A pipeline leakage aperture identification method based on improved LCD

Abstract

The present application relates to the technical field of pipeline leakage, and more particularly to a pipeline leakage aperture identification method based on improved LCD, comprising: collecting pipeline infrasound leakage signal data under different leakage apertures; using prony to construct a difference equation to calculate a matching coefficient, performing LCD decomposition on the extended signal to obtain a plurality of ISC components; distinguishing effective components from ineffective components through mutual information entropy of adjacent ISC components, and performing signal reconstruction; analyzing the reconstructed signal, and extracting dominant feature parameters representing the leakage aperture; using the dominant feature parameters to construct a pipeline leakage feature vector; using the Euclidean distance to calculate the distance between the reconstructed signal leakage feature vector and the to-be-tested leakage signal feature vector, and obtaining the leakage aperture corresponding to the minimum distance. The present application solves the problems of existing neural network algorithms, such as the need for a large amount of data and low recognition efficiency; and the problem of using a single feature parameter, which cannot comprehensively reflect the leakage signal characteristics to ensure accuracy.

Description

Technical Field

[0001] This invention relates to the field of pipeline leakage technology, and in particular to a pipeline leakage aperture identification method based on an improved LCD. Background Technology

[0002] With the rapid development of the national economy and the increasing demand for energy such as natural gas, buried pipelines have become an important infrastructure in modern society as a powerful transportation tool. However, pipeline leaks occur frequently due to aging, environmental corrosion, and damage from third parties. Urban gas pipelines are mostly buried, and early leaks are often difficult to detect due to weak signals, ultimately leading to casualties, economic losses, and environmental damage. Therefore, identifying the diameter of pipeline leaks is crucial for effectively preventing pipeline leaks, fires, and explosions, reducing disaster losses, and enhancing urban public safety.

[0003] In infrasound detection of pipeline leaks, the low-frequency characteristics of infrasound make it susceptible to interference from environmental and media factors during propagation, resulting in various types of noise. Therefore, before identifying the leak aperture, the infrasound leak signal needs to be processed to reduce noise interference and highlight signal characteristics for accurate leak aperture identification. In pipeline infrasound leak signal processing, Empirical Mode Decomposition (EMD) and Local Means Decomposition (LMD) methods are suitable for analyzing non-stationary signals, offering high decomposition efficiency and strong adaptability, but they suffer from mode aliasing and endpoint effects. Local Feature Scale Decomposition (LCD) outperforms EMD and LMD methods in terms of mode aliasing and iteration count, but it still suffers from endpoint effects, significantly impacting the accuracy of leak signal analysis.

[0004] Meanwhile, current research on infrasound leakage signals in pipelines mostly focuses on leak detection and source location, with limited research on leak aperture prediction. Mei L's work, "Leak Identification Based on CS-ResNet under different leakage apertures for Water-Supply Pipeline," compresses the signal before inputting it into a neural network for leak identification. Although the signal is compressed, the data volume remains enormous, resulting in low identification efficiency. Sun J's work, "Natural gas pipeline leak aperture identification and location based on local mean decomposition analysis," uses the LMD method to process the leak signal, extracts the RMS entropy, and inputs it into a vector machine for leak aperture identification. However, this method relies on human experience and is complex. Lang X's work, "Pipeline leak aperture recognition based on wavelet packet analysis and deep belief network with ICR," uses the denoising velocity of the ultrasonic signal as a feature parameter input into a deep belief network for leak aperture identification. This method uses a single feature parameter, which cannot comprehensively reflect the characteristics of the leak signal to ensure accuracy. Summary of the Invention

[0005] To address the shortcomings of existing algorithms, this invention constructs a difference equation based on the Prony algorithm principle, extends the two ends of the leakage signal, and then performs LCD decomposition on the signal. This effectively improves the endpoint effect, prevents signal distortion, and avoids affecting the leakage aperture identification effect. Furthermore, addressing the problem that current methods for identifying leakage apertures are cumbersome and use only a single feature parameter, failing to quickly and comprehensively reflect the characteristics of the leakage signal to ensure accurate aperture identification, this invention analyzes multiple feature signals, selects advantageous feature parameters that characterize the leakage aperture, and uses the Euclidean distance method for aperture identification. This comprehensively reflects the characteristics of the leakage aperture, objectively and effectively identifies the aperture, and achieves fast and good results.

