Laser-ultrasound-based method and system for identifying and assessing risks of damage to heating pipes

Abstract

The application discloses a kind of heat supply pipeline damage identification and risk assessment method based on laser ultrasound, comprising: the method for establishing heat supply pipeline digital twin model using mechanism modeling and data identification;Based on heat supply pipeline digital twin model simulation, laser ultrasound is used to detect the defect damage part of heat supply pipeline, and the simulation data of heat supply pipeline damage are obtained;In the actual scene of the heat supply pipeline to be identified, a laser ultrasound identification device is added to obtain the actual data of the heat supply pipeline damage;The simulation data of heat supply pipeline damage and the actual data of heat supply pipeline damage are used as pipeline damage identification samples, sample denoising is carried out, and a heat supply pipeline damage quantitative identification model is established using a machine learning algorithm;Based on the heat supply pipeline damage quantitative identification model, the heat supply pipeline damage prediction value is obtained, and the heat supply pipeline damage risk assessment is carried out using the fuzzy cloud theory method.

Description

Technical Field

[0001] This invention belongs to the field of heating pipeline health management technology, specifically relating to a method for identifying and assessing the damage to heating pipelines based on laser ultrasound. Background Technology

[0002] With the continuous development of society and the economy, centralized heating, as one of my country's important infrastructures, is also booming. To achieve clean heating, utilizing waste heat from suburban thermal power plants is a major trend, leading to the increasing application of long-distance heating pipelines. Therefore, ensuring the safe, efficient, and energy-saving operation of heating pipelines is crucial, as damage to these pipelines will seriously affect the public's livelihood in ensuring the heating system's effectiveness.

[0003] During operation, heating pipelines can suffer damage such as cracks and pits due to uneven external stress and internal fluid corrosion. If long-term heating pipelines are not inspected and maintained in a timely manner, the damage can worsen, threatening production and daily life safety. Furthermore, to more clearly visualize the damage risks to heating pipelines during use, risk assessments are necessary.

[0004] Current damage identification technologies have limited dimensionality, only capable of identifying surface damage in pipelines and unable to perform in-depth damage identification. Compared to traditional ultrasonic identification and detection methods, laser ultrasonic technology offers advantages such as real-time, dynamic detection of heating pipeline damage, high identification sensitivity, and higher recognition accuracy for minute damage. Therefore, applying laser ultrasonic technology to the field of heating pipeline damage identification is of great research significance. However, how to use laser ultrasonic technology for rapid, accurate, and quantitative analysis of heating pipeline damage, and how to combine the damage status of heating pipelines with an assessment model to conduct reliability risk assessment of the damage status, are urgent problems to be solved.

[0005] Based on the above technical problems, it is necessary to design a method for damage identification and risk assessment of heating pipelines based on laser ultrasound. Summary of the Invention

[0006] The technical problem to be solved by this invention is to overcome the shortcomings of the prior art and provide a laser ultrasound-based method for identifying and assessing the damage to heating pipelines. This method can quantitatively analyze the damage to heating pipelines through laser ultrasound signal characteristics and machine learning algorithms, effectively improving the efficiency of laser ultrasound-based identification of heating pipeline damage. In addition, by using fuzzy cloud theory to assess the risk of heating pipeline damage, this method makes full use of the quantitative analysis of heating pipeline damage, combines it with other assessment indicators, and integrates the subjective and objective factors of each indicator in the assessment system. This makes the weights of each evaluation indicator more consistent with engineering practice and takes into account the fuzziness and randomness in the assessment process, ensuring the reliability of the risk assessment of heating pipelines.

[0007] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0008] This invention provides a method for damage identification and risk assessment of heating pipelines based on laser ultrasound. The method includes:

[0009] A digital twin model of the heating pipeline was established using mechanistic modeling and data identification methods.

[0010] Based on the digital twin model of the heating pipeline, laser ultrasound is used to detect defects and damage in the heating pipeline and obtain simulation data of the heating pipeline damage.

[0011] In the actual scenario of the heating pipeline to be identified, a laser ultrasonic identification device is added to obtain actual data on the damage to the heating pipeline;

[0012] The simulated data of heating pipeline damage and the actual data of heating pipeline damage are used as pipeline damage identification samples. The samples are denoised and a machine learning algorithm is used to establish a quantitative identification model of heating pipeline damage.

[0013] Based on the quantitative identification model for heating pipeline damage, the predicted value of heating pipeline damage is obtained, and the fuzzy cloud theory method is used to assess the risk of heating pipeline damage.

[0014] Furthermore, the method of establishing a digital twin model of the heating pipeline using mechanistic modeling and data identification includes:

[0015] Construct physical entity models, logical models, and simulation models of the heating pipeline; among them,

[0016] The establishment of the logical model includes: establishing a controllable closed-loop logical model based on the logical mechanism relationship of the physical entity of the heating pipeline, and mapping the physical model to the logical model;

[0017] The establishment of the simulation model includes: building a simulation model of the heating pipeline based on the collected operation data, status data, and physical property data of the heating pipeline; optimizing the parameters of the simulation model according to the error between the predicted value and the actual value output by the simulation model; the simulation model also includes a laser ultrasonic simulation unit, which analyzes the interaction process between the laser and the tested heating pipeline from the principle of laser-induced ultrasonic waves, and models and simulates the physical process of laser-induced ultrasonic waves.

[0018] The physical entity model, logical model, and simulation model are fused together to construct a system-level digital twin model of the physical entity of the heating pipeline in virtual space.

[0019] The real-time operating data of the heating pipeline under multiple operating conditions is input into the system-level digital twin model. The simulation results of the system-level digital twin model are adaptively identified and corrected using the reverse identification method to obtain the identified and corrected digital twin model of the heating pipeline.

[0020] Furthermore, the laser ultrasonic identification device includes: a laser ultrasonic emitting unit, a laser ultrasonic signal acquisition and receiving unit, and a laser controller display unit. During the damage identification process of the heating pipe under test, the laser controller display unit synchronously controls the laser signal emitting unit and the laser ultrasonic signal acquisition unit respectively: controlling the triggering of the laser in the laser ultrasonic emitting unit and the electric scanning mirror to focus and scan the laser source; storing the signal acquired by the laser ultrasonic signal acquisition and receiving unit after filtering, amplification, and A / D conversion; and visually displaying the waveform data of the received ultrasonic signal through inversion. The laser ultrasonic signal acquisition and receiving unit adopts an indirect contact identification and detection method, using an ultrasonic probe to contact the surface of the heating pipe under test to achieve laser ultrasonic identification and detection, and using a filter, amplifier, and A / D converter to filter, amplify, and convert the signal.

[0021] Furthermore, the step of using the simulated damage data and the actual damage data of the heating pipeline as pipeline damage identification samples, performing sample denoising, and establishing a quantitative identification model for heating pipeline damage using machine learning algorithms includes:

[0022] The simulated data of heating pipeline damage and the actual data of heating pipeline damage are used as pipeline damage identification samples. Sample data preprocessing is performed, including missing value processing, outlier processing and data standardization.

[0023] The preprocessed samples are denoised using a combination of empirical mode decomposition and wavelet thresholding: the noisy sample signal is decomposed using empirical mode decomposition to obtain high-frequency IMF components, low-frequency IMF components and residual signals, and then the high-frequency IMF components are denoised using wavelet thresholding. Finally, the high-frequency IMF components, low-frequency IMF components and residual signals are reconstructed to obtain the denoised sample.

[0024] After denoising, the laser ultrasonic signal is subjected to wavelet packet decomposition to extract relevant features. The wavelet packet energy features that are strongly correlated with the damage size of the heating pipeline are then extracted. The wavelet packet energy features and the time-domain and frequency-domain features of the laser ultrasonic signal are then used to construct a feature vector.

