into the class.
Training set is D = { (E) 1 ,y 1 ),(E 2 ,y 2 ),…,(E m ,y m ) Therein of
Is an n-dimensional feature vector of which the class is y i ∈Y＝{c 1 ,c 2 ,…c N Where i =1,2, … N.
K nearest neighbor algorithm finds out training set and judgment sample through distance measurementThe distance between two signal vectors in the feature space of the nearest K training samples is the reflection of the similarity degree of the K training samples, and the Euclidean distance L is adopted p And (3) calculating:
the characteristic vector of the signal to be identified is E x Finding the sum E in the training set according to the distance x K nearest neighbor points, and the set containing the k points is marked as N k (E x ) Determining E according to a decision rule of the majority vote x Class y of (2):
wherein I is an indicator function, y i ＝c j Is 1, otherwise is 0.
Specifically, in the present embodiment, the tuning model hyperparameter K =3.
And 5, inputting the acquired field pipeline vibration signals into a water supply network leakage detection model for identification and classification, and outputting a judgment result of whether the pipeline has leakage or not.
The embodiment also provides a water supply network leakage detection device, which comprises a computer memory, a computer processor and a computer program stored in the computer memory and executable on the computer processor, wherein the computer memory adopts the water supply network leakage detection model;
the computer processor, when executing said computer program, performs the steps of: and inputting the acquired field pipeline vibration signals into a water supply network leakage detection model, and outputting a result of whether the corresponding pipeline has leakage or not through analysis and calculation.
As shown in fig. 2, which is a confusion matrix diagram of the identification result of the apparatus, the horizontal axis direction is a test value, and the vertical axis direction is a real value, it can be known from the diagram that the test samples are counted 269 in total, wherein the accuracy of the identification is 0.84 corresponding to the total correct identification 226, and the identification effect of the water supply network leakage detection apparatus provided in this embodiment is better based on the identification task under the large volume data of the water supply network.
Claims (10)
1. A water supply pipe network leakage detection method based on pipeline vibration signals is characterized by comprising the following steps:
step 1, acquiring a pipeline vibration signal of a water supply pipeline, and constructing time-vibration signal data;
step 2, analyzing the time-vibration signals obtained in the step 1 to obtain a plurality of single scalar characteristic values;
step 3, forming a feature vector of the vibration signal according to the plurality of single scalar eigenvalues obtained by analysis in the step 2, labeling the feature vector according to whether the pipeline from which the vibration signal is sourced is lost, and forming a sample set by the label, the feature vector and corresponding pipeline information;
step 4, performing supervised learning training on the pre-constructed classification model by using the sample set obtained in the step 3, and after adjusting model parameters, obtaining a water supply network leakage detection model for judging whether the pipeline has leakage;
and 5, inputting the acquired field pipeline vibration signals into a water supply network leakage detection model for identification and classification, and outputting a judgment result of whether the pipeline has leakage or not.
2. The method for detecting water supply pipe network leakage according to claim 1, wherein the pipe vibration signal is acquired by a sensor disposed on the surface of the pipe or the surface of the valve.
3. The method for detecting the leakage of the water supply network management system based on the pipeline vibration signal as claimed in claim 1, wherein the step 2 further comprises a preprocessing process of filtering the time-vibration signal data by a band-pass filter and intercepting a frequency domain range related to the pipeline leakage.
4. The method as claimed in claim 3, wherein the frequency range is 100-1500Hz.
5. The method for detecting leakage of a water supply pipe network based on the pipeline vibration signal as claimed in claim 1, wherein the single scalar eigenvalue obtained by the analysis in the step 2 comprises energy root mean square of the vibration signal, average Teager energy operator, information entropy, sample entropy and fractal dimension.
6. The water supply pipe network leakage detection method according to claim 1, wherein the pipe information in step 3 comprises pipe material, pipe diameter and leakage position.
7. The method for detecting water supply network leakage according to claim 1, wherein the classification model in step 4 is based on K-nearest neighbor classification algorithm, and majority voting is used as a classification decision.
