Pipeline damage monitoring method and device

Abstract

An embodiment of the present application provides a pipeline damage monitoring method and device. The method includes: obtaining a monitoring signal of a pipeline to be measured during its in-service period and a pre-determined reference Mahalanobis space; extracting features from the monitoring signal to obtain a monitoring feature vector of the monitoring signal; determining a monitoring Mahalanobis distance of the monitoring signal according to the monitoring feature vector and a monitoring feature matrix constructed based on the monitoring feature vector; comparing the monitoring Mahalanobis distance with the reference Mahalanobis space to obtain a deviation degree between the monitoring Mahalanobis distance and the reference Mahalanobis space; and determining a damage state of the pipeline to be measured according to the deviation degree.

Description

Technical Field

[0001] The present application relates to the field of non-destructive testing, and relates to, but is not limited to, a pipeline damage monitoring method and device. Background Art

[0002] In recent years, with the rapid economic development, the demand for energy such as oil and gas has been increasing day by day. On the one hand, due to advantages such as small land occupation, large transportation volume, and low and stable transportation cost, pressure pipelines have become the main means of energy transportation; on the other hand, due to the further expansion of the pipe network coverage area and the relatively high construction years of some in-service pipelines, pipelines are prone to wear and tear defects, resulting in an increasing risk of operation of transportation equipment. Once an accident occurs, it will seriously affect the safety of public life and property, causing immeasurable losses and adverse effects.

[0003] Pipeline corrosion is an important cause of pipeline accidents. Pipeline corrosion is divided into external corrosion and internal corrosion. Among them, external corrosion is mainly erosion corrosion, soil corrosion, etc.; internal corrosion is mainly caused by corrosive components in the transportation medium. Pipeline corrosion is likely to cause pipeline damage, resulting in perforation and pipe burst, triggering a series of safety accidents such as medium leakage.

[0004] The ultrasonic guided wave technology is a non-destructive testing method for the detection of pressure pipelines at present. However, due to the complexity and concealment of pipelines, and the increasing scale of pipeline systems and complex operating environments, traditional ultrasonic non-destructive testing methods, such as single-point testing methods like ultrasonic testing, electromagnetic flaw detection, or penetrant flaw detection, cannot meet the requirements for timeliness, reliability, large range, economy, and safety in structural health detection. Therefore, higher requirements are put forward for the safety of in-service pipelines, and it is an urgent problem to propose reliable and effective online monitoring technologies and methods for the healthy operation state of pipelines. Summary of the Invention

[0005] In view of this, the embodiments of the present application provide a pipeline damage monitoring method and device.

[0006] In a first aspect, the embodiments of the present application provide a pipeline damage monitoring method, the method including:

[0007] Obtain the monitoring signal of the pipeline to be tested during its in-service period and a pre-determined reference Mahalanobis space;

[0008] Extract features from the monitoring signal to obtain the monitoring feature vector of the monitoring signal;

[0009] Determine the monitoring Mahalanobis distance of the monitoring signal according to the monitoring feature vector and the monitoring feature matrix constructed based on the monitoring feature vector;

[0010] Compare the monitored Mahalanobis distance with the reference Mahalanobis space to obtain the degree of deviation between the monitored Mahalanobis distance and the reference Mahalanobis space;

[0011] Determine the damage state of the pipeline to be measured according to the degree of deviation.

[0012] In some embodiments, the steps of determining the reference Mahalanobis space include:

[0013] Obtain the healthy echo signal and the damaged echo signal of the pipeline to be measured;

[0014] Perform adaptive noise complete ensemble empirical mode decomposition (CEEMD) on the healthy echo signal and the damaged echo signal to obtain the healthy characteristic parameters of the healthy echo signal and the damaged characteristic parameters of the damaged echo signal;

[0015] Construct a healthy characteristic matrix corresponding to the healthy echo signal according to the healthy characteristic parameters;

[0016] Calculate the Mahalanobis distance of the healthy echo signal under the healthy characteristic matrix, and integrate the Mahalanobis distances of the healthy echo signals to obtain the Mahalanobis space;

[0017] Based on the damaged characteristic parameters, calculate the reference Mahalanobis distance of the healthy echo signal through the Taguchi optimization method, and integrate the reference Mahalanobis distances to obtain the reference Mahalanobis space.

[0018] In some embodiments, before obtaining the healthy echo signal and the damaged echo signal of the pipeline to be measured, the method further includes:

[0019] Set monitoring points on the pipeline to be measured, where each monitoring point corresponds to a guided wave transducer for receiving echo signals;

[0020] The steps of obtaining the healthy echo signal and the damaged echo signal of the pipeline to be measured include:

[0021] Through a guided wave signal acquisition system, excite and receive a first group of echo signals at the monitoring point, and determine the first group of echo signals as the original healthy echo signal;

[0022] Simulate the damaged condition of the pipeline through a target damage object placed on the pipeline to be measured;

[0023] Under the condition that the pipeline to be measured is damaged, excite and receive a second group of echo signals at the monitoring point, and determine the second group of echo signals as the original damaged echo signal;

[0024] Establish a signal reference library according to the original healthy echo signal and the original damaged echo signal;

[0025] Perform signal preprocessing on the original healthy echo signal and the original damaged echo signal, and use wavelet denoising to filter the preprocessed original healthy echo signal and the preprocessed original damaged echo signal;

[0026] Determine the filtered original healthy echo signal as the healthy echo signal of the pipeline to be measured, and determine the filtered original damaged echo signal as the damaged echo signal of the pipeline to be measured.

[0027] In some embodiments, the number of signals in the signal reference library is 2N, where the number of the original healthy echo signals and the number of the original damaged echo signals are both N, and N is an integer greater than or equal to 1.

[0028] In some embodiments, the echo signal includes a healthy echo signal and a damaged echo signal;

[0029] Perform CEEMD decomposition on the healthy echo signal and the damaged echo signal to obtain the healthy characteristic parameters of the healthy echo signal and the damaged characteristic parameters of the damaged echo signal, including:

[0030] Perform CEEMD decomposition on the echo signal to obtain a plurality of intrinsic mode functions IMF, where the general calculation formula of the intrinsic mode function IMF is:

[0031]

[0032] where, is the (k + 1)-th intrinsic mode function obtained through CEEMD decomposition; E k (·) is the k-th signal mode function obtained through CEEMD decomposition;

[0033] Add Gaussian white noise to the echo signal x(n) to obtain the processed echo signal x (i) :

[0034] x (i) = x n + α k w (i) ;

[0035] where, w i =(i = 1, 2,..., n) is the Gaussian white noise added during the (i)-th CEEMD decomposition; Each order of empirical mode decomposition needs to be performed L times, and the α k coefficient represents the noise coefficient selected during each order of mode decomposition;

[0036] Obtain the L intrinsic mode functions IMF of the current order obtained by performing L times of CEEMD decomposition on the echo signal;

[0037] Perform a weighted average on the L intrinsic mode functions (IMFs) to obtain the intrinsic mode functions at each stage of CEEMD decomposition.

[0038]

[0039] Subtract the intrinsic mode function obtained by CEEMD decomposition from the echo signal. The result is determined as the residual signal, which is expressed as:

[0040]

[0041] Perform CEEMD decomposition on the residual signal r k until the residual signal r k cannot be decomposed any further, so that the original signal x i is decomposed into multiple intrinsic mode functions (IMFs) and a residual signal r; where the original signal x i is expressed as:

[0042]

[0043] where the residual signal r is the signal remaining after CEEMD decomposition; K is the number of times of CEEMD decomposition;

[0044] Determine the correlation coefficient between the k-th intrinsic mode function (IMF) and the echo signal to obtain a sequence of correlation coefficients between the IMFs and the echo signal.

[0045] Extract signal features from the echo signal according to the correlation coefficient to obtain the health feature parameters of the healthy echo signal and the damaged feature parameters of the damaged echo signal.

[0046] In some embodiments, determining the correlation coefficient between the k-th intrinsic mode function (IMF) and the echo signal to obtain a sequence of correlation coefficients between the IMFs and the echo signal includes:

[0047] The correlation coefficient re between the intrinsic mode function (IMF) and the processed echo signal x (i) is:

[0048]

[0049] where re k is the correlation coefficient between the k-th intrinsic mode function (IMF) and the processed echo signal x (i) ; is the mean of the echo signal with added Gaussian white noise; IMF k is the intrinsic mode function of the k-th mode decomposition; is the average value of the intrinsic mode components for the k-th multiple decomposition; is the standard deviation of the echo signal with added Gaussian white noise; σ IMF is the standard deviation of the k-th intrinsic mode component; E[] is the statistical average;

[0050] According to the correlation coefficient between each intrinsic mode component IMF and the echo signal, for the correlation coefficient re k perform screening and sorting to obtain a sequence of correlation coefficients:

[0051] re1' >> re2' >> re3' >> … >> re k '.