[0025] After the feature vectors are input into the ELM model optimized by the Grey Wolf Optimization Algorithm for learning and training, a quantitative identification model for heating pipeline damage is established.

[0026] Furthermore, the denoising process for high-frequency IMF components using the wavelet threshold denoising method includes:

[0027] Select appropriate wavelet basis functions to perform wavelet transform on noisy high-frequency IMF components; perform thresholding on the transformed wavelet coefficients, where coefficients below the threshold are generated by noise signals and coefficients above the threshold are generated by the original signal; discard wavelet coefficients below the threshold and retain coefficients above the threshold for signal denoising.

[0028] The denoised samples are then subjected to wavelet packet decomposition to extract relevant features from the laser-ultrasonic signal. Wavelet packet energy features with strong correlation to the damage size of the heating pipeline are identified. These wavelet packet energy features, along with the time-domain and frequency-domain features of the laser-ultrasonic signal, are then combined to construct a feature vector, including:

[0029] The laser ultrasonic signal is decomposed by wavelet packet decomposition. The decomposed damage signal generates wavelet packet coefficients in the corresponding frequency bands. By utilizing the different energy distributions of the wavelet packet decomposed signal, the energy distribution characteristics of the laser ultrasonic signal in each frequency band can be obtained, which is the wavelet packet energy characteristic.

[0030] The time difference between the peaks and troughs of the reflected echoes from damages of different sizes in the laser-ultrasound signal is used as the time-domain feature of the damage; the spectral energy of damages of different sizes is obtained by spectral transformation of the laser-ultrasound transmitted wave signal as the frequency-domain feature of the damage.

[0031] The wavelet packet energy features, the damage time-domain features, and the damage frequency-domain features are collectively constructed into a feature vector.

[0032] Furthermore, the step of inputting the feature vector into the ELM model optimized by the GWO (Grey Wolf) optimization algorithm for learning and training to establish a quantitative identification model for heating pipeline damage includes:

[0033] Set the initial parameters of the GWO gray wolf optimization algorithm, including the number of wolves, the maximum number of iterations, and the range of optimization parameter values; initialize the positions of gray wolves at levels α, β, and δ, and the objective function values;

[0034] The ELM model parameters are randomly generated, and the mean squared error is used as the objective function value. These parameters are compared with the objective function values ​​of gray wolves at each level, and the larger objective function value is used to replace the objective function.

[0035] The positions and objective function values ​​of gray wolves at levels α, β, and δ are continuously updated. The position of the optimal gray wolf individual is the best value for the input layer weights and hidden layer thresholds of the ELM model.

[0036] After training the ELM model by inputting the feature vectors into the optimized parameters, a quantitative identification model for heating pipeline damage is established.

[0037] Furthermore, the heating pipeline damage identification and risk assessment method also includes: obtaining a laser ultrasonic pipeline damage visualization image through the laser ultrasonic identification device, and after denoising, enhancing, segmenting and extracting features from the image, establishing a qualitative identification model for heating pipeline damage by using a machine learning algorithm for learning and training.

[0038] The method for denoising the image, which employs a median denoising method based on wavelet threshold change, includes: performing median filtering on the original image, performing wavelet layer-by-layer transformation to obtain multiple sub-images, setting a threshold according to the generated coefficient matrix, performing median filtering on each sub-image again, and restoring the image through wavelet reconstruction based on the newly generated information matrix to obtain a denoised image.

[0039] The image enhancement process employs a Retinex enhancement algorithm based on adaptive weights, comprising: applying multiple Retinex enhancement algorithms to the same image, and then weighting and summing the calculation results according to different weights to obtain the enhanced image, as shown below: r i (x,y) represents the output of the re-channel i of the Retinex enhancement algorithm; I i (x,y) represents the pixel value of channel i in the original image; F k (x, y) represents the Gaussian wrap function; k is the number of times the wrap function is calculated, i.e., the number of elements involved in the weighted summation; w k The weight of each element;

[0040] When segmenting the image, the image segmentation algorithm used includes at least threshold-based image segmentation, edge-based image segmentation, region-based image segmentation, clustering-based segmentation, and neural network-based segmentation.

[0041] When extracting features from an image, the eigenvalues ​​of the gray-level co-occurrence matrix are calculated using gray-level co-occurrence matrix analysis to describe the image features, including at least the second moment of the angle, contrast, entropy, inverse moment, variance, mean sum, variance sum, clustering shadow, and significant clustering.

[0042] The establishment of a qualitative identification model for heating pipeline damage after learning and training using machine learning algorithms includes:

[0043] The penalty function factor C and kernel width σ of the SVM model are optimized using the hybrid frog leaping optimization algorithm (SFLA).

[0044] Initialize the frog population, divide the n frogs in the population into n meme groups, calculate the individual fitness value and sort them, frog i (i = 1, 2, ..., n) corresponds to the i-th meme group, and the (n+1)-th frog enters the frog population of meme group 1 for grouping.

[0045] Update the frog's jump step size and position in the meme group, represented as: D i =r(P b (k)-P a (k)); P a (k+1)=P a (k)+D i ;D i P represents the distance the frog moves. b and P a Let r be the frog at the position corresponding to the best and worst fitness in the current meme group; r∈[0,1];

[0046] When the new solution is better, replace the worst individual; otherwise, use the frog P in the best position. g Replace P b The optimal frog position is determined through iterative processes, which are the penalty function factor C and kernel width σ of the SVM model.

[0047] The extracted image features are input into the optimized SVM model for training, and a qualitative identification model for heating pipeline damage is established.

[0048] Furthermore, the step of obtaining predicted values ​​of heating pipe damage based on the quantitative identification model for heating pipe damage, and using fuzzy cloud theory to conduct a risk assessment of heating pipe damage, includes:

[0049] Based on the quantitative identification model for heating pipeline damage, the predicted value of heating pipeline damage is obtained. The predicted value of heating pipeline damage is combined with the pre-calculated fatigue index of heating pipeline, geological index of pipeline location, and remaining life index of pipeline as the risk assessment index of heating pipeline. A risk assessment grading standard is formulated, and a risk assessment index system for heating pipeline is constructed, including Level 1 healthy state, Level 2 good state, Level 3 general state, Level 4 poor state, and Level 5 dangerous state.

[0050] The subjective weight W1 is determined by using the fuzzy hierarchical analysis method, and the objective weight W2 is determined by using the coefficient of variation method. Finally, the combined weight W of each evaluation indicator is determined by using the least squares comprehensive weighting method to combine the subjective and objective weights of each evaluation indicator.

[0051] Based on cloud model theory, cloud model parameters for each corresponding level are calculated according to the grading standards of various evaluation indicators. A forward cloud generator is then used to generate the subordinate clouds for each evaluation indicator. The cloud model parameters include the expected E... x Entropy E n and hyperentropy He ;

[0052] The measured data to be evaluated are substituted into the X-conditional cloud generator to obtain the membership degree μ of each evaluation index for each level. Combined with the combined weight W, the membership degree vector of the overall level evaluation is calculated. Then, the risk assessment level S is determined based on the principle of maximum membership; the X-conditional cloud generator: when the quantitative value x is known, combined with the cloud digital characteristics, according to μ(x)=exp[-(xE) x ) 2 / 2(E n ′) 2 The cloud droplet drop(x,μ) was calculated; E n ′~N(E x E n 2 );

[0053] Based on fuzzy entropy theory, the fuzzy entropy E of the comprehensive risk level assessment of heating pipelines is calculated, and the complexity of the comprehensive risk level assessment results is analyzed as an auxiliary parameter for the risk level evaluation, thus obtaining the final risk assessment result (S, E) of the heating pipelines; the fuzzy entropy... n is the total number of grades, μ i is the membership degree of level i in the risk assessment of heating pipelines, and m′ is the standardization coefficient.

