A pipeline leakage positioning method and application based on multimodal sensor transient signal fusion

Abstract

The application discloses a pipeline leakage positioning method based on multimodal sensor transient signal fusion, comprising the following steps: step one, obtaining a difference value according to pressure sensor measurement values before and after leakage, and continuously measuring to obtain N pressure difference vectors; step two, constraining an expected signal power according to a constraint condition and minimizing total power to obtain a corresponding possible leakage position x; step three, obtaining a quality function of the possible leakage position x; and step four, determining the leakage position by a maximum probability value; the pipeline leakage positioning method provided by the application has good performance in single-leakage and multi-leakage cases; in the single-leakage case, the fusion algorithm provides accurate positioning estimation, even in the presence of high-level noise; for the multi-leakage case, the fusion method is superior to other three methods; therefore, the fusion method has good performance in leakage detection and positioning.

Description

Technical Field

[0001] This invention relates to the field of pipeline leak detection, and specifically to a pipeline leak location method based on the fusion of transient signals from multimodal sensors. Background Technology

[0002] Pipeline transportation plays a crucial role in industrial production and daily life. As an important infrastructure in my country, its advantages, such as large capacity, simple construction, low cost, and ease of control, are unparalleled. However, pipelines face problems such as corrosion, aging, and external forces. These problems, catalyzed by internal and external factors, lead to frequent pipeline leaks. Pipeline leaks not only cause enormous waste of resources but also inflict serious economic losses. Therefore, ensuring the safety and reliability of pipeline systems and the timely detection and location of pipeline leaks are of paramount importance in industrial production and daily life.

[0003] Leak detection and location research has been ongoing for decades, and various commercially available leak detection technologies have been developed, ranging from simple physical detection to acoustic techniques such as vibration signals, negative pressure waves, and sound waves. In the past decade, transient-based technologies have attracted significant attention in leak detection and location. Fluid transient-based defect detection methods are a promising and versatile approach that has been extensively studied in recent years. This method identifies and locates defects in piping systems by actively introducing hydraulic waves, measuring the pressure response at a specified location, and analyzing the measured signals.

[0004] However, in practice, the transient wave reflections from small leaks are very weak and easily affected by uncertainties such as noise, traffic, mechanical equipment, and turbulence. Therefore, the application of existing methods is often limited. From an information theory perspective, there are two methods to improve the robustness of leak location. The first method is to repeat the experiment multiple times, using the average of multiple transient measurements; wavelet transform methods can accurately locate small leaks. However, repeating transient experiments is time-consuming, interrupts water supply, and can easily cause pipeline structural fatigue, making it impractical. The second solution is to conduct only one transient experiment, but measure and use pressure signals from multiple sensors at different locations. As more sensors are installed in newly built urban pipe networks to monitor their health, multi-sensor strategies are becoming increasingly practical. Since leak characteristics are deterministic and appear in the signals of each sensor, while other fluctuations are random, multi-sensor strategies hold promise for improving leak detection in noisy environments.

[0005] Multi-sensor fusion systems and methods have been applied to sensor networks, automatic control, and video / image processing. Probability theory, fuzzy set theory, and Dempster-Shafer evidence theory have been used to quantify data imperfections in information fusion. Probability distributions represent data uncertainty, fuzzy set theory can represent data fuzziness, and Dempster-Shafer evidence theory (DSET) is more general and flexible, capable of describing uncertain and fuzzy data. In leak location problems, due to system complexity, the pressure data measured by each sensor is affected by noise or uncertainty, but provides evidence about the presence and location of the leak. This type of evidence is difficult to model using probability distributions because, in practical applications, even the distribution of arbitrary uncertainty in the data is often unknown. Furthermore, transient wave data is readily apparent. Summary of the Invention

[0006] This invention designs and develops a pipeline leak location method based on the fusion of transient signals from multimodal sensors. The purpose of this invention is to solve the problems of inaccurate location of small leaks and susceptibility to interference by setting up multimodal sensors.