[0052] In some embodiments, signal feature extraction is performed on the echo signal according to the correlation coefficient to obtain the healthy feature parameters of the healthy echo signal and the damaged feature parameters of the damaged echo signal, including:

[0053] The l intrinsic mode components IMF with a correlation coefficient re k greater than a preset threshold are determined as the feature components of the processed echo signal x (i) ;

[0054] According to the feature components, signal feature extraction is performed on the processed echo signal x (i) to obtain the energy of the intrinsic mode component corresponding to each IMF, where the energy E of the intrinsic mode component i is expressed as:

[0055]

[0056] where the total signal energy E is the sum of the energies E of the l IMFs i and p i = E / E i , and the energy entropy of each IMF component is:

[0057]

[0058] The energy entropy of the IMF component is determined as the signal energy feature parameter of the echo signal; among them, the signal energy feature parameter of the echo signal includes the healthy signal energy feature parameter of the healthy echo signal and the damaged signal energy feature parameter of the damaged echo signal;

[0059] Extract parameters from the echo signal to obtain time-domain characteristic parameters and frequency-domain characteristic parameters of the echo signal; among them, the time-domain characteristic parameters of the echo signal include: the healthy time-domain characteristic parameters of the healthy echo signal and the damaged time-domain characteristic parameters of the damaged echo signal; the frequency-domain characteristic parameters of the echo signal include: the healthy frequency-domain characteristic parameters of the healthy echo signal and the damaged frequency-domain characteristic parameters of the damaged echo signal;

[0060] Determine the healthy characteristic parameters of the healthy echo signal by using the healthy signal energy characteristic parameters, the healthy time-domain characteristic parameters, and the healthy frequency-domain characteristic parameters;

[0061] Determine the damaged characteristic parameters of the damaged echo signal by using the damaged signal energy characteristic parameters, the damaged time-domain characteristic parameters, and the damaged frequency-domain characteristic parameters.

[0062] In some embodiments, construct a healthy characteristic matrix corresponding to the healthy echo signal according to the healthy characteristic parameters, including:

[0063] Construct a feature vector of the healthy echo signal according to the healthy characteristic parameters of the healthy echo signal in the signal reference library to obtain the feature vector X of the healthy echo signal base,n :

[0064]

[0065] Among them, X base,n represents the feature vector of the nth healthy echo signal; represents the first 8 characteristic parameters of the feature vector; represents the 9th to 15th characteristic parameters of the feature vector; represents the last l + 1 characteristic parameters of the feature vector;

[0066] Combine the obtained feature vectors of the healthy echo signal to construct a healthy signal characteristic matrix:

[0067]

[0068] Among them, n represents the number of healthy echo signals; p represents the pth characteristic parameter in the feature vector; X np represents the value of the pth characteristic parameter of the nth healthy echo signal, i = 1, 2, …n; j = 1, 2, …p.

[0069] In some embodiments, calculate the Mahalanobis distance of the healthy echo signal under the healthy characteristic matrix, and integrate the Mahalanobis distances of the healthy echo signals to obtain a Mahalanobis space, including:

[0070] Perform standardization processing on the pre-acquired reference space according to the healthy signal characteristic matrix to obtain a standardized reference space:

[0071]

[0072] wherein, is the mean value of the j-th signal feature parameter in the feature matrix; δ j is the standard deviation of the j-th signal feature parameter of the feature matrix, i = 1, 2, 3, … n;

[0073] Based on the standardized reference space, a standardized reference space is established:

[0074]

[0075] Based on the standardized reference space, calculate the Mahalanobis distance d of the healthy echo signal under the standardized reference space M,normal :

[0076]

[0077] wherein, S represents the correlation coefficient matrix of the standardized healthy signal feature matrix;

[0078] Using the eigenvector of the damaged echo signal, calculate the Mahalanobis distance of the damaged echo signal under the healthy signal feature matrix constructed by the healthy echo signal;

[0079] Statistically analyze the Mahalanobis distance corresponding to each healthy echo signal to obtain a statistical result;

[0080] Based on the statistical result, determine the numerical range of the Mahalanobis distance corresponding to the healthy echo signal as the Mahalanobis space of the healthy echo signal.

[0081] In some embodiments, the steps of determining the reference Mahalanobis space include:

[0082] Using the Taguchi optimization method, establish a two-level orthogonal array L n (2 p );

[0083] Through the signal-to-noise ratio gain of the feature variables in the eigenvector of the healthy echo signal and the two-level orthogonal array L n (2 p ), screen the sensitive feature parameters of the healthy feature parameters in the eigenvector to obtain the screened healthy feature parameters; wherein, the signal-to-noise ratio gain △ is expressed as:

[0084] △ = SNR sel - SNR unsel ;

[0085] wherein, SNR sel represents the signal-to-noise ratio using the feature variable; SNR unselDenote the signal-to-noise ratio of the unused feature variable; when Δ is greater than 0, select the feature variable; when Δ is less than or equal to 0, remove the feature variable;

[0086] Construct the final feature vector X of the healthy echo signal according to the filtered healthy feature parameters final :

[0087] X final =[x1, x2, x3, … x q ;

[0088] According to the final feature vector X final , calculate the healthy Mahalanobis distance of the healthy echo signal;

[0089] Statistically analyze the healthy Mahalanobis distance of each healthy echo signal to obtain the statistical result;

[0090] According to the statistical result, determine the numerical range of the healthy Mahalanobis distance as the reference Mahalanobis space.

[0091] In some embodiments, when the monitored Mahalanobis distance of the monitoring signal of the pipeline to be measured is greater than the Mahalanobis distance in the reference Mahalanobis space, and the deviation degree between the monitored Mahalanobis distance and the Mahalanobis distance in the reference Mahalanobis space is greater than the deviation threshold, it is determined that the pipeline to be measured is in a damaged state.

[0092] In a second aspect, an embodiment of the present application provides a pipeline damage monitoring device, and the device includes:

[0093] An acquisition module, configured to acquire the monitoring signal of the pipeline to be measured during operation and the pre-determined reference Mahalanobis space;

[0094] A feature extraction module, configured to extract features from the monitoring signal to obtain the monitoring feature vector of the monitoring signal;

[0095] A first determination module, configured to determine the monitored Mahalanobis distance of the monitoring signal according to the monitoring feature vector and the monitoring feature matrix constructed based on the monitoring feature vector;

[0096] A comparison module, configured to compare the monitored Mahalanobis distance with the reference Mahalanobis space to obtain the deviation degree between the monitored Mahalanobis distance and the reference Mahalanobis space;

[0097] A second determination module, configured to determine the damage state of the pipeline to be measured according to the deviation degree.

[0098] The pipeline damage monitoring method and device provided by the embodiments of the present application determine the Mahalanobis distance of the monitoring signal when the pipeline to be measured is in service, and compare the Mahalanobis distance with the pre-determined reference Mahalanobis space to determine the deviation degree between the monitoring Mahalanobis distance and the reference Mahalanobis space, so as to further determine the damage state of the pipeline to be measured. In this way, the embodiments of the present application use the feature matrix to calculate the Mahalanobis distance. The Mahalanobis distance takes into account the correlation between variables, is not affected by the dimension of the eigenvector, and the embodiments of the present application perform multi-feature information parameter fusion to judge the health state of the pipeline to be measured, mine the information of the data collected from the pipeline to be measured for a long time, improve the guided wave monitoring accuracy and reliability, and have practical significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0099] In the drawings (which are not necessarily drawn to scale), like reference numerals may describe like components in different views. Like reference numerals with different letter suffixes may represent different examples of like components. The drawings generally illustrate, by way of example and not limitation, the various embodiments discussed herein.

[0100] Figure 1 is a schematic flow chart of the pipeline damage monitoring method provided by the embodiments of the present application;

[0101] Figure 2 is a schematic flow chart of the pipeline damage monitoring method provided by the embodiments of the present application;

[0102] Figure 3 is a schematic structural diagram of the pipeline damage monitoring device provided by the embodiments of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0103] The following will describe the exemplary embodiments of the present application in more detail with reference to the drawings. Although the exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the specific embodiments set forth herein. On the contrary, these embodiments are provided so that the present application can be more thoroughly understood and the scope of the present application can be fully communicated to those skilled in the art.

[0104] In the following description, numerous specific details are given to provide a more thorough understanding of the present application. However, it will be apparent to one of ordinary skill in the art that the present application may be practiced without one or more of these details. In other instances, some well-known technical features are not described in order to avoid obscuring the present application; that is, not all features of the actual embodiments are described here, and the well-known functions and structures are not described in detail.