[0006] The technical solution adopted in this invention is: a method for identifying the diameter of pipe leaks based on an improved LCD, comprising the following steps:

[0007] Step 1: Under a fixed pressure, conduct a leakage test on the experimental pipeline and collect infrasonic leakage signals X under different leakage orifice diameters. i (t) data;

[0008] Step 2: Construct a difference equation using Prony to calculate the matching coefficients. Extend the signal based on the matching coefficients and perform LCD decomposition on the extended signal to obtain several ISC components.

[0009] While LCD decomposition can resolve issues such as over-envelope, under-envelope, and negative frequency generation in EMD, the decomposition results still suffer from endpoint effects. Endpoint effects arise because the two ends of the signal are non-extreme points; decomposition errors are introduced from these endpoints, affecting the results and causing severe distortion. The Prony algorithm is used to construct a difference equation to increase the number of extreme points at both ends of the signal, thereby suppressing the impact of endpoint effects. The specific process is as follows:

[0010] Step 21: Locate the infrasound leakage signal X i The times corresponding to all maxima and minima in (t) are denoted as t. 1max ,t 2max ,…,t kmax (k=1…N) and t 1min ,t 2min ,…,t kmin (k = 1…M).

[0011] Step 22: Let t be the time corresponding to the maximum and minimum values ​​that need to be expanded at the endpoints. max and t min ;

[0012] If N = M or N = M-1, then:

[0013]

[0014] If N = M + 1, then:

[0015]

[0016] Step 23: Select the infrasound leakage signal X i (t) represents the y extreme points at the endpoints, denoted as x(1)…x(y); assuming that the time interval between adjacent extreme points is equal, denoted as t=|t max -t min |; Construct the difference equation for the Lth extreme point using the Prony algorithm:

[0017] x(L)=b1x(L-1)+b2x(L-2)+…+b y x(Ly) (3)

[0018] Where L represents the extended extreme points, b1, b2, ..., b y These are the equation coefficients;

[0019] Construct equation coefficients b1, b2, ..., b yA set of linear matrix equations, from which the coefficients b1, b2, ..., b are obtained. y Then, substitute the values ​​into the difference equation and solve to obtain the Lth extreme point. If L is a maximum value, include L in the range of selected extreme points, and repeat the above steps to obtain the (L+1)th minimum point, and vice versa.

[0020] Step 24, (t) min ,x(L)),(t max Let x(L+1) and the two pairs of extreme points near the endpoints be considered as a time series, denoted as Q; where r contains the number of samples of the time-domain signal; let X i (t) is divided into several time series of length Q, denoted as Q1, Q2, ..., Q. s Calculate the matching coefficient between Q and each time series using the following formula:

[0021]

[0022] Select R w The waveform corresponding to the minimum value is extended to X. i (t) before.

[0023] For infrasound leakage signal X i (t) Perform the above steps at both ends to extend the signal. Then, perform LCD decomposition on the extended signal to reduce the endpoint effect.

[0024] Step 3: Process the ISC components decomposed in Step 2. Distinguish between effective and invalid components by the mutual information entropy of adjacent ISC components. After removing invalid components, reconstruct the signal to obtain the feature signal X. i '(t).

[0025] Infrasound leakage signal X i (t) After the improved LCD decomposition, several ISC components are obtained, which can be divided into effective components and invalid components. In traditional LCD decomposition, effective and invalid components are generally distinguished by observing the waveforms of the ISC components empirically. This method is subjective and has a large error.