[0054] Furthermore, the determination of subjective weight W1 using fuzzy hierarchical analysis includes:

[0055] Construct a fuzzy consistency matrix of risk assessment indicators for heating pipelines:

[0056] Construct a fuzzy complementary judgment matrix C for the risk assessment index system of heating pipelines, and obtain r by summing its rows. i , for r i Process to obtain r ij To ensure it meets the requirements of fuzzy consistency matrix elements, the calculation process for fuzzy consistency processing is as follows: n is the number of evaluation indicators;

[0057] According to r ij A fuzzy, consistent judgment matrix for the risk assessment indicators of heating pipelines is formed, represented as follows:

[0058]

[0059] The subjective weight of the i-th heating pipeline risk assessment indicator is calculated and expressed as follows: To sum the rows of the fuzzy consistency judgment matrix R;

[0060] Consistency check: The weight feature matrix is ​​calculated as follows: wij =w i / (w i +w j ); i,j=1,2,…,n; calculate the compatibility index between the weight feature matrix and the set fuzzy consistency judgment matrix, expressed as: The closer I is to zero, the stronger the consistency of the judgment matrix;

[0061] The weight vector W1 = [w] passed the consistency test 11 ,w 12 ,…,w 1n [This refers to the subjective weights of each risk assessment indicator;]

[0062] The determination of the objective weight W2 using the coefficient of variation method includes:

[0063] The sample matrix B = (b...) is constructed using sample data of n evaluation indicators from m groups of evaluation objects. ij ) m×n b ij Let be the measured value of the j-th evaluation indicator for the i-th evaluation subject; calculate the mean of each indicator. Standard deviation

[0064] The weight value of the j-th heating pipeline risk assessment indicator is calculated and expressed as:

[0065] Obtain the weight vector W2 = [w 21 ,w 22 ,…,w 2n [This refers to the objective weights of each risk assessment indicator;]

[0066] The method of determining the combined weight W of each evaluation indicator by combining the subjective and objective weights of each indicator using the least squares comprehensive weighting method includes:

[0067] Set the combination weights W = [w1, w2, ..., w n The optimization objective function is:

[0068]

[0069]

[0070] The objective function is solved using the Lagrange method. The Lagrange multiplier λ is used to apply equality constraints to the objective function, and the Lagrange function is constructed as follows:

[0071]

[0072] Establish parameters for the Lagrange function with respect to w. iThe first-order partial derivatives of λ and λ, set to zero, are used to calculate the extremum of the objective function:

[0073]

[0074] Solving the above formula yields the combined weights W = [w1, w2, ..., w...] for each evaluation indicator. n ].

[0075] This invention also provides a laser-ultrasound-based system for identifying and assessing the damage to heating pipelines, the system comprising:

[0076] The digital twin model building unit is used to build a digital twin model of the heating pipeline using mechanistic modeling and data identification methods.

[0077] The damage simulation data acquisition unit is used to simulate the detection of defects and damage in the heating pipeline using laser ultrasound based on the digital twin model of the heating pipeline, and to acquire damage simulation data of the heating pipeline.

[0078] The actual damage data acquisition unit is used to acquire actual damage data of heating pipelines in the actual scenario of the heating pipeline to be identified by adding a laser ultrasonic identification device.

[0079] The damage quantitative identification unit is used to take the simulated damage data and the actual damage data of the heating pipeline as pipeline damage identification samples, perform sample denoising, and establish a quantitative identification model of heating pipeline damage using machine learning algorithms.

[0080] The damage risk assessment unit is used to obtain the predicted value of heating pipe damage based on the quantitative identification model of heating pipe damage, and to conduct heating pipe damage risk assessment using the fuzzy cloud theory method.

[0081] The beneficial effects of this invention are:

[0082] This invention establishes a digital twin model of a heating pipeline using mechanistic modeling and data identification methods. Based on this digital twin model, laser ultrasound is used to simulate the detection of defects and damage in the heating pipeline, obtaining simulated damage data. In a real-world scenario of the heating pipeline to be identified, a laser ultrasound identification device is added to obtain actual damage data. The simulated and actual damage data are used as pipeline damage identification samples, and denoising is performed on these samples. A quantitative identification model for heating pipeline damage is then established using machine learning algorithms. Based on this quantitative identification model, the damage of the heating pipeline is obtained. The system predicts damage values ​​and uses fuzzy cloud theory to assess the risk of damage to heating pipelines. It can quantitatively analyze the damage to heating pipelines using laser ultrasonic signal characteristics and machine learning algorithms, effectively improving the efficiency of laser ultrasonic damage identification. Furthermore, by employing fuzzy cloud theory for the risk assessment of heating pipeline damage, it fully utilizes the quantitative analysis of pipeline damage, combines it with other assessment indicators, and integrates the subjective and objective factors of each indicator in the assessment system. This makes the weights of each evaluation indicator more consistent with engineering practice and takes into account the fuzziness and randomness inherent in the assessment process, ensuring the reliability of the heating pipeline risk assessment.

[0083] Other features and advantages will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained through the structures particularly pointed out in the description and the drawings.

[0084] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0085] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0086] Figure 1 This is a schematic diagram of the process for a laser-ultrasound-based method for identifying and assessing the damage to heating pipelines according to the present invention.

[0087] Figure 2 This is a schematic diagram of the structure of a laser ultrasound-based heating pipeline damage identification and risk assessment system according to the present invention. Detailed Implementation

[0088] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0089] Example 1

[0090] Figure 1 This is a schematic diagram of a laser ultrasound-based method for identifying and assessing the damage to heating pipelines, as described in this invention.

[0091] like Figure 1 As shown in Embodiment 1, this invention provides a method for identifying and assessing the damage and risks of heating pipelines based on laser ultrasound. The method includes:

[0092] A digital twin model of the heating pipeline was established using mechanistic modeling and data identification methods.

[0093] Based on the digital twin model of the heating pipeline, laser ultrasound is used to detect defects and damage in the heating pipeline and obtain simulation data of the heating pipeline damage.

[0094] In the actual scenario of the heating pipeline to be identified, a laser ultrasonic identification device is added to obtain actual data on the damage to the heating pipeline;

[0095] The simulated data of heating pipeline damage and the actual data of heating pipeline damage are used as pipeline damage identification samples. The samples are denoised and a machine learning algorithm is used to establish a quantitative identification model of heating pipeline damage.

[0096] Based on the quantitative identification model for heating pipeline damage, the predicted value of heating pipeline damage is obtained, and the fuzzy cloud theory method is used to assess the risk of heating pipeline damage.

[0097] It should be noted that the basic principle of laser ultrasonic testing is that the laser irradiates the surface of the heating pipe being tested, causing the surface of the workpiece to be locally heated. The surface temperature of the workpiece rises rapidly, and due to uneven heat dissipation on the workpiece surface, a high-temperature expansion zone is generated around the workpiece. The existence of the high-temperature expansion zone causes the surrounding area of ​​the heating pipe to interact, thereby forming elastic stress waves on the surface of the heating pipe, which is ultrasonic waves.

[0098] In this embodiment, the method of establishing a digital twin model of the heating pipeline using mechanism modeling and data identification includes:

[0099] Construct physical entity models, logical models, and simulation models of the heating pipeline; among them,

[0100] The establishment of the logical model includes: establishing a controllable closed-loop logical model based on the logical mechanism relationship of the physical entity of the heating pipeline, and mapping the physical model to the logical model;

[0101] The establishment of the simulation model includes: building a simulation model of the heating pipeline based on the collected operation data, status data, and physical property data of the heating pipeline; optimizing the parameters of the simulation model according to the error between the predicted value and the actual value output by the simulation model; the simulation model also includes a laser ultrasonic simulation unit, which analyzes the interaction process between the laser and the tested heating pipeline from the principle of laser-induced ultrasonic waves, and models and simulates the physical process of laser-induced ultrasonic waves.