[0007] The technical solution provided by this invention is as follows: A pipeline leak location method based on the fusion of transient signals from multimodal sensors includes the following steps: Step 1: Obtain the difference based on the pressure sensor readings before and after the leak. And by continuously measuring, N pressure difference vectors are obtained. ; Step 2, according to the above By constraining the desired signal power and minimizing the total power, the corresponding possible leakage location x is obtained; Wherein, the constraint condition is

[0008] as well as The desired signal power

[0009]

[0010] In the formula, the optimal weight vector w is the solution vector of the optimization problem, R is the estimator of the covariance matrix of the received signal, G(x) is the position function, β is the control parameter, and c is an arbitrary constant; Step 3: Obtain the mass function of the possible leak location x.

[0011]

[0012]

[0013]

[0014] Data fusion is performed on the mass function of the possible leak location x:

[0015]

[0016]

[0017]

[0018] In the formula, This is the maximum power. The minimum power is represented by "L" and "NL" to indicate whether there is leakage or not. Let the leakage mass function be... For non-leakage mass function, For the joint mass function, To identify the frame, power set K is the conflict coefficient. Let m1 and m2 be an empty set, and m1 and m2 be a recognition frame. Two mass functions on; Step 4: The location of the leak is determined by the maximum probability.

[0019] Preferably, step one further includes: selecting J frequencies, denoted as... ;as well as The pressure head at the pressure sensor is obtained according to different given frequencies.

[0020] Preferably, in step two, the objective function is constructed using the Lagrange method. Later obtained

[0021] In the formula, λ is the Lagrange multiplier. It is the smallest eigenvalue of the covariance matrix R. is the second smallest eigenvalue of R, k is a constant coefficient, J is the frequency number, M is the pressure sensor serial number, and I is the identity matrix; Wherein, the optimal weight vector w is a matrix The minimum eigenvalue corresponds to the eigenvector of λ, and the possible leakage location x is obtained by calculating the expected signal power.

[0022] Preferably, in step three, the mass function of the possible leakage location x is fused using the Dempster combination rule.

[0023] The pipeline leak location method based on the fusion of transient signals from multimodal sensors is applied to pipeline leak detection.

[0024] The beneficial effects of this invention are as follows: To detect a small leak in a water pipe with noise levels of 100 or less, this invention proposes a leak localization scheme that fuses multi-sensor data within the Dempster-Shafer evidence framework. The performance of the fusion method in single-leakage and multi-leakage scenarios is discussed through comparison with three spectrum-based methods. In the case of a single leak, the fusion algorithm provides accurate localization estimates, even with high levels of noise. For multi-leakage scenarios, the fusion method outperforms the other three methods. Therefore, the fusion method exhibits excellent performance in leak detection and localization. Attached Figure Description

[0025] Figure 1 This is a schematic diagram of the transient pipeline model described in this invention.

[0026] Figure 2a This is a schematic diagram of the calculation results for locating a single leak using the MFP method at a signal-to-noise ratio of 0dB, as described in Test Example 1 of this invention.

[0027] Figure 2b This is a schematic diagram of the calculation results for locating a single leak using Capon's BF method at a signal-to-noise ratio of 0 dB, as described in Test Example 1 of this invention.

[0028] Figure 2c This is a schematic diagram of the calculation results for locating a single leak using the MUSIC method at a signal-to-noise ratio of 0dB, as described in Test Example 1 of this invention.

[0029] Figure 2d This is a schematic diagram of the calculation results for locating a single leak using the DS-MUSIC-like method at a signal-to-noise ratio of 0dB, as described in Test Example 1 of this invention.

[0030] Figure 3a This is a schematic diagram of the calculation results for locating a single leak using the MFP method at a signal-to-noise ratio of 0dB, as described in Test Example 2 of this invention.

[0031] Figure 3b This is a schematic diagram of the calculation results for locating a single leak using Capon's BF method at a signal-to-noise ratio of 0 dB, as described in Test Example 2 of this invention.

[0032] Figure 3cThis is a schematic diagram of the calculation results for locating a single leak using the MUSIC method at a signal-to-noise ratio of 0dB, as described in Test Example 2 of this invention.