[0105] Based on the problems and deficiencies in the related art, the embodiments of the present application provide a pipeline damage monitoring method. By determining the Mahalanobis distance of the monitoring signal when the pipeline to be measured is in service and comparing the Mahalanobis distance with the pre-determined reference Mahalanobis space, the deviation degree between the monitoring Mahalanobis distance and the reference Mahalanobis space is determined to determine the damage state of the pipeline to be measured, which not only enriches the monitoring methods for pipeline structure damage, but also can realize the online automatic monitoring of the pipeline operation health status.

[0106] Figure 1 It is a schematic flow chart of the pipeline damage monitoring method provided by the embodiments of the present application. As Figure 1 shown, the embodiments of the present application realize pipeline damage monitoring through the following steps:

[0107] Step S101: Obtain the monitoring signal when the pipeline to be measured is in service and the pre-determined reference Mahalanobis space.

[0108] In some embodiments, obtaining the monitoring signal when the pipeline to be measured is in service may be to set monitoring points on the pipeline to be measured that needs to be monitored, and install a guided wave transducer at the monitoring point. The guided wave transducer is used to receive the echo signal, and the guided wave signal acquisition system is used to excite and receive the monitoring signal when the pipeline to be measured is in service at the monitoring point.

[0109] In some embodiments, the monitoring signal when the pipeline to be measured is in service may be an ultrasonic monitoring signal obtained by periodically detecting the in-service pipeline to be measured using a guided wave transducer. The pipeline to be measured being in service means that the pipeline to be measured is a pipeline in use. For example, some natural gas pipelines in use.

[0110] In some embodiments, a guided wave is an elastic wave whose propagation is affected by the constraints of the waveguide geometry. The guided wave can propagate along the waveguide pipeline for a relatively long distance, and a single detection point can achieve full coverage of the detected structure and large-scale online monitoring and detection, which can meet the requirements of rapid, efficient, monitoring and detection of pipeline damage assessment. The guided wave technology is also applied in the non-destructive testing and online monitoring of various industries and fields due to its characteristics such as long detection distance, large detection range and full coverage detection.

[0111] In some embodiments, obtaining the pre-determined reference Mahalanobis space is completed through the following steps:

[0112] The first set of echo signals s1'(n) is excited and received at the monitoring points of the pipeline under test by a guided wave signal acquisition system; a bonded iron block (i.e., the target damage object) is placed on the pipeline under test to simulate the damaged condition of the pipeline, and the second set of echo signals s2'(n) is excited and received at the monitoring points. The echo signals are classified according to the pipeline condition. The first set of echo signals s1'(n) is the original healthy echo signal x'1(n), and the second set of echo signals s2'(n) is the original damaged echo signal x'2(n). The original healthy echo signal x'1(n) and the original damaged echo signal x'2(n) jointly establish a signal reference library.

[0113] The ultrasonic guided wave monitoring device transducer (i.e., the guided wave transducer) receives the echo signal s(n) (including the first set of echo signals s1'(n) and the second set of echo signals s2'(n)). Among them, n represents the ordinal number of signal acquisition. The echo signal s(n) is divided into the first set of echo signals s2'(n) (i.e., the original healthy echo signal x'1(n)) when the pipeline is in a healthy and undamaged state and the second set of echo signals s2'(n) (i.e., the original damaged echo signal x'2(n)) when the pipeline is in a simulated damaged state according to the pipeline condition.

[0114] In some embodiments, the number of signals in the signal reference library is 2N, and the numbers of the original healthy echo signal x'1(n) and the original damaged echo signal x'2(n) are N respectively.

[0115] The echo signals collected by the guided wave signal acquisition system are affected by multiple factors. For example, other noise sources in the field environment and pipeline boundary conditions and other factors will interfere with the recognition of the monitoring signals. Therefore, before signal feature extraction, the collected signals need to be preprocessed.

[0116] In some embodiments, the original healthy echo signal x'1(n) and the original damaged echo signal x'2(n) (i.e., the echo signal x'(n)) are preprocessed to eliminate the DC component of the signal and normalize the amplitude. The echo signal x(n) after signal preprocessing is shown in formula (1):

[0117]

[0118] where M represents the signal length of the echo signal x'(n); x'(n) max represents the maximum value of the signal amplitude in the echo signal x'(n); n represents the time series points of the echo signal x'(n).

[0119] The preprocessing of the echo signal x'(n) in the embodiments of the present application can reduce the interference of the hardware and the external environment on the collected signals and improve the discrimination accuracy of the pipeline damage monitoring method.

[0120] In some embodiments, after preprocessing the echo signal, the wavelet denoising method can be used to remove the clutter in the preprocessed echo signal x(n), and the filtered healthy echo signal x1(n) and the filtered damaged echo signal x2(n) can be obtained respectively.

[0121] In some embodiments, after obtaining the filtered healthy echo signal x1(n) and the filtered damaged echo signal x2(n), the time-domain and frequency-domain characteristic parameters in the filtered healthy echo signal x1(n) are extracted item by item. The filtered healthy echo signal x1(n) is subjected to CEEMD decomposition. From the intrinsic mode components (IMFs) of the healthy echo signal obtained from the decomposition, the signal characteristic parameters of the healthy echo signal are further extracted; the time-domain and frequency-domain characteristic parameters in the filtered damaged echo signal x2(n) are extracted item by item, and the filtered damaged echo signal x2(n) is subjected to CEEMD decomposition. From the IMFs of the damaged echo signal obtained from the decomposition, the signal characteristic parameters of the damaged echo signal are further extracted.

[0122] In some embodiments, CEEMD decomposition refers to Complete Ensemble Empirical Mode Decomposition (CEEMD). The CEEMD decomposition algorithm can reduce the computational amount and further eliminate mode mixing and false components in the echo signal.

[0123] The CEEMD decomposition of the echo signal can be achieved through the following steps:

[0124] In some embodiments, the general formula (2) for calculating the intrinsic mode component IMF is as follows:

[0125]

[0126] Where, is the (k + 1)th intrinsic mode component obtained through CEEMD decomposition; E k (·) is the kth signal mode component obtained through CEEMD decomposition.

[0127] In some embodiments, Gaussian white noise is added to the healthy echo signal and the damaged echo signal to obtain the processed echo signal x (i) , x (i) = x n + α0w (i) , where, w i =(i = 1, 2,..., n) is the Gaussian white noise added during the i-th CEEMD decomposition.

[0128] In the embodiments of the present application, each stage of empirical mode decomposition needs to be performed L times, and the α k coefficient represents the noise coefficient selected during each stage of mode decomposition.

[0129] The embodiments of the present application set an ideal noise coefficient α k to improve the clarity of empirical mode decomposition.

[0130] In the embodiments of the present application, the processed echo signal x (i) is subjected to empirical mode decomposition a total of L times, and the L intrinsic mode functions IMF obtained by the decomposition are weighted and averaged to obtain the intrinsic mode function of this decomposition stage as shown in formula (3):

[0131]

[0132] In some embodiments, the echo signal is subtracted from the intrinsic mode function obtained by CEEMD decomposition to obtain a residual signal

[0133] The residual signal r k is subjected to CEEMD decomposition until the residual signal r k cannot be decomposed, so that the processed echo signal is decomposed into multiple intrinsic mode functions IMF and a residual signal r, where the residual signal r is the residual signal after CEEMD decomposition, and the original signal x i is expressed as shown in formula (4):

[0134]

[0135] In some embodiments, the correlation strengths of the characteristic information carried by each IMF component are different, and the intrinsic mode functions IMF can be sorted by the correlation coefficient re, where the intrinsic mode function IMF and the processed echo signal x (i) The correlation coefficient re between them is shown in formula (5):

[0136]

[0137] where re k is the correlation coefficient between the k-th intrinsic mode function IMF and the processed echo signal x (i) ; is the mean value of the echo signal with added Gaussian white noise; IMF k is the intrinsic mode function of the k-th mode decomposition; is the average value of the intrinsic mode functions obtained by multiple decompositions for the k-th time; [[ID=5�]] is the standard deviation of the echo signal with added Gaussian white noise; σ IMFis the standard deviation of the k-th intrinsic mode component; E[] is the statistical average value.