[0102] The physical entity model, logical model, and simulation model are fused together to construct a system-level digital twin model of the physical entity of the heating pipeline in virtual space.

[0103] The real-time operating data of the heating pipeline under multiple operating conditions is input into the system-level digital twin model. The simulation results of the system-level digital twin model are adaptively identified and corrected using the reverse identification method to obtain the identified and corrected digital twin model of the heating pipeline.

[0104] In this embodiment, the laser ultrasonic identification device includes: a laser ultrasonic emitting unit, a laser ultrasonic signal acquisition and receiving unit, and a laser controller display unit. During the damage identification process of the heating pipe under test, the laser controller display unit synchronously controls the laser signal emitting unit and the laser ultrasonic signal acquisition unit respectively: controlling the triggering of the laser in the laser ultrasonic emitting unit and the electric scanning mirror to focus and scan the laser source; storing the signal acquired by the laser ultrasonic signal acquisition and receiving unit after filtering, amplification, and A / D conversion; and visually displaying the waveform data of the received ultrasonic signal through inversion. The laser ultrasonic signal acquisition and receiving unit adopts an indirect contact identification and detection method, using an ultrasonic probe to contact the surface of the heating pipe under test to achieve laser ultrasonic identification and detection, and using a filter, amplifier, and A / D converter to filter, amplify, and convert the signal.

[0105] In this embodiment, the step of using the simulated damage data and the actual damage data of the heating pipeline as pipeline damage identification samples, performing sample denoising, and establishing a quantitative identification model for heating pipeline damage using machine learning algorithms includes:

[0106] The simulated data of heating pipeline damage and the actual data of heating pipeline damage are used as pipeline damage identification samples. Sample data preprocessing is performed, including missing value processing, outlier processing and data standardization.

[0107] The preprocessed samples are denoised using a combination of empirical mode decomposition and wavelet thresholding: the noisy sample signal is decomposed using empirical mode decomposition to obtain high-frequency IMF components, low-frequency IMF components and residual signals, and then the high-frequency IMF components are denoised using wavelet thresholding. Finally, the high-frequency IMF components, low-frequency IMF components and residual signals are reconstructed to obtain the denoised sample.

[0108] After denoising, the laser ultrasonic signal is subjected to wavelet packet decomposition to extract relevant features. The wavelet packet energy features that are strongly correlated with the damage size of the heating pipeline are then extracted. The wavelet packet energy features and the time-domain and frequency-domain features of the laser ultrasonic signal are then used to construct a feature vector.

[0109] After the feature vectors are input into the ELM model optimized by the Grey Wolf Optimization Algorithm for learning and training, a quantitative identification model for heating pipeline damage is established.

[0110] In this embodiment, the denoising process for high-frequency IMF components using the wavelet threshold denoising method includes:

[0111] Select appropriate wavelet basis functions to perform wavelet transform on noisy high-frequency IMF components; perform thresholding on the transformed wavelet coefficients, where coefficients below the threshold are generated by noise signals and coefficients above the threshold are generated by the original signal; discard wavelet coefficients below the threshold and retain coefficients above the threshold for signal denoising.

[0112] The denoised samples are then subjected to wavelet packet decomposition to extract relevant features from the laser-ultrasonic signal. Wavelet packet energy features with strong correlation to the damage size of the heating pipeline are identified. These wavelet packet energy features, along with the time-domain and frequency-domain features of the laser-ultrasonic signal, are then combined to construct a feature vector, including:

[0113] The laser ultrasonic signal is decomposed by wavelet packet decomposition. The decomposed damage signal generates wavelet packet coefficients in the corresponding frequency bands. By utilizing the different energy distributions of the wavelet packet decomposed signal, the energy distribution characteristics of the laser ultrasonic signal in each frequency band can be obtained, which is the wavelet packet energy characteristic.

[0114] The time difference between the peaks and troughs of the reflected echoes from damages of different sizes in the laser-ultrasound signal is used as the time-domain feature of the damage; the spectral energy of damages of different sizes is obtained by spectral transformation of the laser-ultrasound transmitted wave signal as the frequency-domain feature of the damage.

[0115] The wavelet packet energy features, the damage time-domain features, and the damage frequency-domain features are collectively constructed into a feature vector.

[0116] It's important to note that Empirical Mode Decomposition (EMD) separates fluctuations at different scales from a signal, adaptively generating a series of data sequences arranged from high to low frequencies. These sequences are called Intrinsic Mode Functions (IMFs). IMFs are stationary signals with simple components, allowing for analysis to extract crucial information from the original signal. In practical applications, noise often resides in the high-frequency region. Traditional denoising methods discard the high-frequency components and reconstruct the low-frequency components to achieve noise reduction, but this approach loses the original signal in the high-frequency region. Therefore, combining EMD with wavelet thresholding results in better denoising performance.

[0117] In this embodiment, the step of inputting the feature vector into the ELM model optimized by the GWO (Grey Wolf) optimization algorithm for learning and training to establish a quantitative identification model for heating pipeline damage includes:

[0118] Set the initial parameters of the GWO gray wolf optimization algorithm, including the number of wolves, the maximum number of iterations, and the range of optimization parameter values; initialize the positions of gray wolves at levels α, β, and δ, and the objective function values;

[0119] The ELM model parameters are randomly generated, and the mean squared error is used as the objective function value. These parameters are compared with the objective function values ​​of gray wolves at each level, and the larger objective function value is used to replace the objective function.

[0120] The positions and objective function values ​​of gray wolves at levels α, β, and δ are continuously updated. The position of the optimal gray wolf individual is the best value for the input layer weights and hidden layer thresholds of the ELM model.

[0121] After training the ELM model by inputting the feature vectors into the optimized parameters, a quantitative identification model for heating pipeline damage is established.

[0122] In this embodiment, the heating pipeline damage identification and risk assessment method further includes: obtaining a laser ultrasonic pipeline damage visualization image through the laser ultrasonic identification device, and after denoising, enhancing, segmenting and extracting features from the image, establishing a qualitative identification model for heating pipeline damage by using a machine learning algorithm for learning and training.

[0123] The method for denoising the image, which employs a median denoising method based on wavelet threshold change, includes: performing median filtering on the original image, performing wavelet layer-by-layer transformation to obtain multiple sub-images, setting a threshold according to the generated coefficient matrix, performing median filtering on each sub-image again, and restoring the image through wavelet reconstruction based on the newly generated information matrix to obtain a denoised image.

[0124] The image enhancement process employs a Retinex enhancement algorithm based on adaptive weights, comprising: applying multiple Retinex enhancement algorithms to the same image, and then weighting and summing the calculation results according to different weights to obtain the enhanced image, as shown below: r i (x,y) represents the output of the re-channel i of the Retinex enhancement algorithm; I i (x,y) represents the pixel value of channel i in the original image; F k (x, y) represents the Gaussian wrap function; k is the number of times the wrap function is calculated, i.e., the number of elements involved in the weighted summation; w k The weight of each element;

[0125] When segmenting the image, the image segmentation algorithm used includes at least threshold-based image segmentation, edge-based image segmentation, region-based image segmentation, clustering-based segmentation, and neural network-based segmentation.

[0126] When extracting features from an image, the eigenvalues ​​of the gray-level co-occurrence matrix are calculated using gray-level co-occurrence matrix analysis to describe the image features, including at least the second moment of the angle, contrast, entropy, inverse moment, variance, mean sum, variance sum, clustering shadow, and significant clustering.