[0026] Mutual information, a similarity measurement method based on information entropy theory, can capture the nonlinear statistical correlation between variables, thus measuring and reflecting the true dependence between variables to a large extent. Therefore, based on information entropy, mutual information entropy is used to quantitatively distinguish between effective and ineffective components, and its expression is as follows:

[0027]

[0028] Wherein, p(ISC) i ) and p(ISC i+1Let p(ISC) be the marginal probability distribution of two adjacent ISC components. i. ISC i+1 Let be the joint probability distribution of two adjacent ISC components. The larger the mutual information entropy value, the more identical information the adjacent ISC components contain. The dependence of invalid information on each ISC gradually decreases, while the dependence of valid information on each ISC gradually increases. Therefore, when calculating the mutual information entropy between each adjacent ISC component, a local minimum will appear. The first local minimum is the boundary between invalid and valid components.

[0029] For the original signal X i The components decomposed from (t) are ISC1, ISC2, ..., ISC n As the decomposition increases, the number of high-frequency signals in the ISC components gradually decreases, while the number of low-frequency signals gradually increases. ISC1 contains the most useful high-frequency information and contains no low-frequency noise information. n It contains the most low-frequency noise information and no high-frequency useful information. Therefore, when a local minimum occurs in the mutual information entropy between each adjacent ISC component, it means that the adjacent ISC components contain the least amount of the same information. That is, one ISC component contains high-frequency useful information while the other adjacent ISC component contains low-frequency useless information. The first local minimum is the boundary between high-frequency and low-frequency components, and it can also be the boundary between effective and ineffective components.

[0030] By removing invalid components and extracting the valid components as feature components for signal reconstruction, the feature signal X can be obtained. i '(t).

[0031] Step 4: Analyze the characteristic signal X i We analyze '(t) and extract the key characteristic parameters that can characterize the leakage pore size.

[0032] Because pipeline leakage is influenced by multiple factors, a single characteristic parameter cannot comprehensively represent the state of different leakage orifice diameters. Each characteristic parameter represents a physical meaning and has a certain relationship with the pipeline's operating state, and can be used to characterize whether the pipeline is leaking, the leakage orifice diameter, etc. However, the relationship between characteristic parameters and pipeline operating state is complex, and different characteristic parameters have different sensitivities to different pipeline operating states. A single characteristic parameter cannot comprehensively characterize the pipeline's operating state, while characterizing the pipeline's operating state with all characteristic parameters is very cumbersome and inefficient. Therefore, it is necessary to study different characteristic parameters and select advantageous parameters to efficiently and comprehensively reflect the pipeline's operating state.

[0033] Pipeline leakage can be viewed as a free-flowing jet of the medium within the pipeline, potentially triggering high-pressure impacts within the medium inside the leak hole. Therefore, the leakage generates impact sound pressure, a result of the superposition of shock waves occurring at different moments during pipe wall rupture. Thus, the size of the leak hole can be characterized by the impact characteristics of the leakage sound wave. Furthermore, since both the impulse factor and kurtosis factor can be used to characterize the abrupt changes in the impact characteristics of the leakage sound wave, one must be chosen as the dominant parameter to reflect the pipeline's operating state. Simultaneously, based on the characteristics of pipeline leakage signals, when the pipeline leak hole diameter changes, the frequency and amplitude corresponding to the peak value of the leakage signal also change accordingly. Moreover, due to attenuation during signal transmission, the signal energy differs for different leak hole diameters. Therefore, seven characteristic parameters are selected for study: energy, variance, and root mean square (RMS) in the time domain; peak frequency in the frequency domain; and impulse factor, kurtosis factor, and kurtosis in the waveform parameters.

[0034] Curves were fitted using Matlab software, and the fitted curves were compared and analyzed with the actual data curves. Ultimately, the energy value E and root mean square X, the dominant parameters characterizing the leakage pore size, were selected. rms and kurtosis X k .

[0035] Step 5: Utilize the dominant characteristic parameters energy value E and root mean square X rms and kurtosis X k Constructing the pipeline leakage feature vector Y e ;Y e It can be represented as:

[0036] Y e =[EX rms X k (6)

[0037] In the formula, e represents the size of the leakage orifice, in mm.

[0038] Step 6: Calculate the leakage feature vector Y of the measured signal X'(t) using the Euclidean distance algorithm. e 'with Y e The distance is such that the leakage orifice diameter corresponding to the minimum distance is the size of the leakage orifice diameter of the signal to be measured, X'(t).

[0039] The infrasound leakage signal X'(t) to be measured is processed using the above steps to obtain the eigenvector Ye'=

[0040] [E′X rms ′X k ′).

[0041] The leakage feature vector Y of the measured signal X'(t) is calculated using the Euclidean distance algorithm. e 'with Y e The distance is calculated using the following formula:

[0042]

[0043] In the formula, E is the leakage signal X. i The energy of X(t), rms Leakage signal X i The root mean square of (t), X k Leakage signal X i The kurtosis of X'(t), E′ is the energy of the leakage signal X'(t), X rms ′ is the root mean square of the leakage signal X'(t), X k ′ represents the kurtosis of the leakage signal X'(t).

[0044] Euclidean distance refers to the true distance between two points in m-dimensional space, and it is regarded as the similarity of signals; the shorter the distance, the more similar the signals are. In the formula, if O... e The smaller the value, the higher the overlap between the feature vectors. Calculate the Euclidean distance O between Ye' and Ye. e The arrangement determines the minimum O e If the value is specified, the corresponding leakage orifice diameter is the size of the leakage orifice diameter of the signal to be measured, X(t). Identifying the leakage orifice diameter of the pipeline can improve the accuracy of leakage orifice diameter identification, so as to formulate an efficient emergency repair plan and protect the safety of residents' lives and property.

[0045] The beneficial effects of this invention are:

[0046] 1. It can reduce signal distortion caused by the endpoint effect during LCD decomposition, quantitatively distinguish between effective and invalid components, and effectively improve the accuracy of leakage aperture identification by constructing a leakage aperture estimation model by selecting the optimal feature parameters and identifying leakage aperture through the Euclidean distance algorithm. Attached Figure Description

[0047] Figure 1 This is a flowchart illustrating the present invention;

[0048] Figure 2 This is a schematic diagram of the experimental setup;

[0049] Figure 3 This is a time-frequency diagram of the infrasound leakage signal from the pipeline;

[0050] Figure 4 It is the ISC component diagram of the signal;

[0051] Figure 5 These are root mean square curves for different leakage orifice diameters;

[0052] Figure 6 It is a variance curve of different leakage orifice diameters;

[0053] Figure 7This is a curve of the margin factor for different leakage orifice diameters;

[0054] Figure 8 These are peak frequency curves for different leakage orifice diameters;

[0055] Figure 9 These are energy curves for different leakage orifice diameters;

[0056] Figure 10 These are kurtosis curves for different leakage orifice diameters;

[0057] Figure 11 This is a pulse factor curve for different leakage orifice diameters;

[0058] Figure 12 This is a numerical diagram of the Euclidean distance for different leakage orifice diameters. Detailed Implementation

[0059] The present invention will be further described below with reference to the accompanying drawings and embodiments. The drawings are simplified schematic diagrams, which only illustrate the basic structure of the present invention in a schematic manner, and therefore only show the components related to the present invention.

[0060] like Figure 1 As shown, a method for identifying the diameter of a pipe leak based on an improved LCD includes the following steps:

[0061] This invention employs, as follows Figure 2 The experimental setup shown mainly includes a buried non-metallic pipeline, a pressure stabilizing tank, a medium distributor, inlet and outlet valves, various sensors, and leak ports at different locations. The medium distributor is used to supply fluid into the pipeline. After opening the inlet valve, fluid enters and fills the entire pipeline. The pressure stabilizing pipe is used to stabilize the pressure inside the pipeline, and the outlet valve is used to exhaust air at the end of the experiment. The sensors include four infrasound sensors, three pressure sensors, and three flow sensors. Each sensor is located at a different position in the pipeline to receive infrasound leakage signals and pressure and flow parameters at different locations. In this experiment, infrasound sensors were installed at infrasound sensor placement ports 1 and 4, respectively. The pressure and fluid flow rate inside the pipeline were adjusted based on the pressure and flow sensors to stabilize the pressure at 0.3 MPa and the flow rate at 1 m³ / s. 3 Meanwhile, the experiment was conducted using leak point 2 as the leak point, and the aperture size of leak point 2 was changed. The infrasonic leakage signals collected under different leak apertures were output through the PC terminal.