[0138] According to the correlation coefficient between each intrinsic mode component IMF and the echo signal, the IMF correlation coefficient re k is sorted by screening to obtain a correlation coefficient sequence: re1' >> re2' >> re3' >> … >> re k ', where the IMF component corresponding to each correlation coefficient is

[0139] The l intrinsic mode components IMF with the correlation coefficient re k greater than the preset threshold are determined as the characteristic components of the processed echo signal x (i) . Based on the characteristic components, signal feature extraction is performed on the processed echo signal x (i) to obtain the energy of the intrinsic mode component corresponding to each IMF. Among them, the energy E of the intrinsic mode component i is as shown in formula (6):

[0140]

[0141] where the preset threshold can be 0.1; the IMF component corresponding to the correlation coefficient of the echo signal is The signal energy E within the preset time period is the sum of the energies E of l IMFs i and p i = E / E i . The energy entropy of each IMF component is as shown in formula (7):

[0142]

[0143] In the embodiment of the present application, the energy entropy of the IMF component is determined as the signal energy characteristic parameter of the echo signal. The echo signal includes a healthy echo signal and a damaged echo signal. Repeat the above steps to perform CEEMD decomposition, selection of intrinsic mode IMF components, and frequency domain feature extraction on the damaged echo signals in the signal reference library to obtain the signal energy characteristic parameters of the damaged echo signals. The signal energy characteristic parameter is the energy entropy parameter extracted from the first l high-correlation intrinsic mode components IMF.

[0144] Time domain characteristic parameters and frequency domain characteristic parameters are extracted from the filtered healthy echo signal x1(n) in the signal reference library. Among them, the time domain characteristic parameters include the mean square error standard deviation energy fraction kurtosis skewness peak factor waveform factor and pulse factor The frequency-domain characteristic parameters include the center frequency peak frequency root mean square frequency frequency skewness frequency standard deviation frequency kurtosis and root mean square of frequency Extract the time-domain characteristic parameters and frequency-domain characteristic parameters from the filtered damaged echo signal x2(n). The signal energy characteristic parameter is also the energy entropy parameter extracted from the first l high-correlation intrinsic mode components IMF.

[0145] In some embodiments, the dimensional signal characteristic parameters in the echo signal can be made dimensionless, as shown in formula (8):

[0146] x n = log 10 (x n ) (8)

[0147] where x n is the dimensional signal characteristic parameter.

[0148] In some embodiments, the characteristic parameters of the healthy echo signal can be combined to form the healthy characteristic signal feature vector X base , construct a feature matrix, calculate the Mahalanobis distance of each signal feature vector under this feature matrix, and integrate the Mahalanobis distance to obtain the Mahalanobis space.

[0149] In the embodiments of the present application, the steps to obtain the Mahalanobis space include:

[0150] Extract signal parameters and construct signal vectors for the healthy echo signals in the signal reference library to obtain the feature vector X of the nth healthy echo signal base,n , as shown in formula (9):

[0151]

[0152] where X base,n represents the feature vector of the nth healthy echo signal; represents the first 8 characteristic parameters of the feature vector; represents the 9th to 15th characteristic parameters of the feature vector; represents the last l characteristic parameters of the feature vector; l represents the energy entropy obtained by extracting L high-correlation intrinsic mode components.

[0153] Combine the feature vectors of the obtained n healthy echo signals to construct a healthy signal feature matrix X, as shown in formula (10):

[0154]

[0155] Among them, n represents the number of healthy echo signals, p represents the p-th characteristic parameter in the eigenvector, and X np represents the p-th characteristic parameter value of the n-th healthy echo signal, where i = 1, 2, …, n; j = 1, 2, …, p.

[0156] According to the healthy signal feature matrix X, perform standardization processing on the pre-acquired reference space to obtain the standardized reference space, as shown in formula (11):

[0157]

[0158] Among them, is the mean value of the j-th signal characteristic parameter in the feature matrix; δ j is the standard deviation of the j-th signal characteristic parameter of the feature matrix, where i = 1, 2, 3, …, n.

[0159] According to the standardized reference space, establish the standardized reference space to obtain the Mahalanobis distance d of the healthy echo signal in the standardized reference space M,normal , as shown in formula (12):

[0160]

[0161] Among them, S represents the correlation coefficient matrix of the standardized healthy signal feature matrix, that is, the covariance matrix.

[0162] Statistically analyze the Mahalanobis distance corresponding to each healthy echo signal, and determine the numerical range involved in the Mahalanobis distance as the Mahalanobis space of the healthy echo signal.

[0163] Use the signal feature vector extracted from the damaged echo signal x'2(n) to calculate the Mahalanobis distance of the damaged echo signal under the feature matrix constructed by the healthy echo signals.

[0164] In the embodiment of the present application, the Mahalanobis distance of the damaged echo signal x'2(n) will be significantly greater than the Mahalanobis distance obtained by calculating the healthy echo signal.

[0165] In some embodiments, the Taguchi optimization method can be used to establish an orthogonal table, and through the signal-to-noise ratio gain of the characteristic variables in the eigenvector, screen the sensitive characteristic parameters of the eigenvector, filter out some of the characteristic parameters that are insensitive to pipeline damage in the healthy characteristic parameters, select the screened sensitive signal characteristic parameters of each healthy echo signal to reconstitute the signal feature vector, and re-construct the reference feature matrix X to calculate the reference Mahalanobis distance d M and the reference Mahalanobis space.

[0166] In the embodiments of the present application, the reference Mahalanobis space can be formed through the following steps:

[0167] In some embodiments, the eigenvector of the healthy echo signal is a p-dimensional vector. By selecting the two-level orthogonal array L n (2 p ) to minimize the characteristic variables of the characteristic parameters, sensitive eigenvector recognition is performed; wherein, in the two-level orthogonal array L n (2 p ), "1" represents selecting the eigenvector, and "2" represents not selecting the eigenvector.

[0168] In some embodiments, during calculation, the signal eigenvectors using the characteristic variables and not using the characteristic variables are respectively formed, and the characteristic matrices under using the characteristic variables and not using the characteristic variables are constructed. The Mahalanobis distances of the damaged echo signals under using the characteristic variables and not using the characteristic variables are calculated, and the signal-to-noise ratio difference is used to reflect the sensitivity of using the selected characteristic variables to damage recognition, as shown in formula (13):

[0169]

[0170] Wherein, m is the number of damaged echo signals, and d M，k is the Mahalanobis distance of the k-th damaged echo signal.

[0171] The recognition effect of the signal with the selected characteristic variables is judged by the signal-to-noise ratio gain △, wherein the signal-to-noise ratio gain △ is expressed as formula (14):

[0172] △ = SNR sel -SNR unsel (14)

[0173] Wherein, SNR sel represents the signal-to-noise ratio using the selected variable; SNR unsel represents the signal-to-noise ratio without using the selected variable; when Δ is greater than 0, the characteristic variable is selected; when Δ is less than or equal to 0, the characteristic variable is removed.

[0174] Through the signal-to-noise ratio gain of the characteristic variables in the eigenvector of the healthy echo signal, the sensitive characteristic parameters in the healthy characteristic parameters of the eigenvector are screened to obtain the screened healthy characteristic parameters. The eigenvector X final = [x1, x2, x3,... x q is reconstructed by the screened healthy characteristic parameters, and the healthy Mahalanobis distance of the healthy echo signal is recalculated, and the healthy Mahalanobis distance is statistically analyzed to obtain the statistical result; according to the statistical result, the numerical range of the healthy Mahalanobis distance is determined as the reference Mahalanobis space.

[0175] Step S102: Extract the features of the monitoring signal to obtain the monitoring feature vector of the monitoring signal.

[0176] In the embodiment of the present application, the feature parameters of the monitoring signal are extracted by step S101 to form the signal feature vector X collect,1 .

[0177] Step S103: Determine the monitoring Mahalanobis distance of the monitoring signal according to the monitoring feature vector and the monitoring feature matrix constructed based on the monitoring feature vector.

[0178] Step S104: Compare the monitoring Mahalanobis distance with the reference Mahalanobis space to obtain the deviation degree between the monitoring Mahalanobis distance and the reference Mahalanobis space.

[0179] Step S105: Determine the damage state of the pipeline to be measured according to the deviation degree.

[0180] According to the signal feature vector X collect,1 , calculate the Mahalanobis distance d of the monitoring signal (i.e., the monitoring signal) monitoring , compare the Mahalanobis distance d of the monitoring signal monitoring with the deviation degree between the reference Mahalanobis space, and judge the running health state of the pipeline to be measured by monitoring the deviation degree of the Mahalanobis distance of the monitoring signal from the reference Mahalanobis space.

[0181] In some embodiments, the monitoring signal can be continuously collected at the monitoring points set on the pipeline to be measured, the feature parameters of the m-th monitoring signal are extracted, and the Mahalanobis distance d of the monitoring signal is calculated monitoring , to judge the running health state of the pipeline to be measured.