[0127] The establishment of a qualitative identification model for heating pipeline damage after learning and training using machine learning algorithms includes:

[0128] The penalty function factor C and kernel width σ of the SVM model are optimized using the hybrid frog leaping optimization algorithm (SFLA).

[0129] Initialize the frog population, divide the n frogs in the population into n meme groups, calculate the individual fitness value and sort them, frog i (i = 1, 2, ..., n) corresponds to the i-th meme group, and the (n+1)-th frog enters the frog population of meme group 1 for grouping.

[0130] Update the frog's jump step size and position in the meme group, represented as: D i =r(P b (k)-P a (k)); P a (k+1)=P a (k)+D i ;D i P represents the distance the frog moves. b and P a Let r be the frog at the position corresponding to the best and worst fitness in the current meme group; r∈[0,1];

[0131] When the new solution is better, replace the worst individual; otherwise, use the frog P in the best position. g Replace Pb The optimal frog position is determined through iterative processes, which are the penalty function factor C and kernel width σ of the SVM model.

[0132] The extracted image features are input into the optimized SVM model for training, and a qualitative identification model for heating pipeline damage is established.

[0133] It should be noted that a combination of wavelet thresholding and median filtering is used for denoising. For the salt-and-pepper noise present in the original image, a median filter is first performed to obtain an image with almost no salt-and-pepper noise removed. The median-filtered image is then subjected to a wavelet transform, and a threshold is set based on the generated coefficient matrix to smooth the Gaussian noise. Finally, an inverse wavelet transform is performed on the processed coefficient matrix, and the image is reconstructed based on the newly generated coefficient matrix, thus eliminating the mixed noise interference from the original image.

[0134] The core of the traditional Retinex enhancement algorithm is determining the wrap function. However, when a heating pipe is damaged, the image brightness information changes significantly, making it difficult to obtain the optimal result through a single calculation of the wrap function, leading to unsatisfactory image enhancement. To better complete the image enhancement process and achieve higher resolution and contrast in the ultrasonic visualization image of the heating pipe, an adaptive weight-based Retinex enhancement algorithm is adopted. This algorithm allows the convolution function to traverse the entire scale range to ensure the most ideal enhancement effect.

[0135] The SFLA–SVM hybrid algorithm combines the advantages of the former, such as strong global optimization and fast convergence, and makes full use of the advantages of SVM algorithm, such as fast computation performance and high prediction accuracy, and has good robustness for predicting damage to heating pipelines.

[0136] In this embodiment, obtaining predicted values ​​of heating pipe damage based on the quantitative identification model for heating pipe damage, and conducting a risk assessment of heating pipe damage using fuzzy cloud theory, includes:

[0137] Based on the quantitative identification model for heating pipeline damage, the predicted value of heating pipeline damage is obtained. The predicted value of heating pipeline damage is combined with the pre-calculated fatigue index of heating pipeline, geological index of pipeline location, and remaining life index of pipeline as the risk assessment index of heating pipeline. A risk assessment grading standard is formulated, and a risk assessment index system for heating pipeline is constructed, including Level 1 healthy state, Level 2 good state, Level 3 general state, Level 4 poor state, and Level 5 dangerous state.

[0138] The subjective weight W1 is determined by using the fuzzy hierarchical analysis method, and the objective weight W2 is determined by using the coefficient of variation method. Finally, the combined weight W of each evaluation indicator is determined by using the least squares comprehensive weighting method to combine the subjective and objective weights of each evaluation indicator.

[0139] Based on cloud model theory, cloud model parameters for each corresponding level are calculated according to the grading standards of various evaluation indicators. A forward cloud generator is then used to generate the subordinate clouds for each evaluation indicator. The cloud model parameters include the expected E... x Entropy E n and hyperentropy H e ;

[0140] The measured data to be evaluated are substituted into the X-conditional cloud generator to obtain the membership degree μ of each evaluation index for each level. Combined with the combined weight W, the membership degree vector of the overall level evaluation is calculated. Then, the risk assessment level S is determined based on the principle of maximum membership; the X-conditional cloud generator: when the quantitative value x is known, combined with the cloud digital characteristics, according to μ(x)=exp[-(xE) x ) 2 / 2(E n ′) 2 The cloud droplet drop(x,μ) was calculated; E n ′~N(E x E n 2 );

[0141] Based on fuzzy entropy theory, the fuzzy entropy E of the comprehensive risk level assessment of heating pipelines is calculated, and the complexity of the comprehensive risk level assessment results is analyzed as an auxiliary parameter for the risk level evaluation, thus obtaining the final risk assessment result (S, E) of the heating pipelines; the fuzzy entropy... n is the total number of grades, μ i is the membership degree of level i in the risk assessment of heating pipelines, and m′ is the standardization coefficient.

[0142] It should be noted that the cloud model is implemented by constructing a cloud generator. Based on different inputs and outputs, it is mainly divided into forward cloud generators and reverse cloud generators. A forward cloud generator is an algorithm that converts qualitative concepts into quantitative representations. In the context of heating pipeline risk assessment, this manifests as transforming risk level intervals into a cloud model composed of multiple regularly distributed cloud droplets. A reverse cloud generator, on the other hand, converts quantitative relationships into qualitative concepts. In the context of heating pipeline risk assessment, this manifests as transforming a large amount of sample data into a cloud model that reflects the overall properties of the sample.

[0143] The risk assessment of heating pipelines involves a large amount of information, and the classification of various assessment indicators varies greatly. It is usually difficult to make a qualitative conclusion based on a single level assessment, which is not enough to truly and comprehensively reflect the status of heating pipelines. Therefore, fuzzy entropy E is introduced into the risk assessment of heating pipelines to conduct complexity analysis on the comprehensive evaluation results of the cloud model, which serves as an auxiliary evaluation parameter for level assessment and yields the final two-dimensional assessment results.

[0144] Based on the aforementioned fuzzy entropy principle, the correspondence between the complexity of the comprehensive assessment results and the fuzzy entropy E is defined, with values ​​of obvious, relatively obvious, fuzzy, relatively obvious, and obvious. When the calculation result is obvious or relatively obvious, it indicates that the differences in the classification of various assessment indicators are small, and the complexity of the heating pipeline risk assessment results is low; conversely, when the calculation result is fuzzy, it indicates that the differences in the classification of various assessment indicators are large, and the complexity of the heating pipeline risk assessment results is high. The X-conditional cloud model, comprehensively considering the randomness and fuzziness of heating pipeline risk assessment, uses fuzzy entropy as an auxiliary evaluation parameter to analyze the complexity of the comprehensive assessment results, thus realizing an improvement in the two-dimensional evaluation mode of heating pipeline risk assessment.

[0145] In this embodiment, determining the subjective weight W1 using fuzzy hierarchical analysis includes:

[0146] Construct a fuzzy consistency matrix of risk assessment indicators for heating pipelines:

[0147] Construct a fuzzy complementary judgment matrix C for the risk assessment index system of heating pipelines, and obtain r by summing its rows. i , for r i Process to obtain r ij To ensure it meets the requirements of fuzzy consistency matrix elements, the calculation process for fuzzy consistency processing is as follows: n is the number of evaluation indicators;

[0148] According to r ij A fuzzy, consistent judgment matrix for the risk assessment indicators of heating pipelines is formed, represented as follows:

[0149]

[0150] The subjective weight of the i-th heating pipeline risk assessment indicator is calculated and expressed as follows: To sum the rows of the fuzzy consistency judgment matrix R;

[0151] Consistency check: The weight feature matrix is ​​calculated as follows: w ij =w i / (w i +w j ); i,j=1,2,…,n; calculate the compatibility index between the weight feature matrix and the set fuzzy consistency judgment matrix, expressed as: The closer I is to zero, the stronger the consistency of the judgment matrix;

[0152] The weight vector W1 = [w] passed the consistency test 11 ,w 12 ,…,w 1n [This refers to the subjective weights of each risk assessment indicator;]

[0153] The determination of the objective weight W2 using the coefficient of variation method includes:

[0154] The sample matrix B = (b...) is constructed using sample data of n evaluation indicators from m groups of evaluation objects. ij ) m×n b ij Let be the measured value of the j-th evaluation indicator for the i-th evaluation subject; calculate the mean of each indicator. Standard deviation

[0155] The weight value of the j-th heating pipeline risk assessment indicator is calculated and expressed as:

[0156] Obtain the weight vector W2 = [w 21 ,w 22 ,…,w 2n [This refers to the objective weights of each risk assessment indicator;]

[0157] The method of determining the combined weight W of each evaluation indicator by combining the subjective and objective weights of each indicator using the least squares comprehensive weighting method includes:

[0158] Set the combination weights W = [w1, w2, ..., w n The optimization objective function is:

[0159]

[0160]

[0161] The objective function is solved using the Lagrange method. The Lagrange multiplier λ is used to apply equality constraints to the objective function, and the Lagrange function is constructed as follows:

[0162]

[0163] Establish parameters for the Lagrange function with respect to w. i The first-order partial derivatives of λ and λ, set to zero, are used to calculate the extremum of the objective function:

[0164]

[0165] Solving the above formula yields the combined weights W = [w1, w2, ..., w...] for each evaluation indicator. n ].

[0166] It should be noted that the traditional analytic hierarchy process (AHP) lacks clear principles in constructing the judgment matrix, often resulting in inconsistencies in the selection of scales between indicators. Therefore, the judgment matrix needs to undergo consistency checks to prove its effectiveness. This process involves continuous adjustments to the judgment matrix and repeated calculations of the largest eigenvalue of higher-order matrices, making the overall calculation complex. The FAHP method employs fuzzy consistency in constructing the judgment matrix to minimize subsequent adjustments. The coefficient of variation (COP) method is a commonly used objective weighting method, assigning different weights based on the degree of fluctuation of all sample values ​​for each indicator around the mean. Indicators with smaller differences between sample values ​​are considered to have a smaller impact on the overall evaluation and are therefore assigned smaller weights, while indicators with larger differences are assigned larger weights. The FAHP method, which involves subjective weighting, relies on practical experience to set the judgment matrix and calculate subjective weights, while the COP method, which involves objective weighting, relies on the mean and standard deviation of the indicator sample data to calculate objective weights. Both methods have their advantages, but also have inherent limitations. By using the least squares method to comprehensively process subjective and objective weights, taking into account both subjective experience and the patterns of sample data, a combined weight optimization model is proposed. Compared with the single weighting method, it can obtain more reasonable simplified weights for indicators.

[0167] Example 2

[0168] Figure 2 This is a schematic diagram of a laser ultrasound-based heating pipeline damage identification and risk assessment system involved in this invention.

[0169] like Figure 2 As shown in Embodiment 2, this embodiment provides a laser-ultrasound-based system for identifying and assessing the damage to heating pipelines. The system includes:

[0170] The digital twin model building unit is used to build a digital twin model of the heating pipeline using mechanistic modeling and data identification methods.

[0171] The damage simulation data acquisition unit is used to simulate the detection of defects and damage in the heating pipeline using laser ultrasound based on the digital twin model of the heating pipeline, and to acquire damage simulation data of the heating pipeline.

[0172] The actual damage data acquisition unit is used to acquire actual damage data of heating pipelines in the actual scenario of the heating pipeline to be identified by adding a laser ultrasonic identification device.

[0173] The damage quantitative identification unit is used to take the simulated damage data and the actual damage data of the heating pipeline as pipeline damage identification samples, perform sample denoising, and establish a quantitative identification model of heating pipeline damage using machine learning algorithms.

[0174] The damage risk assessment unit is used to obtain the predicted value of heating pipe damage based on the quantitative identification model of heating pipe damage, and to conduct heating pipe damage risk assessment using the fuzzy cloud theory method.

[0175] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can also be implemented in other ways. The system embodiments described above are merely illustrative; for example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0176] In addition, the functional modules in the various embodiments of the present invention can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0177] If the functionality is implemented as a software module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0178] Based on the above-described preferred embodiments of the present invention, and through the foregoing description, those skilled in the art can make various changes and modifications without departing from the inventive concept. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for damage identification and risk assessment of heating pipelines based on laser ultrasound, characterized in that, The method for identifying and assessing the damage to heating pipelines includes: A digital twin model of the heating pipeline was established using mechanistic modeling and data identification methods. Based on the digital twin model of the heating pipeline, laser ultrasound is used to detect defects and damage in the heating pipeline and obtain simulation data of the heating pipeline damage. In the actual scenario of the heating pipeline to be identified, a laser ultrasonic identification device is added to obtain actual data on the damage to the heating pipeline; The simulated damage data and actual damage data of the heating pipelines are used as pipeline damage identification samples. Sample denoising is performed, and a quantitative identification model for heating pipeline damage is established using machine learning algorithms, including: The simulated data of heating pipeline damage and the actual data of heating pipeline damage are used as pipeline damage identification samples. Sample data preprocessing is performed, including missing value processing, outlier processing and data standardization. The preprocessed samples are denoised using a combination of empirical mode decomposition and wavelet thresholding: the noisy sample signal is decomposed using empirical mode decomposition to obtain high-frequency IMF components, low-frequency IMF components and residual signals, and then the high-frequency IMF components are denoised using wavelet thresholding. Finally, the high-frequency IMF components, low-frequency IMF components and residual signals are reconstructed to obtain the denoised sample. After denoising, the laser ultrasonic signal is subjected to wavelet packet decomposition to extract relevant features. The wavelet packet energy features that are strongly correlated with the damage size of the heating pipeline are then extracted. The wavelet packet energy features and the time-domain and frequency-domain features of the laser ultrasonic signal are then used to construct a feature vector. After the feature vectors are input into the ELM model optimized by the Grey Wolf Optimization Algorithm for learning and training, a quantitative identification model for heating pipeline damage is established. Specifically, wavelet packet decomposition is used to extract relevant features from the laser-ultrasonic signal, mining the wavelet packet energy features that strongly correlate with the damage size of the heating pipeline. These wavelet packet energy features, along with the time-domain and frequency-domain features of the laser-ultrasonic signal, are then combined to construct a feature vector, including: The laser ultrasonic signal is decomposed by wavelet packet decomposition. The decomposed damage signal generates wavelet packet coefficients in the corresponding frequency bands. By utilizing the different energy distributions of the wavelet packet decomposed signal, the energy distribution characteristics of the laser ultrasonic signal in each frequency band can be obtained, which is the wavelet packet energy characteristic. The time difference between the peaks and troughs of the reflected echoes from damages of different sizes in the laser-ultrasound signal is used as the time-domain feature of the damage; the spectral energy of damages of different sizes is obtained by spectral transformation of the laser-ultrasound transmitted wave signal as the frequency-domain feature of the damage. The wavelet packet energy features, the damage time-domain features, and the damage frequency-domain features are collectively constructed into a feature vector; Furthermore, the heating pipeline damage identification and risk assessment method further includes: obtaining a laser ultrasonic pipeline damage visualization image through the laser ultrasonic identification device, and after denoising, enhancing, segmenting and extracting features from the image, establishing a qualitative identification model for heating pipeline damage by using a machine learning algorithm for learning and training. Based on the quantitative identification model for heating pipeline damage, predicted values ​​of heating pipeline damage are obtained, and a fuzzy cloud theory method is used to assess the risk of heating pipeline damage, including: Based on the quantitative identification model for heating pipeline damage, the predicted value of heating pipeline damage is obtained. The predicted value of heating pipeline damage is combined with the pre-calculated fatigue index of heating pipeline, geological index of pipeline location, and remaining life index of pipeline as the risk assessment index of heating pipeline. A risk assessment grading standard is formulated, and a risk assessment index system for heating pipeline is constructed, including Level 1 healthy state, Level 2 good state, Level 3 general state, Level 4 poor state, and Level 5 dangerous state. Determining Subjective Weights Using Fuzzy Hierarchical Analysis Then, the objective weights are determined using the coefficient of variation method. Then, the least squares comprehensive weighting method is used to combine the subjective and objective weights of each evaluation indicator to determine the combined weight of each evaluation indicator. ; Based on cloud model theory, cloud model parameters for each corresponding level are calculated according to the grading standards of various evaluation indicators. A forward cloud generator is then used to generate the subordinate clouds for each evaluation indicator. The cloud model parameters include the expected... ,entropy and hyperentropy ; Substitute the measured data to be evaluated into the X-conditional cloud generator to solve for the membership degree of each evaluation index at each level. Combined with weights Calculate the membership vector of the comprehensive evaluation level. Then, the risk assessment level S is determined based on the principle of maximum membership; the X-conditional cloud generator: when the quantitative value x is known, combined with cloud digital characteristics, according to Calculate cloud droplets ; ; Based on fuzzy entropy theory, the fuzzy entropy E of the comprehensive risk level assessment of heating pipelines is calculated, and the complexity of the comprehensive risk level assessment results is analyzed as an auxiliary parameter for the risk level evaluation, thus obtaining the final risk assessment result (S, E) of the heating pipelines; the fuzzy entropy... , ; The total number of grades, For the membership degree of level i in the risk assessment of heating pipelines, The coefficient is the standardized coefficient.