[0062] Step 1: Conduct experiments on the leaking pipeline under a pressure condition of 0.3 MPa. During the experiments, the leak orifice diameter was varied, and infrasonic signal data were collected for different orifice diameters at 0.3 MPa pressure: 1 mm, 2 mm, 3 mm, and 4 mm. The time-domain and frequency-domain plots of one set of collected leak signals are shown below. Figure 3 As shown. From Figure 3 As can be seen, the collected infrasound leakage signal from the pipeline contains a significant amount of noise due to various interference factors. Signal noise can interfere with signal analysis and even obscure valuable information. Therefore, it is necessary to improve the LCD to extract features from the original signal data, remove the interference of cluttered noise signals, and highlight the signal characteristics in order to effectively identify the size of the pipeline leak.

[0063] Step 2: Analyze the infrasound leakage signal X. i Analyze (t) to find all maxima and minima, where the times corresponding to the maxima and minima are t, respectively. 1max =0.02s,t 2max =0.05s,…,t 340max =0.94s,t 341max =0.96s and t 1min =0.04s,t 2min =0.08s,…,t 340min =0.95s,t 341min =0.98s; Since the number of maxima and minima of this signal set is the same, let the difference between the times corresponding to the maxima and minima that need to be expanded at the endpoint of this signal be 0.98s;

[0064] Selecting infrasound leakage signal X i (t) Five extreme points at the endpoints, denoted as x(1)…x(5); assuming that the time interval between adjacent extreme points is equal, denoted as t=|t max -t min |=0.031s; Using MATLAB programming, the difference equation of the extreme point is constructed according to the Prony algorithm, and the extreme point is obtained by solving the equation.

[0065] Since the last extreme point on the right side of the signal is a minimum with a time interval of 0.98s, the first extreme point of the extended signal is a maximum. Furthermore, since the time interval between adjacent extreme points is 0.031s, the time point of the first minimum and the time point of the first maximum on the right side of the extended signal are 1.011s and 1.042s, respectively. The extended maximum (1.011, 89) and minimum (1.042, -77), along with the two pairs of extreme points near the endpoint, are denoted as a time series Q, where 89 and -77 are the time-domain signal amplitudes. The value of Q is the sum of the first minimum and the second maximum near the endpoint (t...). 340max The time difference of x(2) is approximately 1.042-0.94=0.102s≈0.1s.

[0066] Due to the leakage signal X of this infrasound segment i(t) The time is 1 second, therefore X i (t) is divided into 10 time series of length Q, denoted as Q1, Q2, ..., Q. 10 Each sequence segment has 100 sampling points; the matching coefficients between Q and the 10 time series segments are calculated, and the values ​​are shown in Table 1.

[0067] Table 1 Time Series Matching Coefficients

[0068]

[0069] As shown in Table 1, the matching coefficient R between Q4 and Q is... w At the very least, select the waveform corresponding to Q4 and extend it to X. i Before (t), the extended signal is decomposed using LCD, and the decomposition result is as follows. Figure 4 As shown.

[0070] Step 3: Mutual information is a similarity measurement method based on information entropy theory. It can capture the nonlinear statistical correlation between variables, and thus can measure and reflect the true dependence between variables to a large extent. Therefore, based on information entropy, mutual information entropy can be used to quantitatively calculate and distinguish between effective and ineffective parts.