[0033] Figure 3d This is a schematic diagram of the calculation results for locating a single leak using the DS-MUSIC-like method at a signal-to-noise ratio of 0dB, as described in Test Example 2 of this invention.

[0034] Figure 4a This is a schematic diagram of the calculation results for locating a single leak using the MFP method at a signal-to-noise ratio of -40dB, as described in Test Example 2 of this invention.

[0035] Figure 4b This is a schematic diagram of the calculation results for locating a single leak using Capon's BF method at a signal-to-noise ratio of -40dB, as described in Test Example 2 of this invention.

[0036] Figure 4c This is a schematic diagram of the calculation results for locating a single leak using the MUSIC method at a signal-to-noise ratio of -40dB, as described in Test Example 2 of this invention.

[0037] Figure 4d This is a schematic diagram of the calculation results for locating a single leak using the DS-MUSIC-like method at a signal-to-noise ratio of -40dB, as described in Test Example 2 of this invention.

[0038] Figure 5a This is a schematic diagram of the calculation results for locating two leaks using the MFP method at a signal-to-noise ratio of -30dB, as described in Test Example 3 of this invention.

[0039] Figure 5b This is a schematic diagram of the calculation results for locating two leaks using Capon's BF method at a signal-to-noise ratio of -30dB, as described in Test Example 3 of this invention.

[0040] Figure 5c This is a schematic diagram of the calculation results for locating two leaks using the MUSIC method at a signal-to-noise ratio of -30dB, as described in Test Example 3 of this invention.

[0041] Figure 5d This is a schematic diagram of the calculation results for locating two leaks using the DS-MUSIC-like method at a signal-to-noise ratio of -30dB, as described in Test Example 3 of this invention.

[0042] Figure 6a This is a schematic diagram of the calculation results for locating two leaks using the MFP method at a signal-to-noise ratio of -30dB, as described in Test Example 4 of this invention.

[0043] Figure 6bThis is a schematic diagram of the calculation results for locating two leaks using Capon's BF method at a signal-to-noise ratio of -30dB, as described in Test Example 4 of this invention.

[0044] Figure 6c This is a schematic diagram of the calculation results for locating two leaks using the MUSIC method at a signal-to-noise ratio of -30dB, as described in Test Example 4 of this invention.

[0045] Figure 6d This is a schematic diagram of the calculation results for locating two leaks using the DS-MUSIC-like method at a signal-to-noise ratio of -30dB, as described in Test Example 4 of this invention.

[0046] Figure 7a This is a schematic diagram of the calculation results for locating two leaks using the DS-MUSIC-like method at a signal-to-noise ratio of -30dB, as described in Test Example 5 of this invention (fusion data sensor positions are 1600m, 1800m and 2000m; 1500m, 1600m and 2000m).

[0047] Figure 7b This is a schematic diagram of the calculation results for locating two leaks using the DS-MUSIC-like method at a signal-to-noise ratio of -30dB, as described in Test Example 5 of this invention (fusion data sensor positions are 800m, 1600m and 2000m; 1500m, 1600m and 2000m).

[0048] Figure 7c This is a schematic diagram of the calculation results for locating two leaks using the DS-MUSIC-like method at a signal-to-noise ratio of -30dB, as described in Test Example 5 of this invention (fusion data sensor positions are 800m, 1600m and 2000m; 600m, 1600m and 2000m). Detailed Implementation

[0049] The present invention will now be described in further detail with reference to the accompanying drawings, so that those skilled in the art can implement it based on the description.

[0050] This invention provides a pipeline leak location method based on the fusion of transient signals from multimodal sensors (DS-MUSIC-Like pipeline leak location method), comprising the following steps: Step 1: Select J frequencies, and denote them as follows: The pressure head at the pressure sensor is obtained according to different given frequencies; Step 2: Obtain the difference based on the pressure sensor readings before and after the leak. And by continuously measuring, N pressure difference vectors are obtained. ; Step 3, according to the above By constraining the desired signal power and minimizing the total power, the corresponding possible leakage location x is obtained; Wherein, the constraint condition is

[0051] as well as The desired signal power

[0052]

[0053] In the formula, the optimal weight vector w is the solution vector of the optimization problem, R is the estimator of the covariance matrix of the received signal, G(x) is the position function, β is the control parameter, and c is an arbitrary constant; Step 4: Obtain the mass function of the possible leak location x.