2. The method for identifying and assessing the damage to heating pipelines according to claim 1, characterized in that, The method of establishing a digital twin model of heating pipelines using mechanism modeling and data identification includes: Construct physical entity models, logical models, and simulation models of the heating pipeline; among them, The establishment of the logical model includes: establishing a controllable closed-loop logical model based on the logical mechanism relationship of the physical entity of the heating pipeline, and mapping the physical model to the logical model; The establishment of the simulation model includes: building a simulation model of the heating pipeline based on the collected operation data, status data, and physical property data of the heating pipeline; optimizing the parameters of the simulation model according to the error between the predicted value and the actual value output by the simulation model; the simulation model also includes a laser ultrasonic simulation unit, which analyzes the interaction process between the laser and the tested heating pipeline from the principle of laser-induced ultrasonic waves, and models and simulates the physical process of laser-induced ultrasonic waves. The physical entity model, logical model, and simulation model are fused together to construct a system-level digital twin model of the physical entity of the heating pipeline in virtual space. The real-time operating data of the heating pipeline under multiple operating conditions is input into the system-level digital twin model. The simulation results of the system-level digital twin model are adaptively identified and corrected using the reverse identification method to obtain the identified and corrected digital twin model of the heating pipeline.

3. The method for identifying and assessing the damage to heating pipelines according to claim 1, characterized in that, The laser ultrasonic identification device includes: a laser ultrasonic emitting unit, a laser ultrasonic signal acquisition and receiving unit, and a laser controller display unit. During damage identification of the heating pipe under test, the laser controller display unit synchronously controls the laser ultrasonic emitting unit and the laser ultrasonic signal acquisition and receiving unit: it controls the triggering of the laser in the laser ultrasonic emitting unit and the electric scanning mirror to focus and scan the laser source; it stores the signal acquired by the laser ultrasonic signal acquisition and receiving unit after filtering, amplification, and A / D conversion; and it visualizes the waveform data of the received ultrasonic signal through inversion. The laser ultrasonic signal acquisition and receiving unit adopts an indirect contact identification and detection method, using an ultrasonic probe to contact the surface of the heating pipe under test to achieve laser ultrasonic identification and detection, and using filters, amplifiers, and A / D converters to filter, amplify, and convert the signal.

4. The method for identifying and assessing the damage to heating pipelines according to claim 1, characterized in that, The method of using wavelet threshold denoising to denoise high-frequency IMF components includes: Select appropriate wavelet basis functions to perform wavelet transform on the noisy high-frequency IMF components; perform thresholding on the transformed wavelet coefficients, where coefficients below the threshold are generated by noise signals and coefficients above the threshold are generated by the original signal; discard wavelet coefficients below the threshold and retain coefficients above the threshold for signal denoising.

5. The method for damage identification and risk assessment of heating pipelines according to claim 1, characterized in that, The step of establishing a quantitative identification model for heating pipeline damage by inputting feature vectors into an ELM model optimized by the Grey Wolf optimization algorithm for learning and training includes: Set the initial parameters of the GWO (Grey Wolf) optimization algorithm, including the number of wolves, the maximum number of iterations, and the range of optimization parameter values; initialize. , , The location and objective function value of gray wolves at each level; The ELM model parameters are randomly generated, and the mean squared error is used as the objective function value. These parameters are compared with the objective function values ​​of gray wolves at each level, and the larger objective function value is used to replace the objective function. Continuously updated , , The positions and objective function values ​​of gray wolves at each level, and the position of the optimal gray wolf individual, are the best values ​​for the input layer weights and hidden layer thresholds of the ELM model; After training the ELM model by inputting the feature vectors into the optimized parameters, a quantitative identification model for heating pipeline damage is established.

6. The method for damage identification and risk assessment of heating pipelines according to claim 1, characterized in that, When denoising the image, a median denoising method based on wavelet threshold change is adopted, including: performing median filtering on the original image, performing wavelet layer transformation to obtain multiple sub-images, setting a threshold according to the generated coefficient matrix, performing median filtering on each sub-image again, and restoring the image through wavelet reconstruction operation according to the newly generated information matrix to obtain a denoised image. The image enhancement process employs a Retinex enhancement algorithm based on adaptive weights, comprising: applying multiple Retinex enhancement algorithms to the same image, and then weighting and summing the calculation results according to different weights to obtain the enhanced image, as shown below: , This is for the output of the Retinex enhancement algorithm's multiple channels i; The pixel value of channel i in the original image; It is a Gaussian wrapping function; To calculate the number of times the wrap function is used, i.e., the number of elements involved in the weighted summation; The weight of each element; When segmenting the image, the image segmentation algorithm used includes at least threshold-based image segmentation, edge-based image segmentation, region-based image segmentation, clustering-based segmentation, and neural network-based segmentation. When extracting features from an image, the eigenvalues ​​of the gray-level co-occurrence matrix are calculated using gray-level co-occurrence matrix analysis to describe the image features, including at least the second moment of the angle, contrast, entropy, inverse moment, variance, mean sum, variance sum, clustering shadow, and significant clustering. The establishment of a qualitative identification model for heating pipeline damage after learning and training using machine learning algorithms includes: The penalty function factor C and kernel width of the SVM model are optimized using the hybrid frog leaping optimization algorithm (SFLA). : Initialize the frog population, divide the n frogs in the population into n meme groups, calculate the individual fitness value and sort them, frog i (i=1,2,…,n) corresponds to the i-th meme group, and the (n+1)-th frog enters the frog population of meme group 1 for grouping. The update of the frog's jump step size and position in the meme group is represented as: ; ; This represents the distance the frog travels. and The frogs at the positions corresponding to the best and worst fitness in the current meme group; ; When the new solution is better, replace the worst individual; otherwise, use the frog in the best position. replace The optimal frog position is determined through iterative iteration, which is the penalty function factor C and kernel width of the SVM model. ; The extracted image features are input into the optimized SVM model for training, and a qualitative identification model for heating pipeline damage is established.