[0071] Taking ISC1 and ISC2 as examples, the mutual information entropy P(ISC1;ISC2) = 0.853 of ISC1 and ISC2 is calculated according to formula (5).

[0072] The mutual information entropy between adjacent ISC components is shown in Table 1.

[0073] Table 2 Mutual information entropy values ​​between adjacent ISC components

[0074]

[0075] As shown in Table 1, the first local minimum of the mutual information entropy of each component after the improved LCD decomposition is ISC34, indicating that the first three components are all valid components. By removing invalid components and extracting the valid components as feature components for signal reconstruction, the feature signal X can be obtained. i '(t).

[0076] Step 4: Select 30 sets of leakage signal data. After processing through the above steps, obtain characteristic signals, extract characteristic parameters, and use MATLAB software to fit and obtain fitting curves. At the same time, compare the fitting curves with the actual data curves, select the advantageous parameters that characterize the leakage pore size, and compare the fitting curves of each characteristic parameter with the actual data curves.

[0077] Depend on Figures 5-11 It can be seen that the characteristic parameters with a high degree of overlap between the fitted curve and the actual data curve are the energy value E and the root mean square X.rms and kurtosis X k The higher the curve overlap, the better the fit, and the better the ability of this characteristic parameter to characterize the leakage pore size. Therefore, the energy value E and the root mean square X are selected. rms and kurtosis X k It is an advantageous parameter for characterizing the leakage orifice size.

[0078] Step 5: Determine the dominance parameters, energy value E and root mean square X, as described in Step 4. rms and kurtosis X k Then, based on the fitting curves from MATLAB, the characteristic parameters corresponding to leakage orifice diameters of 1mm, 2mm, 3mm, and 4mm were determined to establish the leakage orifice diameter feature vector; the energy value E and root mean square X corresponding to a 1mm orifice diameter were also determined. rms and kurtosis X k The values ​​are 5878.81, 65.94, and 1.248, respectively; the energy values ​​E and root mean square X corresponding to a 2mm aperture are... rms and kurtosis X k The values ​​are 6208.14, 76.12, and 1.67, respectively; the energy values ​​E and root mean square X corresponding to a 3mm aperture are... rms and kurtosis X k The values ​​are 7457.55, 76.38, and 1.869, respectively; the energy values ​​E and root mean square X corresponding to a 4mm aperture are... rms and kurtosis X k The values ​​are 7500.83, 86.61, and 2.19 respectively; therefore, the characteristic vectors of pipe leakage for leakage orifice diameters of 1mm, 2mm, 3mm, and 4mm are:

[0079] Y1 = [5878.81 65.94 1.248]

[0080] Y2 = [6208.14 76.12 1.67]

[0081] Y3 = [7457.55 76.38 1.869]

[0082] Y4 = [7500.83 86.61 2.19]

[0083] Step 6: Randomly select a group of leakage signals with leakage hole diameters of 1mm, 2mm, 3mm and 4mm respectively, perform operations such as LCD decomposition on the signal, calculate the energy, root mean square and kurtosis, and calculate the Euclidean distance value according to formula (7).

[0084] Depend on Figure 12It can be seen that the minimum Euclidean distance for a 1mm leak hole corresponds to O1, the minimum Euclidean distance for a 2mm leak hole corresponds to O2, the minimum Euclidean distance for a 3mm leak hole corresponds to O3, and the minimum Euclidean distance for a 4mm leak hole corresponds to O4. This shows that the method can accurately identify the leak hole diameter in the pipeline.

[0085] In addition, 30 sets of data were selected for identification using this method, and the identification results are shown in Table 3.

[0086] Table 3. Results of Pipeline Leakage Diameter Identification

[0087]

[0088]

[0089] As shown in Table 3, this method can identify the size of the leak hole in the pipeline with a high accuracy rate. Furthermore, the accuracy rate of leak hole identification gradually increases with the increase of the leak hole diameter.