[0054]

[0055]

[0056]

[0057] The mass function of the possible leak location x is fused using the following formula:

[0058]

[0059]

[0060]

[0061] In the formula, This is the maximum power. The minimum power is represented by "L" and "NL" to indicate whether there is leakage or not. Let the leakage mass function be... For non-leakage mass function, For the joint mass function, To identify the frame, power set K is the conflict coefficient. Let m1 and m2 be an empty set, and m1 and m2 be a recognition frame. Two mass functions on; Step 5: The leak location is determined by the maximum probability, using the following formula:

[0062] In another embodiment, in step two, as Figure 1 As shown, a transient pipeline model is established using a water tank-pipe-valve system as the research object. A pipeline of length l is defined by two water tanks, with the upstream water tank at position 110. Downstream water tank position 120 Assume there is a sensor 140 near the downstream water tank 120, its location is... The location of leak point 130 is determined by express.

[0063] Based on the law of conservation of mass and Newton's second law, by taking a specific small unit, we can obtain the continuity equation and momentum equation for one-dimensional unsteady flow, and thus the basic differential equations for flow rate and pressure changes during the flow process inside the pipe.

[0064]

[0065] In the formula, P is the pipe pressure, V is the water velocity, ρ is the water density, a is the pressure wave velocity, and g is the acceleration due to gravity. Let f be the angle between the pipe and the horizontal plane, f be the Darcy-Weisbach friction coefficient, D be the inner diameter of the pipe, t be time, and x be the distance from the upstream end of the pipe. According to the transfer matrix method, at the pressure sensor in the pipeline The state variables can be represented as

[0066] In the formula, q(x) and h(x) are the flow rate function and pressure function of x, respectively; in,

[0067]

[0068]

[0069] In the formula, C d and A L Here, μ represents the flow coefficient and the flow area of ​​the leakage opening, respectively; μ is the propagation function; and i is the number of pipe segments. For pressure head; For a given angular frequency The pressure head at the pressure sensor can be obtained as follows:

[0070]

[0071]

[0072] make This is expressed as the difference between the pressure sensor readings before and after the leak, and is calculated as follows:

[0073] It can be obtained The general model is

[0074] This model shows that the head pressure difference is related to the location of the leak, so we will use [the following]... To conduct the test.

[0075] In another embodiment, step three employs an algorithm for locating pipe leaks within a beamforming framework, namely the MUSIC-Like algorithm. This algorithm utilizes data collected from pressure sensors. The optimization problem, which involves constraining the desired signal power while minimizing the total power, can be defined as follows:

[0076]

[0077] In the formula, the weight vector w is the solution vector of the optimization problem, R is the estimator of the covariance matrix of the received signal, G(x) is the position function, β is the control parameter, and c is an arbitrary constant; N pressure difference vectors are obtained through continuous measurement. The covariance matrix of the data is obtained.

[0078] Solve using the Lagrange method and construct the objective function.

[0079] In the formula, λ is the Lagrange multiplier; Differentiating w in equation (15) and setting the result to 0, we obtain

[0080] Equation (16) can be rewritten as follows

[0081] in,

[0082] In the formula, It is the smallest eigenvalue of the covariance matrix R. is the second smallest eigenvalue of R, k is a constant coefficient, J is the frequency number, M is the pressure sensor serial number, and I is the identity matrix; As can be seen from equation (16), the optimization problem defined in equation (12) is transformed into an eigenvalue problem, and the optimal weight vector w is a matrix. The smallest eigenvalue corresponds to the eigenvector of λ, and the corresponding position is obtained by calculating the signal power spectrum, i.e.