[0182] In some embodiments, when the monitoring Mahalanobis distance of the monitoring signal of the pipeline to be measured is greater than the Mahalanobis distance in the reference Mahalanobis space, and the deviation degree between the monitoring Mahalanobis distance and the Mahalanobis distance in the reference Mahalanobis space is greater than the deviation threshold, it is determined that the pipeline to be measured is in a damaged state.

[0183] In the embodiment of the present application, the magnitude of the Mahalanobis distance of the healthy echo signal approaches 1. When the pipeline to be measured is damaged and has defects, the Mahalanobis distance of the monitoring signal of the pipeline to be measured is greater than the Mahalanobis distance in the reference Mahalanobis space; among them, the more serious the damage degree of the pipeline to be measured, the greater the Mahalanobis distance of the monitoring signal.

[0184] In some embodiments, pipeline guided wave monitoring signals are mostly non - linear and non - stationary signals. It is difficult for time - domain characteristic parameters and frequency - domain characteristic parameters to comprehensively characterize pipeline damage information. In the embodiments of the present application, in addition to extracting time - domain characteristic parameters and frequency - domain characteristic parameters, the signal frequency energy is also analyzed, and signal energy characteristic parameters are supplemented to comprehensively judge the pipeline health status, improving the signal identification accuracy.

[0185] Compared with the traditional EMD empirical mode decomposition, in the embodiments of the present application, the complete ensemble empirical mode decomposition with adaptive white noise (CEEMD) is used to decompose and reconstruct the signal. White noise is added during the signal decomposition process, which can achieve the accurate reconstruction of the original signal, ensure the decomposition accuracy, and overcome the problem of EMD mode mixing, and is applicable to the analysis of pipeline monitoring signals.

[0186] In the embodiments of the present application, the Mahalanobis distance is calculated using the covariance matrix to construct a Mahalanobis space. The Mahalanobis distance takes into account the correlation between variables and is not affected by the dimension of the eigenvector. The Taguchi method is used to select and optimize sensitive characteristic parameters, greatly reducing the number of characteristic variables, achieving the purpose of dimensionality reduction and reducing the amount of calculation. And when optimizing the characteristic parameters, no additional new parameters need to be set. Only by continuously measuring the data can the signal - to - noise ratio gain of the characteristic variables be obtained for characteristic optimization.

[0187] Combining the advantages of guided wave technology, the embodiments of the present application use the Mahalanobis - Taguchi method to fuse multiple characteristic information parameters to judge the pipeline health status, mine the information of the data collected from the long - term detection of the pipeline, improve the guided wave monitoring accuracy and reliability, and have practical significance.

[0188] The embodiments of the present application further provide a pipeline damage monitoring method, as Figure 2 shown, Figure 2 is a schematic flow chart of the pipeline damage monitoring method provided by the embodiments of the present application. The pipeline damage monitoring method provided by the embodiments of the present application is implemented through the following steps:

[0189] Step S201: Filter the echo signal to obtain a filtered signal.

[0190] In some embodiments, a monitoring point is set on the pipeline to be measured, and the monitoring point corresponds to a guided wave transducer for receiving echo signals. Through the guided wave signal acquisition system, the first group of echo signals is excited and received at the monitoring point, and the first group of echo signals is determined as the original healthy echo signal; by placing a bonded iron block (i.e., the target damage object) on the pipeline to be measured to simulate the damaged situation of the pipeline, the second group of echo signals is excited and received at the monitoring point when the pipeline to be measured is damaged, and the second group of echo signals is determined as the original damaged echo signal. The original healthy echo signal and the original damaged echo signal are determined as the echo signals.

[0191] Filtering the echo signal means performing signal preprocessing on the echo signal and filtering the preprocessed echo signal using wavelet denoising to obtain a filtered signal.

[0192] Step S202: Extract the time-domain characteristic parameters and frequency-domain characteristic parameters of the filtered signal.

[0193] In some embodiments, the time-domain characteristic parameters include mean square error standard deviation energy fraction kurtosis skewness kurtosis factor waveform factor and pulse factor The frequency-domain characteristic parameters include center frequency peak frequency root mean square frequency frequency skewness frequency standard deviation frequency kurtosis and frequency root mean square

[0194] Step S203: Perform CEEMD decomposition on the filtered signal to obtain IMF intrinsic mode components.

[0195] Step S204: Extract the signal energy characteristic parameters of the filtered signal according to the IMF intrinsic mode components.

[0196] In some embodiments, perform CEEMD decomposition on the filtered signal, and from the obtained intrinsic mode components IMF, screen the intrinsic mode components IMF using the correlation coefficient to obtain the first l intrinsic mode components IMF with a correlation coefficient greater than 0.1. According to the first l intrinsic mode components IMF, further extract the signal energy characteristic parameter energy entropy of the echo signal, and determine the energy entropy of the intrinsic mode components IMF as the signal energy characteristic parameter of the filtered signal.

[0197] Step S205: Obtain the feature vector of the echo signal according to the signal energy characteristic parameter, time-domain characteristic parameter, and frequency-domain characteristic parameter.

[0198] Step S206: Construct a Mahalanobis space according to the feature vector.

[0199] According to the signal energy characteristic parameter, time-domain characteristic parameter, and frequency-domain characteristic parameter, respectively construct the feature vectors of the healthy echo signal and the damaged echo signal, and use the healthy echo signal to establish a feature matrix.

[0200] Perform standardization processing on the reference space to obtain the standardized reference space, as shown in formula (15):

[0201]

[0202] Based on the standardized reference space, establish a standardized reference space Calculate the Mahalanobis distance d of the healthy echo signal in the standardized reference space M,normal , as shown in formula (16):

[0203]

[0204] Integrate the Mahalanobis distance of the healthy echo signal to construct a Mahalanobis space

[0205] Step S207: Establish an orthogonal table for the characteristic variables in the eigenvector

[0206] Step S208: Calculate the signal-to-noise ratio of the characteristic variables

[0207] Step S209: Construct a reference Mahalanobis space according to the signal-to-noise ratio of the characteristic variables

[0208] Adopt the Taguchi method to screen sensitive characteristic variables through the signal-to-noise ratio gain of the characteristic variables, screen out some characteristic parameters that are insensitive to pipeline damage in the healthy characteristic parameters, reconstitute the signal characteristic vector by selecting the sensitive signal characteristic parameters from each healthy echo signal, and reconstruct the reference characteristic matrix, and calculate the reference Mahalanobis distance d M and the reference Mahalanobis space

[0209] Each time of calculation, use the selected characteristic parameters to construct a characteristic matrix, respectively form the signal characteristic vectors with and without using this characteristic variable each time of calculation, and construct the characteristic matrices in their respective cases, calculate the Mahalanobis distances of the damaged echo signals with and without using this characteristic variable, and use the signal-to-noise ratio gain to reflect the sensitivity of using the selected variable to damage identification, as shown in formula (17):

[0210]

[0211] Judge the recognition effect of the signal of the selected characteristic variable through the gain △ of the signal-to-noise ratio

[0212] Through the signal-to-noise ratio gain of the characteristic variables in the eigenvector of the healthy echo signal, screen the sensitive characteristic parameters from the healthy characteristic parameters in the eigenvector to obtain the screened healthy characteristic parameters. Reconstruct the eigenvector X final = [x1, x2, x3,... x q , recalculate the healthy Mahalanobis distance of the healthy echo signal, statistically integrate the healthy Mahalanobis distance, and establish a reference Mahalanobis space

[0213] Step S210: Monitor the monitoring signal of the pipeline to be measured.

[0214] In some embodiments, monitoring the monitoring signal of the pipeline to be measured may be to excite and receive the monitoring signal of the pipeline to be measured during service through a guided wave signal acquisition system at the monitoring point.

[0215] Step S211: Calculate the Mahalanobis distance of the monitoring signal.

[0216] In the embodiments of the present application, the Mahalanobis distance of the monitoring signal is calculated through the aforementioned steps of calculating the Mahalanobis distance.

[0217] Step S212: Judge the operation condition of the pipeline according to the Mahalanobis distance of the monitoring signal and the reference Mahalanobis space.

[0218] In some embodiments, when the monitoring Mahalanobis distance of the monitoring signal of the pipeline to be measured is greater than the Mahalanobis distance in the reference Mahalanobis space, and the deviation degree between the monitoring Mahalanobis distance and the Mahalanobis distance in the reference Mahalanobis space is greater than the deviation threshold, it is determined that the pipeline to be measured is in a damaged state.

[0219] In the embodiments of the present application, the magnitude of the Mahalanobis distance of the healthy echo signal approaches 1. When the pipeline to be measured is damaged and has defects, the Mahalanobis distance of the monitoring signal of the pipeline to be measured is greater than the Mahalanobis distance in the reference Mahalanobis space; wherein, the more serious the damage degree of the pipeline to be measured, the greater the Mahalanobis distance of the monitoring signal.

[0220] The embodiments of the present application provide a pipeline damage monitoring device, which can execute the pipeline damage monitoring method provided in any of the above embodiments.

[0221] Figure 3 It is a schematic structural diagram of the pipeline damage monitoring device provided by the embodiments of the present application. As Figure 3 shown, the pipeline damage monitoring device may include an acquisition module 301, a feature extraction module 302, a first determination module 303, a comparison module 304, and a second determination module 305, wherein:

[0222] The acquisition module 301 is configured to acquire the monitoring signal of the pipeline to be measured during service and the pre-determined reference Mahalanobis space;

[0223] The feature extraction module 302 is configured to extract features from the monitoring signal to obtain the monitoring feature vector of the monitoring signal;

[0224] The first determination module 303 is configured to determine the monitoring Mahalanobis distance of the monitoring signal according to the monitoring feature vector and the monitoring feature matrix constructed based on the monitoring feature vector;

[0225] A comparison module 304, configured to compare the monitored Mahalanobis distance with the reference Mahalanobis space to obtain the deviation degree between the monitored Mahalanobis distance and the reference Mahalanobis space;

[0226] A second determination module 305, configured to determine the damage state of the pipeline to be measured according to the deviation degree.

[0227] In some embodiments, the pipeline damage monitoring device further includes: a first acquisition module, configured to acquire a healthy echo signal and a damaged echo signal of the pipeline to be measured; a decomposition module, configured to perform an adaptive noise complete ensemble empirical mode CEEMD decomposition on the healthy echo signal and the damaged echo signal to obtain healthy characteristic parameters of the healthy echo signal and damaged characteristic parameters of the damaged echo signal; a construction module, configured to construct a healthy characteristic matrix corresponding to the healthy echo signal according to the healthy characteristic parameters; a calculation module, configured to calculate the Mahalanobis distance of the healthy echo signal under the healthy characteristic matrix and integrate the Mahalanobis distance of the healthy echo signal to obtain a Mahalanobis space; an integration module, configured to calculate the reference Mahalanobis distance of the healthy echo signal by using a Taguchi optimization method based on the damaged characteristic parameters and integrate the reference Mahalanobis distance to obtain the reference Mahalanobis space.

[0228] In some embodiments, the pipeline damage monitoring device further includes: a setting module, configured to set a monitoring point on the pipeline to be measured, where the monitoring point corresponds to a guided wave transducer for receiving an echo signal;

[0229] The first acquisition module is further configured to, through a guided wave signal acquisition system, excite and receive a first set of echo signals at the monitoring point, and determine the first set of echo signals as the original healthy echo signal; simulate the pipeline damage condition through a target damage object placed on the pipeline to be measured; in the case where the pipeline to be measured is damaged, excite and receive a second set of echo signals at the monitoring point, and determine the second set of echo signals as the original damaged echo signal; establish a signal reference library according to the original healthy echo signal and the original damaged echo signal; perform signal preprocessing on the original healthy echo signal and the original damaged echo signal, and filter the preprocessed original healthy echo signal and the preprocessed original damaged echo signal by using a wavelet denoising method; determine the filtered original healthy echo signal as the healthy echo signal of the pipeline to be measured, and determine the filtered original damaged echo signal as the damaged echo signal of the pipeline to be measured.

[0230] In some embodiments, the echo signal includes a healthy echo signal and a damaged echo signal; the decomposition module is further configured to perform CEEMD decomposition on the healthy echo signal and the damaged echo signal to obtain healthy characteristic parameters of the healthy echo signal and damaged characteristic parameters of the damaged echo signal, including:

[0231] Perform CEEMD decomposition on the echo signal to obtain multiple intrinsic mode functions (IMFs). Among them, the general calculation formula for the intrinsic mode function IMF is:

[0232]

[0233] Where is the (k + 1)-th intrinsic mode function obtained through CEEMD decomposition; E k (·) is the k-th signal mode function obtained through CEEMD decomposition;

[0234] Add Gaussian white noise to the echo signal x(n) to obtain the processed echo signal x (i) :

[0235] x (i) = x n + α k w (i) ;

[0236] Where w i =(i = 1, 2,..., n) is the Gaussian white noise added during the i-th CEEMD decomposition; Each order of empirical mode decomposition needs to be performed L times, and the α k coefficient represents the noise coefficient selected during each order of mode decomposition;

[0237] Obtain the L intrinsic mode functions IMF of the current order obtained by performing L times of CEEMD decomposition on the echo signal;

[0238] Perform weighted averaging on the L intrinsic mode functions IMF to obtain the intrinsic mode function of each CEEMD decomposition stage

[0239]

[0240] Determine the result of subtracting the intrinsic mode function obtained by CEEMD decomposition from the echo signal as the residual signal, and the residual signal is expressed as:

[0241]

[0242] Perform CEEMD decomposition on the residual signal r k until the residual signal r k cannot be decomposed any further, so that the original signal x i is decomposed into multiple intrinsic mode functions IMF and the residual signal r; Among them, the original signal x i is expressed as:

[0243]

[0244] Among them, the residual signal r is the signal remaining after CEEMD decomposition; K is the number of times of CEEMD decomposition;

[0245] Determine the correlation coefficient between the k-th intrinsic mode function IMF and the echo signal, and obtain the correlation coefficient sequence of the intrinsic mode function IMF and the echo signal;

[0246] Extract the signal features of the echo signal according to the correlation coefficient, and obtain the health feature parameters of the healthy echo signal and the damaged feature parameters of the damaged echo signal.

[0247] In some embodiments, the decomposition module is further configured to calculate the correlation coefficient re between the intrinsic mode function IMF and the processed echo signal x (i) as follows:

[0248]

[0249] where re k is the correlation coefficient between the k-th intrinsic mode function IMF and the processed echo signal x (i) ; is the mean value of the echo signal with added Gaussian white noise; IMF k is the intrinsic mode function of the k-th mode decomposition; is the average value of the intrinsic mode functions of the k-th multiple decompositions; is the standard deviation of the echo signal with added Gaussian white noise; σ IMF is the standard deviation of the k-th intrinsic mode function; E[] is the statistical average value;

[0250] According to the correlation coefficient between each intrinsic mode function IMF and the echo signal, screen and sort the correlation coefficient re k to obtain the correlation coefficient sequence:

[0251] re1' >> re2' >> re3' >> … >> re k '.

[0252] In some embodiments, the decomposition module is further configured to determine the l intrinsic mode functions IMF with the correlation coefficient re k greater than the preset threshold as the characteristic components of the processed echo signal x (i) ;

[0253] Extract the signal features of the processed echo signal x (i) according to the characteristic components, and obtain the energy of the intrinsic mode function corresponding to each IMF, where the energy E of the intrinsic mode function i is expressed as:

[0254]

[0255] Among them, the total signal energy E is the sum of the energies E of l IMFs, that is, the total energy of the received signal, p i = E / E i ,and the energy entropy of each IMF component i is as follows: :

[0256]

[0257] Determine the energy entropy of the IMF component as the signal energy characteristic parameter of the echo signal; among them, the signal energy characteristic parameter of the echo signal includes the healthy signal energy characteristic parameter of the healthy echo signal and the damaged signal energy characteristic parameter of the damaged echo signal;

[0258] Extract parameters from the echo signal to obtain the time-domain characteristic parameters and frequency-domain characteristic parameters of the echo signal; among them, the time-domain characteristic parameters of the echo signal include: the healthy time-domain characteristic parameters of the healthy echo signal and the damaged time-domain characteristic parameters of the damaged echo signal; the frequency-domain characteristic parameters of the echo signal include: the healthy frequency-domain characteristic parameters of the healthy echo signal and the damaged frequency-domain characteristic parameters of the damaged echo signal;

[0259] Determine the healthy signal energy characteristic parameter, the healthy time-domain characteristic parameter, and the healthy frequency-domain characteristic parameter as the healthy characteristic parameters of the healthy echo signal;

[0260] Determine the damaged signal energy characteristic parameter, the damaged time-domain characteristic parameter, and the damaged frequency-domain characteristic parameter as the damaged characteristic parameters of the damaged echo signal.

[0261] In some embodiments, the construction module is further configured to construct a feature vector based on the healthy characteristic parameters of the healthy echo signals in the signal reference library to obtain the feature vector X of the healthy echo signal base,n :

[0262]

[0263] where X base,n represents the feature vector of the nth healthy echo signal; represents the first 8 characteristic parameters of the feature vector; represents the 9th to 15th characteristic parameters of the feature vector; represents the last l + 1 characteristic parameters of the feature vector;

[0264] Combine the obtained feature vectors of the healthy echo signals to construct a healthy signal feature matrix:

[0265]

[0266] Among them, n represents the number of healthy echo signals; p represents the p-th characteristic parameter in the eigenvector; X np represents the value of the p-th characteristic parameter of the n-th healthy echo signal, where i = 1, 2, …, n; j = 1, 2, …, p.