[0090] Based on the above-described preferred embodiments of the present invention, and through the foregoing description, those skilled in the art can make various changes and modifications without departing from the inventive concept. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for identifying the diameter of pipe leaks based on an improved LCD, characterized in that, Includes the following steps: Step 1: Collect infrasound leakage signal data of pipelines with different leakage orifice diameters; Step 2: Construct a difference equation using Prony to calculate the matching coefficients. Extend the signal based on the matching coefficients and perform LCD decomposition on the extended signal to obtain several ISC components. Step Two (Specific) include: Step 21: Locate the infrasound leakage signal X i ( t The times corresponding to all maxima and minima in ); Step 22: Let the times corresponding to the maximum and minimum values ​​of the extended endpoints be denoted as . t max and t min Calculate the difference between the times corresponding to the maximum and minimum values; The formula for the difference between the times corresponding to the maximum and minimum values ​​is: like N = M or N = M -1, then: (1) like N = M +1, then: (2) in, t kmax , t kmin for k The maximum and minimum values ​​at time 1. M , N The total number of times when the maximum and minimum values ​​are reached; Step 23: Select infrasound leakage signal X i ( t The y extreme points at the endpoints are used to construct the th extreme point according to the Prony algorithm. L Solving the difference equation at the nth extreme point yields the nth extreme point. L There are several extreme points; No. L The difference equations for the extreme points are: (3) in, L Represents the extended extreme point, b 1, b 2,…, b y These are the equation coefficients; Solving for the first L The extreme points include: like L To be the maximum value, L If a point is included within the range of selected extreme points, the process is repeated until the first extreme point is obtained. L +1 local minimum point; like L To be the minimum value, L If a point is included within the range of selected extreme points, the process is repeated until the first extreme point is obtained. L +1 maximum point; Step 24, put ( t min ,x( L )),( t max ,x( L +1)) and the two pairs of extreme points near the endpoints are taken as a time series Q, and X i ( t The time series Q is divided into several segments of the same length as Q. s Calculate time series Q and time series Q. s The matching coefficient is used to select the waveform corresponding to the smallest matching coefficient value, and then extended to... X i ( t )forward; The formula for the matching coefficient is: (4) in, r The number of samples for the time-domain signal; Step 3: Distinguish between effective and invalid components by the mutual information entropy of adjacent ISC components, and reconstruct the signal after removing invalid components; Step 4: Analyze the reconstructed signal and extract the key characteristic parameters that characterize the leakage aperture; Step 5: Construct a pipeline leakage feature vector using the dominant feature parameters; Step 6: Use Euclidean distance to calculate the distance between the leakage feature vector of the reconstructed signal and the feature vector of the leakage signal to be measured, and obtain the leakage aperture corresponding to the minimum distance.

2. The method for identifying pipe leakage aperture based on an improved LCD according to claim 1, characterized in that, The expression for distinguishing between effective and invalid components by the mutual information entropy of adjacent ISC components is: （5） in, p ( ISC i )and p ( ISC i+1 ) represents two adjacent ISC Marginal probability distribution of components p ( ISC i. , ISC i+1 ) represents the joint probability distribution of two adjacent ISC components.

3. The method for identifying pipe leakage aperture based on an improved LCD according to claim 1, characterized in that, Advantageous characteristic parameters include: energy value E Root mean square X rms and kurtosis X k .

4. The method for identifying pipe leakage aperture based on an improved LCD according to claim 1, characterized in that, The pipeline leakage feature vector is: (6) In the formula, e This represents the size of the leaking orifice.

5. The method for identifying pipe leakage aperture based on an improved LCD according to claim 1, characterized in that, The formula for calculating the distance between the leakage feature vector of the reconstructed signal and the feature vector of the leakage signal to be measured using Euclidean distance is as follows: (7) In the formula, E Leakage signal X i ( t ) energy, X rms Leakage signal X i ( t The root mean square of ) X k Leakage signal X i ( t ) cliff, Leakage signal to be tested X’ ( t ) energy, Leakage signal to be tested X’ ( t The root mean square of ) Leakage signal to be tested X’ ( t ) (the steepness of the cliff).

Images