[0083] In another embodiment, step four, obtaining the quality function for each location and performing data fusion on the quality function, includes: Dempster-Shafer evidence theory, also known as belief function theory, was initially proposed by A.P. Dempster to solve multi-valued mapping problems using upper and lower bound probabilities. G. Shafer further developed the evidence theory, introducing the concept of a belief function, forming a mathematical approach to handling uncertain reasoning using "evidence" and "combination." DS theory is a generalization of Bayesian inference methods, requiring no prior knowledge of probabilities and effectively representing "uncertainty."

[0084] Hypothetical recognition framework It is a finite set containing n mutually exclusive elements. yes The power set of can be represented as:

[0085] mass function of the uncertainty of an event Quantization, with the following constraints:

[0086] Where A is Any subset of m is called the focal element of m if m(A) > 0, and The core element of m It is an empty set. The mass value represents the degree to which the corresponding evidence supports A. Specifically, the mass function is completely negligible if and only if m(Θ) = 1; The quality function m(A) quantifies the confidence level of the actual value; it can be expressed as the relevant confidence function and likelihood function for all... It can be obtained

[0087]

[0088] Wherein, the trust function (22) refers to the sum of probabilities of all subsets of the hypothesis, the likelihood function (23) refers to the sum of probabilities that the intersection with the hypothesis is not empty, and the trust function Bel(A) and the likelihood function Pl(A) are the lower and upper bounds of the probability that hypothesis A is true, obtained from the quality function m. And for all Bel(A)≤Pl(A); In evidence theory, for the identification frame Given a hypothesis A, the confidence function Bel(A) and likelihood function Pl(A) are calculated based on BPA to form a confidence interval [Bel(A), Pl(A)], which represents the degree of certainty about the hypothesis. Let m1 and m2 be a recognition frame. The two quality functions on the fetch correspond to two pieces of evidence from independent and fully reliable sources, respectively. Their combination function, denoted according to Dempster's evidence combination rule, is denoted as [evidence / function]. Recorded as:

[0089] in,

[0090] K is used to represent the degree of conflict between two pieces of evidence. The larger the K value, the more serious the conflict between the two pieces of evidence, and the less information their combination contains. Generally speaking, for n mass functions The above rules for combining evidence can be generalized as follows:

[0091] K is defined as

[0092] Consider two scenarios for the pipeline: leaking and non-leaking, denoted by "L" and "NL" respectively. Define the identification framework. , power set , Given an empty set, the mass function of a possible leakage location x is defined as follows:

[0093]

[0094]

[0095]

[0096] In the formula, This is the maximum power. This is the minimum power value. Let the leakage mass function be... For non-leakage mass function, This is the joint mass function.

[0097] As long as the distribution of the mass function is known, the Dempster combination rule can be used to fuse information from different sensors. For a potential leak location x, the combined mass function from two sensors is:

[0098]

[0099]

[0100]

[0101] In the formula, K is the conflict coefficient.

[0102] The location of the leak is determined by the maximum probability, i.e.

[0103] Test Example 1 Leakage location performance test under single leak and single sensor conditions This test case considers the estimation of a single leak in the case of a single sensor, and the leak location. Assuming total leakage parameters (effective leakage size) Simulating the propagation of transient waves, sensor position DS fusion algorithm through and The obtained mass functions m1 and m2 are then fused. The resonant frequency and anti-resonant frequency are then determined. Used for leak detection, among which The sample size of the correlation matrix RCM is 620, i.e., N = 620. Under these parameters, the DS fusion algorithm is applied and compared with three other algorithms: MFP, Capon's beamforming (BF), and MUSIC. MFP and Capon are, in principle, beamforming methods, while MUSIC is a representative of subspace-based methods, requiring knowledge of the number of sources to perform accurate subspace decomposition of the correlation matrix. The three beamforming algorithms mentioned above correspond to candidate leak locations. Different designs of the weighting vector for the function. In this case, the spatial power spectrum function reaches its maximum value at the actual leakage location. For example... Figure 2a , 2b As shown in Figures 2c and 2d, when the signal-to-noise ratio is high, setting the signal-to-noise ratio to 0dB yields results from... Figure 2a , 2bThe leak localization results for 2c and 2d show that, unlike the MFP method which has a wide main lobe, Capon's BF and MUSIC methods successfully achieved narrow peaks. The MFP method retains a large number of sidelobes, with high sidelobes at 600m, 800m, and 1200m. Capon's BF and MUSIC methods retain sidelobes at 600m. Compared to the above three methods, the DS-MUSIC-Like method provided by this invention successfully achieved narrow peaks and removed all sidelobes.

[0104] Test Example 2 Leakage location performance test under single leak dual sensor conditions This test case considers the estimation of a single leak in a dual-sensor scenario, including the leak location. Assuming total leakage parameters (effective leakage size) Simulating the propagation of transient waves, sensor position , The DS fusion algorithm, through... and The obtained mass function m1, and The obtained mass function m2 is then fused with mass function m1. For example... Figure 3a , 3b As shown in 3c and 3d, from Figure 3a , 3b The leak localization results for 3c and 3d show that, unlike the MFP method which has a wide main lobe and a high secondary lobe around 1600m, Capon's BF, MUSIC, and DS-MUSIC-like methods all successfully achieved narrow peaks and removed all sidelobes. These three methods are superior to MFP in suppressing lateral lobes and fluctuations. Figure 4a , 4b As shown in 4c and 4d, for a relatively small signal-to-noise ratio, Figure 4a , 4b Figures 4c and 4d show the case with a signal-to-noise ratio of -40 dB. It can be observed that the performance of all algorithms decreases with increasing noise levels. This is reflected in the fact that all four algorithms return results with sidelobes, but can still roughly locate the leak. Although all four methods can accurately estimate the leak, the presence of high sidelobes can misidentify them as leaks, affecting the accuracy of leak location, especially when the number of leaks is unknown. The MUSIC and DS-MUSIC-like methods perform better in suppressing lateral lobes and fluctuations.

[0105] Test Example 3 Leakage location performance test under dual leakage single sensor conditions This test case considers a dual-leakage scenario with a single sensor, i.e. and There are two leaks at this location, with actual dimensions of [missing information]. and Sensor position DS fusion algorithm through and The obtained quality functions m1 and m2 are fused. Assuming the signal-to-noise ratio of the measured noise is -30dB at all frequencies, and the sample size JM = 620, ... Figure 5a , 5b As shown in Figures 5c and 5d, a local maximum exists near each actual leak in each figure, meaning all four methods can accurately locate two leaks. However, the first three algorithms have sidelobes, especially the MFP algorithm, which has a wide main lobe and several sidelobes, particularly a very high sidelobe around 1400m. The presence of high sidelobes can interfere with leak localization, as they may be incorrectly identified as leaks, especially when the number of leaks is unknown. Figure 5a , 5b As can be seen in 5c and 5d, the DS fusion algorithm curve is smooth and has a significant effect on sidelobe suppression.

[0106] Test Example 4 Leakage location performance test under dual leakage and dual sensor conditions This section considers the dual-leakage scenario with dual sensors, i.e. and There are two leaks at this location, with actual dimensions of [missing information]. and The sensor position is... The DS fusion algorithm, through... and The obtained mass function m1, and The obtained quality function m2 is fused with quality function m1. Assuming the signal-to-noise ratio of the measured noise is -30dB at all frequencies and the sample size JM = 620, ... Figure 6a , 6b As shown in Figures 6c and 6d, each method exhibits sidelobes to varying degrees, with higher sidelobes appearing at certain locations, such as 1700m, which can affect the determination of the leak location. Compared to the three methods mentioned above, the DS-MUSIC-Like algorithm effectively suppresses most sidelobes through data fusion and can pinpoint the leak location. Although the DS-MUSIC-Like method can accurately estimate leaks, the presence of high sidelobes can lead to misidentification as leaks, affecting the accuracy of the leak location, especially when the amount of leakage is unknown.