[0267] In some embodiments, the calculation module is further configured to perform a normalization process on a pre-acquired reference space according to the healthy signal feature matrix to obtain a normalized reference space:

[0268]

[0269] Among them, is the mean value of the j-th signal characteristic parameter in the feature matrix; δ j is the standard deviation of the j-th signal characteristic parameter of the feature matrix, where i = 1, 2, 3, …, n;

[0270] According to the normalized reference space, establish a normalized reference space:

[0271]

[0272] According to the normalized reference space, calculate the Mahalanobis distance d of the healthy echo signal in the normalized reference space M,normal :

[0273]

[0274] Among them, S represents the correlation coefficient matrix of the normalized healthy signal feature matrix;

[0275] Use the eigenvector of the damaged echo signal to calculate the Mahalanobis distance of the damaged echo signal under the healthy signal feature matrix constructed by the healthy echo signals;

[0276] Statistically count the Mahalanobis distance corresponding to each healthy echo signal to obtain a statistical result;

[0277] According to the statistical result, determine the numerical range of the Mahalanobis distance corresponding to the healthy echo signal as the Mahalanobis space of the healthy echo signal.

[0278] In some embodiments, the integration module is further configured to use the Taguchi optimization method to establish a two-level orthogonal array L n (2 p );

[0279] Through the signal-to-noise ratio gain of the characteristic variables in the eigenvector of the healthy echo signal and the two-level orthogonal array L n (2 p) Screen the sensitive feature parameters from the healthy feature parameters in the feature vector to obtain the screened healthy feature parameters; where the signal-to-noise ratio gain △ is expressed as:

[0280] △ = SNR sel - SNR unsel ;

[0281] Where, SNR sel represents the signal-to-noise ratio using the feature variable; SNR unsel represents the signal-to-noise ratio without using the feature variable; when Δ is greater than 0, select the feature variable; when Δ is less than or equal to 0, remove the feature variable;

[0282] Construct the final feature vector X of the healthy echo signal according to the screened healthy feature parameters final :

[0283] X final = [x1, x2, x3, … x q ;

[0284] Calculate the healthy Mahalanobis distance of the healthy echo signal according to the final feature vector X final .

[0285] Statistically analyze the healthy Mahalanobis distance of each healthy echo signal to obtain the statistical result;

[0286] Determine the numerical range of the healthy Mahalanobis distance as the reference Mahalanobis space according to the statistical result.

[0287] The embodiment of the present application uses the covariance matrix to calculate the Mahalanobis distance to construct the Mahalanobis space. The Mahalanobis distance takes into account the correlation between variables, is not affected by the dimension of the feature vector, and uses the Taguchi method to select and optimize the sensitive feature parameters, greatly reducing the number of feature variables, achieving the purpose of dimensionality reduction and reducing the amount of calculation. And when optimizing the feature parameters, no additional new parameters need to be set, and only the signal-to-noise ratio gain of the feature variable needs to be obtained through continuous data measurement for feature optimization. The embodiment of the present application combines the advantages of the guided wave technology, adopts the Mahalanobis-Taguchi method to perform multi-feature information parameter fusion to judge the health state of the pipeline, mines the information of the data collected by the long-term detection of the pipeline, improves the guided wave monitoring accuracy and reliability, and has practical significance.

[0288] In several embodiments provided by the present application, it should be understood that the disclosed devices and methods can be implemented in a non-targeted manner. The device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed. In addition, the couplings between the various components shown or discussed are either direct couplings or indirect couplings through some interfaces.

[0289] The units described above as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0290] The features disclosed in several method or device embodiments provided by the present application can be arbitrarily combined without conflict to obtain new method embodiments or device embodiments.

[0291] As described above, only some implementation manners of the embodiments of the present application are provided, but the protection scope of the embodiments of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the embodiments of the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the embodiments of the present application. Therefore, the protection scope of the embodiments of the present application should be subject to the protection scope of the claims.

Claims

1. A method for monitoring pipeline damage, characterized in that, The method includes: Obtaining the monitoring signal of the pipeline under test during its in-service period and a pre-determined reference Mahalanobis space; Performing feature extraction on the monitoring signal to obtain the monitoring feature vector of the monitoring signal; Determining the monitoring Mahalanobis distance of the monitoring signal according to the monitoring feature vector and the monitoring feature matrix constructed based on the monitoring feature vector; Comparing the monitoring Mahalanobis distance with the reference Mahalanobis space to obtain the deviation degree between the monitoring Mahalanobis distance and the reference Mahalanobis space; Determining the damage state of the pipeline under test according to the deviation degree; Among them, the steps of determining the reference Mahalanobis space include: Obtaining the healthy echo signal and the damaged echo signal of the pipeline under test; Performing adaptive noise complete ensemble empirical mode decomposition (CEEMD) on the healthy echo signal and the damaged echo signal to obtain the healthy feature parameters of the healthy echo signal and the damaged feature parameters of the damaged echo signal; Constructing a healthy feature matrix corresponding to the healthy echo signal according to the healthy feature parameters; Calculating the Mahalanobis distance of the healthy echo signal under the healthy feature matrix, and integrating the Mahalanobis distances of the healthy echo signals to obtain a Mahalanobis space; Calculating the reference Mahalanobis distance of the healthy echo signal through the Taguchi optimization method based on the healthy feature parameters, and integrating the reference Mahalanobis distances to obtain the reference Mahalanobis space; Before obtaining the healthy echo signal and the damaged echo signal of the pipeline under test, the method further includes: Setting monitoring points on the pipeline under test, where each monitoring point corresponds to a guided wave transducer for receiving echo signals; The steps of obtaining the healthy echo signal and the damaged echo signal of the pipeline under test include: Exciting and receiving a first group of echo signals at the monitoring point through a guided wave signal acquisition system, and determining the first group of echo signals as the original healthy echo signal; Simulating the damaged condition of the pipeline through a target damage object placed on the pipeline under test; When the pipeline under test is damaged, exciting and receiving a second group of echo signals at the monitoring point, and determining the second group of echo signals as the original damaged echo signal; Establishing a signal reference library according to the original healthy echo signal and the original damaged echo signal; Performing signal preprocessing on the original healthy echo signal and the original damaged echo signal, and filtering the preprocessed original healthy echo signal and the preprocessed original damaged echo signal by using wavelet denoising; Determining the filtered original healthy echo signal as the healthy echo signal of the pipeline under test, and determining the filtered original damaged echo signal as the damaged echo signal of the pipeline under test; Among them, the constructing a healthy feature matrix corresponding to the healthy echo signal according to the healthy feature parameters includes: Construct a feature vector based on the health feature parameters of the healthy echo signals in the signal reference library to obtain the feature vector of the healthy echo signals : ； Among them, represents the eigenvector of the nth healthy echo signal; represents the first 8 eigenparameters of the eigenvector; represents the 9th to 15th eigenparameters of the eigenvector; represents the last eigenparameters of the eigenvector; Combining the feature vectors of the obtained healthy echo signals to construct a healthy signal feature matrix: ； Among them, represents the number of healthy echo signals; represents the th feature parameter in the eigenvector; represents the th healthy echo signal's th feature parameter value, .

2. The method according to claim 1, wherein The echo signal includes a healthy echo signal and a damaged echo signal; performing CEEMD decomposition on the healthy echo signal and the damaged echo signal to obtain the healthy characteristic parameters of the healthy echo signal and the damaged characteristic parameters of the damaged echo signal, including: Performing CEEMD decomposition on the echo signal to obtain a plurality of intrinsic mode functions IMF, where the general calculation formula of the intrinsic mode function IMF is: ； Among them, is the (k + 1)-th intrinsic mode component obtained by CEEMD decomposition; is the k-th signal mode component obtained by CEEMD decomposition; is the residual signal; is the Gaussian white noise added during the i-th CEEMD decomposition; Each empirical mode decomposition needs to be performed L times, The coefficient represents the noise coefficient selected during each mode decomposition; Add Gaussian white noise to the echo signal to obtain the processed echo signal : ； Obtaining the L-th order L intrinsic mode functions IMF obtained by performing L times of CEEMD decomposition on the processed echo signal; Weighted average is performed on L intrinsic mode functions IMF to obtain the average intrinsic mode function at the current order CEEMD decomposition stage : ； Among them, is the i-th intrinsic mode function IMF among the L intrinsic mode components, where i = 1, 2, … L; Subtract the average intrinsic mode component obtained by CEEMD decomposition from the echo signal, and determine the result as the residual signal, which is expressed as: ​ ； For the remaining signal perform CEEMD decomposition until the remaining signal can no longer be decomposed, so that the original signal is decomposed into multiple intrinsic mode functions IMFs and a residual signal r; among them, the original signal is expressed as: ； Wherein, the residual signal r is the signal remaining after CEEMD decomposition; K is the number of times of CEEMD decomposition; Determining the correlation coefficient between the k-th intrinsic mode function IMF and the echo signal to obtain a correlation coefficient sequence of the intrinsic mode function IMF and the echo signal; Performing signal feature extraction on the echo signal according to the correlation coefficient to obtain the healthy characteristic parameters of the healthy echo signal and the damaged characteristic parameters of the damaged echo signal.