7. The method for damage identification and risk assessment of heating pipelines according to claim 1, characterized in that, The subjective weights are determined using fuzzy hierarchical analysis. ,include: Construct a fuzzy consistency matrix of risk assessment indicators for heating pipelines: Construct a fuzzy complementary judgment matrix C for the risk assessment index system of heating pipelines, and obtain it by summing its rows. ,right Processing to obtain To ensure it meets the requirements of fuzzy consistency matrix elements, the calculation process for fuzzy consistency processing is as follows: ; The number of evaluation indicators; according to A fuzzy, consistent judgment matrix for the risk assessment indicators of heating pipelines is formed, represented as follows: ； The subjective weight of the i-th heating pipeline risk assessment indicator is calculated and expressed as follows: ; To sum the rows of the fuzzy consistency judgment matrix R; Consistency check: The weight feature matrix is ​​calculated as follows: ; The compatibility index between the weighted feature matrix and the set fuzzy consistency judgment matrix is ​​calculated and expressed as: ; The closer the matrix is ​​to zero, the stronger the consistency of the judgment matrix. Weight vector that has passed the consistency test The subjective weights of each risk assessment indicator; The method of determining objective weights using the coefficient of variation is mentioned. ,include: A sample matrix is ​​constructed using sample data of n evaluation indicators from m groups of evaluation objects. ; Let be the measured value of the j-th evaluation indicator for the i-th evaluation subject; calculate the mean of each indicator. Standard deviation ; The weight value of the j-th heating pipeline risk assessment indicator is calculated and expressed as: ; Obtain the weight vector The objective weights of each risk assessment indicator; The least squares comprehensive weighting method is used to determine the combined weight of each evaluation indicator by integrating the subjective and objective weights of each indicator. ,include: Set combination weights The optimization objective function is: ； ； The objective function is solved using the Lagrange method, and the Lagrange multipliers are defined. By applying equality constraints to the objective function, we construct the Lagrangian function, which is expressed as: ； Establish the relationship between the Lagrange function and the Lagrange function. and The first-order partial derivatives are used, and the extreme values ​​of the objective function are calculated by setting the partial derivatives to zero: ； Solving the above formula yields the combined weights of each evaluation indicator. .

8. A laser-ultrasound-based system for identifying and assessing the damage and risks of heating pipelines, characterized in that, The heating pipeline damage identification and risk assessment system includes: The digital twin model building unit is used to build a digital twin model of the heating pipeline using mechanistic modeling and data identification methods. The damage simulation data acquisition unit is used to simulate the detection of defects and damage in the heating pipeline using laser ultrasound based on the digital twin model of the heating pipeline, and to acquire damage simulation data of the heating pipeline. The actual damage data acquisition unit is used to acquire actual damage data of heating pipelines in the actual scenario of the heating pipeline to be identified by adding a laser ultrasonic identification device. The damage quantitative identification unit is used to take the simulated damage data and the actual damage data of the heating pipeline as pipeline damage identification samples, perform sample denoising, and establish a quantitative identification model for heating pipeline damage using machine learning algorithms, including: The simulated data of heating pipeline damage and the actual data of heating pipeline damage are used as pipeline damage identification samples. Sample data preprocessing is performed, including missing value processing, outlier processing and data standardization. The preprocessed samples are denoised using a combination of empirical mode decomposition and wavelet thresholding: the noisy sample signal is decomposed using empirical mode decomposition to obtain high-frequency IMF components, low-frequency IMF components and residual signals, and then the high-frequency IMF components are denoised using wavelet thresholding. Finally, the high-frequency IMF components, low-frequency IMF components and residual signals are reconstructed to obtain the denoised sample. After denoising, the laser ultrasonic signal is subjected to wavelet packet decomposition to extract relevant features. The wavelet packet energy features that are strongly correlated with the damage size of the heating pipeline are then extracted. The wavelet packet energy features and the time-domain and frequency-domain features of the laser ultrasonic signal are then used to construct a feature vector. After the feature vectors are input into the ELM model optimized by the Grey Wolf Optimization Algorithm for learning and training, a quantitative identification model for heating pipeline damage is established. Specifically, wavelet packet decomposition is used to extract relevant features from the laser-ultrasonic signal, mining the wavelet packet energy features that strongly correlate with the damage size of the heating pipeline. These wavelet packet energy features, along with the time-domain and frequency-domain features of the laser-ultrasonic signal, are then combined to construct a feature vector, including: The laser ultrasonic signal is decomposed by wavelet packet decomposition. The decomposed damage signal generates wavelet packet coefficients in the corresponding frequency bands. By utilizing the different energy distributions of the wavelet packet decomposed signal, the energy distribution characteristics of the laser ultrasonic signal in each frequency band can be obtained, which is the wavelet packet energy characteristic. The time difference between the peaks and troughs of the reflected echoes from damages of different sizes in the laser-ultrasound signal is used as the time-domain feature of the damage; the spectral energy of damages of different sizes is obtained by spectral transformation of the laser-ultrasound transmitted wave signal as the frequency-domain feature of the damage. The wavelet packet energy features, the damage time-domain features, and the damage frequency-domain features are collectively constructed into a feature vector; Furthermore, the heating pipeline damage identification and risk assessment system also includes: obtaining a laser ultrasonic pipeline damage visualization image through the laser ultrasonic identification device, and after denoising, enhancing, segmenting and extracting features from the image, establishing a qualitative identification model for heating pipeline damage by using a machine learning algorithm for learning and training. The damage risk assessment unit is used to obtain predicted values ​​of heating pipe damage based on the quantitative identification model of heating pipe damage, and to conduct a damage risk assessment of the heating pipe using the fuzzy cloud theory method, including: Based on the quantitative identification model for heating pipeline damage, the predicted value of heating pipeline damage is obtained. The predicted value of heating pipeline damage is combined with the pre-calculated fatigue index of heating pipeline, geological index of pipeline location, and remaining life index of pipeline as the risk assessment index of heating pipeline. A risk assessment grading standard is formulated, and a risk assessment index system for heating pipeline is constructed, including Level 1 healthy state, Level 2 good state, Level 3 general state, Level 4 poor state, and Level 5 dangerous state. Determining Subjective Weights Using Fuzzy Hierarchical Analysis Then, the objective weights are determined using the coefficient of variation method. Then, the least squares comprehensive weighting method is used to combine the subjective and objective weights of each evaluation indicator to determine the combined weight of each evaluation indicator. ; Based on cloud model theory, cloud model parameters for each corresponding level are calculated according to the grading standards of various evaluation indicators. A forward cloud generator is then used to generate the subordinate clouds for each evaluation indicator. The cloud model parameters include the expected... ,entropy and hyperentropy ; Substitute the measured data to be evaluated into the X-conditional cloud generator to solve for the membership degree of each evaluation index at each level. Combined with weights Calculate the membership vector of the comprehensive evaluation level. Then, the risk assessment level S is determined based on the principle of maximum membership; the X-conditional cloud generator: when the quantitative value x is known, combined with cloud digital characteristics, according to Calculate cloud droplets ; ; Based on fuzzy entropy theory, the fuzzy entropy E of the comprehensive risk level assessment of heating pipelines is calculated, and the complexity of the comprehensive risk level assessment results is analyzed as an auxiliary parameter for the risk level evaluation, thus obtaining the final risk assessment result (S, E) of the heating pipelines; the fuzzy entropy... , ; The total number of grades, For the membership degree of level i in the risk assessment of heating pipelines, The coefficient is the standardized coefficient.

Images