[0107] Test Example 5 Leakage location performance test under dual leak sensor position change conditions This test case considers a dual-leakage scenario, i.e., in and There are two leaks at this location, with actual dimensions of [missing information]. and In each test, three sensors are deployed in the pipeline. The mass functions obtained from different sensor locations are fused to evaluate the fusion performance. This is achieved by fusing data collected from three different sensor locations, such as... Figure 7a , 7b As shown in Figure 7c, although each test accurately detected the leak location, the downstream sensor data fusion result had fewer sidelobes and a smoother fusion curve compared to the upstream sensor data fusion result, while the upstream data showed more obvious peaks. Based on transient wave characteristics, it is speculated that the reflected waves received by the upstream sensor caused the detection results to have a large number of sidelobes and peaks. In the fusion experiment, placing the sensor downstream resulted in more accurate leak detection and location. Furthermore, in practical applications, leaks occur randomly and can occur anywhere in the pipeline. The sensor placement strategy considers all possible leaks by averaging the maximum measured information and random leak parameters. Therefore, the sensor location represents the optimal set of all leak locations.

[0108] This invention utilizes uncertain noise and transient wave measurements to address the localization problem of pipeline leaks. A method is proposed to improve leak location accuracy by introducing multi-sensor measurements. Within the framework of the Dempster-Schafer evidence theory, leak information measured by multiple sensors is extracted and fused. A multi-sensor-based leak location algorithm is presented. Simulation and experimental results in Test Examples 1-5 demonstrate that this method can effectively fuse multi-sensor information. This method can detect small leaks hidden under strong noise, improving location accuracy and precision.

[0109] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. They can be applied to various fields suitable for the present invention. For those skilled in the art, other modifications can be easily made. Therefore, without departing from the general concept defined by the claims and their equivalents, the present invention is not limited to the specific details and illustrations shown and described herein.

Claims

1. A pipeline leak location method based on transient signal fusion from multimodal sensors, characterized in that, The steps include the following: Step 1: Obtain the difference based on the pressure sensor readings before and after the leak. And by continuously measuring, N pressure difference vectors are obtained. ; Step 2, according to the above By constraining the desired signal power and minimizing the total power, the corresponding possible leakage location x is obtained; Wherein, the constraint condition is ； ;as well as The desired signal power ; ； In the formula, the optimal weight vector w is the solution vector of the optimization problem, R is the estimator of the covariance matrix of the received signal, G(x) is the position function, β is the control parameter, and c is an arbitrary constant; Step 3: Obtain the mass function of the possible leak location x. ； ； ； ； Data fusion is performed on the mass function of the possible leak location x: ； ； ； ； In the formula, This is the maximum power. The minimum power is represented by "L" and "NL" to indicate whether there is leakage or not. Let the leakage mass function be... For non-leakage mass function, For the joint mass function, To identify the frame, power set K is the conflict coefficient. Let m1 and m2 be an empty set, and m1 and m2 be a recognition frame. Two mass functions on; Step 4: The location of the leak is determined by the maximum probability. .

2. The pipeline leak location method based on multimodal sensor transient signal fusion as described in claim 1, characterized in that, Step one further includes: selecting J frequencies, denoted as... ; and the pressure head at the pressure sensor is obtained according to different given frequencies.

3. The pipeline leak location method based on multimodal sensor transient signal fusion as described in claim 1, characterized in that, In step two, the objective function is constructed using the Lagrange method. Later obtained ; In the formula, λ is the Lagrange multiplier. It is the smallest eigenvalue of the covariance matrix R. is the second smallest eigenvalue of R, k is a constant coefficient, J is the frequency number, M is the pressure sensor serial number, and I is the identity matrix; Wherein, the optimal weight vector w is a matrix The minimum eigenvalue corresponds to the eigenvector of λ, and the possible leakage location x is obtained by calculating the expected signal power.

4. The pipeline leak location method based on multimodal sensor transient signal fusion as described in claim 1, characterized in that, In step three, the mass function of the possible leakage location x is fused using the Dempster combination rule.

5. Application of the pipeline leak location method based on multimodal sensor transient signal fusion as described in claim 1 in pipeline leak detection.

Images