3. The method according to claim 2, wherein Determining the correlation coefficient between the k-th intrinsic mode function IMF and the echo signal to obtain a correlation coefficient sequence of the intrinsic mode function IMF and the echo signal, including: The correlation coefficient between the intrinsic mode function IMF and the processed echo signal is as follows: ； Among them, is the correlation coefficient between the k-th intrinsic mode function IMF and the processed echo signal ; is the mean value of the echo signal with added Gaussian white noise; is the intrinsic mode function of the k-th mode decomposition; is the average value of the intrinsic mode functions obtained from multiple decompositions at the k-th time; is the standard deviation of the echo signal with added Gaussian white noise; is the standard deviation of the k-th intrinsic mode function; E[] represents the statistical average value; According to the correlation coefficient between each intrinsic mode function (IMF) and the echo signal, the correlation coefficient is screened and sorted to obtain a correlation coefficient sequence: 。 4. The method according to claim 3, characterized in that, Performing signal feature extraction on the echo signal according to the correlation coefficient to obtain the healthy characteristic parameters of the healthy echo signal and the damaged characteristic parameters of the damaged echo signal, including: The correlation coefficient greater than the preset threshold intrinsic mode components are determined as the characteristic components of the processed echo signal ; According to the characteristic component, the processed echo signal is subjected to signal feature extraction to obtain the energy of each corresponding intrinsic mode component, where the energy of the intrinsic mode component is expressed as: ； Among them, the total signal energy is the sum of energies . For each component, the energy entropy is as follows: ； The energy entropy of the component is determined as the signal energy characteristic parameter of the echo signal; wherein, the signal energy characteristic parameter of the echo signal includes the healthy signal energy characteristic parameter of the healthy echo signal and the damaged signal energy characteristic parameter of the damaged echo signal; Performing parameter extraction on the echo signal to obtain the time-domain characteristic parameters and frequency-domain characteristic parameters of the echo signal; wherein, the time-domain characteristic parameters of the echo signal include: the healthy time-domain characteristic parameters of the healthy echo signal and the damaged time-domain characteristic parameters of the damaged echo signal; the frequency-domain characteristic parameters of the echo signal include: the healthy frequency-domain characteristic parameters of the healthy echo signal and the damaged frequency-domain characteristic parameters of the damaged echo signal; Determining the healthy signal energy characteristic parameters, the healthy time-domain characteristic parameters, and the healthy frequency-domain characteristic parameters as the healthy characteristic parameters of the healthy echo signal; Determining the damaged signal energy characteristic parameters, the damaged time-domain characteristic parameters, and the damaged frequency-domain characteristic parameters as the damaged characteristic parameters of the damaged echo signal.

5. The method according to claim 1, characterized in that, Calculating the Mahalanobis distance of the healthy echo signal under the healthy characteristic matrix and integrating the Mahalanobis distance of the healthy echo signal to obtain a Mahalanobis space, including: Performing standardization processing on the pre-obtained reference space according to the healthy signal characteristic matrix to obtain a standardized reference space: ； Among them, is the mean value of the th signal feature parameter in the feature matrix; is the standard deviation of the th signal feature parameter of the feature matrix, ; Establishing a standardized reference space according to the standardized reference space: ； Calculate the Mahalanobis distance of the healthy echo signal in the standardized reference space according to the standardized reference space : ； Among them, represents the correlation coefficient matrix of the standardized health signal feature matrix; Using the eigenvector of the damaged echo signal, calculating the Mahalanobis distance of the damaged echo signal under the healthy signal characteristic matrix constructed by the healthy echo signal; Counting the Mahalanobis distance corresponding to each healthy echo signal to obtain a statistical result; According to the statistical result, determining the numerical range of the Mahalanobis distance corresponding to the healthy echo signal as the Mahalanobis space of the healthy echo signal.

6. The method according to claim 5, characterized in that, The steps for determining the reference Mahalanobis space include: Use the Taguchi optimization method to establish a two-level orthogonal array ; Through the signal-to-noise ratio gain of the characteristic variables in the eigenvector of the healthy echo signal and the two-level orthogonal array , the sensitive characteristic parameters of the healthy characteristic parameters in the eigenvector are screened to obtain the screened healthy characteristic parameters; among them, the signal-to-noise ratio gain △ is expressed as: ； Among them, represents the signal-to-noise ratio using the characteristic variable; represents the signal-to-noise ratio without using the characteristic variable; when △ is greater than 0, the said characteristic variable is selected; when △ is less than or equal to 0, the said characteristic variable is removed. Construct the final feature vector of the healthy echo signal according to the screened health feature parameters : ； According to the final eigenvector , calculate the healthy Mahalanobis distance of the healthy echo signal; Counting the healthy Mahalanobis distance of each healthy echo signal to obtain a statistical result; According to the statistical result, determining the numerical range of the healthy Mahalanobis distance as the reference Mahalanobis space.

7. A pipeline damage monitoring device, characterized in that, The device includes: An acquisition module for acquiring monitoring signals of a pipeline under test during service and a pre-determined reference Mahalanobis space; A feature extraction module for extracting features from the monitoring signals to obtain a monitoring feature vector of the monitoring signals; A first determination module for determining a monitoring Mahalanobis distance of the monitoring signals according to the monitoring feature vector and a monitoring feature matrix constructed based on the monitoring feature vector; A comparison module for comparing the monitoring Mahalanobis distance with the reference Mahalanobis space to obtain a deviation degree between the monitoring Mahalanobis distance and the reference Mahalanobis space; A second determination module for determining a damage state of the pipeline under test according to the deviation degree; Wherein, the device further includes a third determination module, and the third determination module is configured to acquire a healthy echo signal and a damaged echo signal of the pipeline under test; perform adaptive noise complete ensemble empirical mode CEEMD decomposition on the healthy echo signal and the damaged echo signal to obtain healthy feature parameters of the healthy echo signal; construct a healthy feature matrix corresponding to the healthy echo signal according to the healthy feature parameters; calculate a Mahalanobis distance of the healthy echo signal under the healthy feature matrix, and integrate the Mahalanobis distances of the healthy echo signal to obtain a Mahalanobis space; calculate a reference Mahalanobis distance of the healthy echo signal through the Taguchi optimization method based on the healthy feature parameters, and integrate the reference Mahalanobis distances to obtain the reference Mahalanobis space; The device further includes: a setting module for setting monitoring points on the pipeline under test, wherein each monitoring point corresponds to a guided wave transducer for receiving echo signals; correspondingly, the third determination module is further configured to, through a guided wave signal acquisition system, excite and receive a first group of echo signals at the monitoring points, and determine the first group of echo signals as the original healthy echo signals; simulate the pipeline damage condition through a target damage object placed on the pipeline under test; in the case of the pipeline under test being damaged, excite and receive a second group of echo signals at the monitoring points, and determine the second group of echo signals as the original damaged echo signals; establish a signal reference library according to the original healthy echo signals and the original damaged echo signals; perform signal preprocessing on the original healthy echo signals and the original damaged echo signals, and filter the preprocessed original healthy echo signals and the preprocessed original damaged echo signals by using a wavelet denoising method; determine the filtered original healthy echo signals as the healthy echo signals of the pipeline under test, and determine the filtered original damaged echo signals as the damaged echo signals of the pipeline under test; Among them, the third determination module is further configured to construct a feature vector according to the health feature parameters of the healthy echo signals in the signal reference library, so as to obtain the feature vectors of the healthy echo signals : ； Among them, represents the eigenvector of the nth healthy echo signal; represents the first 8 eigenparameters of the eigenvector; represents the 9th to 15th eigenparameters of the eigenvector; represents the last eigenparameters of the eigenvector; Combine the obtained feature vectors of the healthy echo signals to construct a healthy signal feature matrix: ； Among them, represents the number of healthy echo signals; represents the th characteristic parameter in the eigenvector; represents the th th characteristic parameter value of the .

Images