8. The water supply pipe network leakage detection method based on the pipe vibration signal as recited in claim 7, wherein the classification model is trained by performing 10-fold cross validation on the feature vectors, selecting parameters of the classification model, and determining a K value of a final classification model.
9. The method as claimed in claim 1, wherein the model for detecting water supply network leakage in step 5 is evaluated according to the input vibration signal with a plurality of dimensional characteristic values, and the evaluation result is subjected to a binary process to output 0 or 1, wherein 0 represents no leakage and 1 represents leakage.
10. A water supply network leak detection apparatus comprising a computer memory, a computer processor, and a computer program stored in said computer memory and executable on said computer processor, wherein said computer memory has employed therein the water supply network leak detection model of claim 1; the computer processor when executing the computer program implements the steps of: and inputting the acquired on-site pipeline vibration signals into a water supply network leakage detection model, and outputting a result of whether the corresponding pipeline has leakage or not through analysis and calculation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210904430.0A CN115146684A (en) | 2022-07-29 | 2022-07-29 | Water supply pipe network leakage detection method and device based on pipeline vibration signals |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210904430.0A CN115146684A (en) | 2022-07-29 | 2022-07-29 | Water supply pipe network leakage detection method and device based on pipeline vibration signals |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115146684A true CN115146684A (en) | 2022-10-04 |
Family
ID=83413529
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210904430.0A Pending CN115146684A (en) | 2022-07-29 | 2022-07-29 | Water supply pipe network leakage detection method and device based on pipeline vibration signals |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115146684A (en) |
-
2022
- 2022-07-29 CN CN202210904430.0A patent/CN115146684A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110440148B (en) | Method, device and system for classifying and identifying leakage acoustic signals | |
| Mechefske et al. | Fault detection and diagnosis in low speed rolling element bearings Part II: The use of nearest neighbour classification | |
| CN114926462B (en) | Intelligent detection method and system for metal material surface defects | |
| CN109858408A (en) | A kind of ultrasonic signal processing method based on self-encoding encoder | |
| CN112985574B (en) | High-precision classification identification method for optical fiber distributed acoustic sensing signals based on model fusion | |
| CN117235661B (en) | AI-based direct drinking water quality monitoring method | |
| CN116805061B (en) | Leakage event judging method based on optical fiber sensing | |
| CN115459868A (en) | Millimeter wave communication performance evaluation method and system in complex environment | |
| CN116738221B (en) | Pressurized pipeline gas analysis method and system | |
| CN114487129A (en) | Flexible material damage identification method based on acoustic emission technology | |
| CN111339986B (en) | Device frequency law mining method and system based on time domain/frequency domain analysis | |
| CN115452957B (en) | Small sample metal damage identification method based on attention prototype network | |
| CN115146684A (en) | Water supply pipe network leakage detection method and device based on pipeline vibration signals | |
| CN110346032A (en) | A kind of Φ-OTDR vibration signal end-point detecting method combined based on constant false alarm with zero-crossing rate | |
| Feng | Condition Classification in Underground Pipes Based on Acoustical Characteristics. Acoustical characteristics are used to classify the structural and operational conditions in underground pipes with advanced signal classification methods | |
| CN115310499A (en) | Industrial equipment fault diagnosis system and method based on data fusion | |
| CN115062891A (en) | Method for quickly positioning leakage of long-distance pressure water delivery pipeline | |
| CN114724036A (en) | Ice crack detection method based on multi-scale learning | |
| CN116992389B (en) | False data detection method and system for Internet of things | |
| CN112528849B (en) | Structure health monitoring method based on inner product matrix and deep learning | |
| CN117113259B (en) | Coal mine state data processing method and system for predicting potential safety hazards | |
| CN116500135A (en) | Rock mass pipeline cracking water-permeable damage mode acoustic emission fractal evaluation method and device | |
| CN117951581A (en) | Pile foundation integrity assessment method based on deep learning and ultrasonic reflection | |
| CN117268814A (en) | Structural damage identification method combining recursion quantitative analysis and principal component analysis | |
| CN117786583A (en) | Fault diagnosis method and device based on variational modal decomposition and computer device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |