A method and system for defect early warning of cryogenic pipeline welds

Abstract

The application provides a defect early warning method and system for a low-temperature pipeline weld. Applied to the technical field of low-temperature pipeline weld detection, the method comprises collecting a multi-modal signal of a pipeline weld area, determining a correction coefficient of acoustic propagation characteristics if there is low-temperature drift, and compensating and correcting a synchronously collected ultrasonic A-scan signal; determining whether there is a defect echo based on the corrected signal; if there is a defect echo, determining a current characteristic state vector of the defect echo; inputting the current characteristic state vector into a model; determining whether the output result simultaneously satisfies a data fitting constraint and a physical law constraint; if the output result satisfies the constraint, determining an evolution track of the defect echo in a future service period; determining a real-time residual pressure-bearing strength of the pipeline based on the evolution track; and determining a safety warning level in combination with a change trend of a defect expansion rate. Thus, the efficiency of defect early warning of a low-temperature pipeline weld is improved.

Description

Technical Field

[0001] This invention relates to the field of cryogenic pipeline weld inspection technology, and in particular to a defect early warning method and system for cryogenic pipeline welds. Background Technology

[0002] Cryogenic pipelines are widely used in energy, chemical, and aerospace industries. Welds, as weak points in the structure, are prone to defects such as cracks and incomplete fusion under cryogenic conditions due to thermal stress and material embrittlement. These defects can dynamically expand with fluctuations in operating conditions, easily leading to leaks, explosions, and other safety accidents, causing significant losses. Developing early warning systems for cryogenic pipeline weld defects allows for real-time perception of defect evolution patterns, early prediction of failure risks, and overcomes the limitations of traditional post-incident inspections. This shift from passive maintenance to proactive prevention ensures the long-term safe and stable operation of cryogenic pipeline systems while reducing maintenance costs, providing crucial technical support for safety production and risk management in related industrial sectors.

[0003] Currently, existing technologies for early warning of weld defects in low-temperature pipelines do not take into account the characteristics of low-temperature operating conditions and mostly use general detection algorithms, resulting in low accuracy in identifying early characteristics of stress concentration and cold cracks in welds at low temperatures. Furthermore, data acquisition and analysis are disconnected, lacking a real-time linkage processing mechanism, leading to significant lag and low early warning efficiency.

[0004] Therefore, there is an urgent need for a defect early warning method and system for cryogenic pipeline welds with high early warning efficiency. Summary of the Invention

[0005] To address the problems existing in the prior art, the present invention provides a method and system for early warning of defects in low-temperature pipeline welds.

[0006] A defect early warning method for cryogenic pipeline welds, comprising: S1, acquire multimodal signals from the pipe weld area, determine whether there is low-temperature drift in the real-time temperature field data relative to the preset reference temperature, if so, determine the correction coefficient of the acoustic propagation characteristics to compensate and correct the synchronously acquired ultrasonic A-scan signal; determine whether there is a defect echo based on the corrected signal. S2, if the defect echo exists, determine the current characteristic state vector of the defect echo; input the current characteristic state vector into the physical information neural network model; determine whether the output of the physical information neural network model simultaneously satisfies the data fitting constraint and the physical law constraint; if satisfied, determine the evolution trajectory and remaining service life of the defect echo in the future service cycle. S3, determine the real-time remaining pressure bearing capacity of the pipeline based on the evolution trajectory; determine the safety warning level based on the real-time remaining pressure bearing capacity, the safety design threshold, and the trend of defect propagation rate; generate corresponding graded control instructions based on the determined safety warning level.

[0007] Furthermore, S1 includes: Calculate the absolute value of the temperature difference between the real-time temperature field data and the preset reference temperature; If the absolute value of the temperature difference is greater than the preset acoustic influence threshold, then the low temperature drift is determined to exist. Based on the absolute value of the temperature difference and the thermo-acoustic coupling equation of the material, the sound velocity compensation amount and attenuation compensation factor are determined. The flight time of the ultrasonic A-scan signal is corrected by using the sound velocity compensation amount, and the amplitude of the ultrasonic A-scan signal is corrected by using the attenuation compensation factor, so as to eliminate the deviation of low temperature environment in defect location and quantification.

[0008] Further, in S2, determining the current characteristic state vector of the defect echo and inputting the current characteristic state vector into the physical information neural network model includes: Extract the energy amplitude and geometric feature dimensions of the defect echo; Obtain the micro-strain monitoring value of the pipeline weld area and calculate the dynamic change rate of the geometric feature size under the action of the strain field; The energy amplitude, geometric feature size, and dynamic change rate are fused to determine the current feature state vector; When the current feature state vector is input into the physical information neural network model, the real-time temperature field data and the low-temperature toughness index of the pipeline base material are input as environmental boundary parameters to constrain the prediction space of the model.

[0009] Furthermore, the determination in S2 of whether the output of the physical information neural network model simultaneously satisfies both data fitting constraints and physical law constraints includes: Compute the data-driven loss between the predicted values ​​of the physical information neural network model and historical measured data; Calculation of physical constraint loss based on fracture mechanics theory; Determine whether the data-driven loss is less than a preset data convergence threshold and whether the physical constraint loss is less than a preset physical consistency threshold; If both of the above conditions are met, then the output of the physical information neural network model is determined to be valid. The calculation of the physical constraint loss is based on the Paris fatigue crack propagation formula, which determines whether the propagation rate predicted by the physical information neural network model matches the theoretically calculated propagation rate.

[0010] Furthermore, in S3, the safety warning level is determined based on the real-time remaining bearing capacity, the safety design threshold, and the trend of defect propagation rate changes, including: Based on the predicted defect depth and length in the evolution trajectory, the effective load-bearing cross-sectional area of ​​the pipe weld region is determined; The real-time residual bearing strength is calculated based on the effective bearing cross-sectional area and the yield strength of the material. The ratio of the real-time remaining pressure bearing capacity to the pipeline design pressure is calculated and denoted as the strength safety factor. The strength safety factor is compared with a first safety design threshold and a second safety design threshold, respectively, wherein the first safety design threshold is greater than the second safety design threshold. Further, determining the safety warning level based on the trend of defect propagation rate includes: If the strength safety factor is greater than the first safety design threshold, and the trend of the defect propagation rate is stable fluctuation, then the safety warning level is determined to be a Level 1 observation warning; based on the Level 1 observation warning, an instruction to maintain the current operating status but be included in the periodic inspection plan is generated; If the strength safety factor is between the second safety design threshold and the first safety design threshold, or if the defect propagation rate shows an accelerating trend, then the safety warning level is determined to be a Level 2 attention warning; based on the Level 2 attention warning, instructions to shorten the detection cycle and prepare maintenance plans are generated.

[0011] Furthermore, the method of determining the security warning level based on the changing trend of the defect propagation rate also includes: If the strength safety factor is less than the second safety design threshold, the safety warning level is determined to be a Level 3 action warning; based on the Level 3 action warning, an automatic maintenance work order is generated and a voltage reduction operation prompt is triggered; Determine whether the evolution trajectory indicates that the defect will become unstable and expand within a preset time, or whether the real-time remaining pressure bearing capacity is close to the burst pressure limit; if either determination condition is met, then determine the safety warning level as a level four emergency warning; based on the level four emergency warning, generate a command to link the emergency shut-off valve and send the highest level alarm.

[0012] A defect early warning system for cryogenic pipeline welds, comprising: The edge computing gateway is used to perform signal acquisition and low temperature drift determination in S1. The edge computing gateway has a built-in data buffer module to determine whether the communication connection with the upper-level server is interrupted. If interrupted, the acquired data is temporarily stored and resumed after the connection is restored. The central analysis server is used to perform model predictions in S2 and early warning level determinations in S3. The early warning execution terminal is used to receive and execute the hierarchical control instructions; when it receives the instruction corresponding to the fourth-level emergency early warning, the early warning execution terminal directly locks the pipeline operation permission and starts the emergency procedure.

[0013] Furthermore, the central analysis server is also configured with a dynamic adaptive optimization module for: Obtain actual defect monitoring data and historical prediction results; Calculate the prediction deviation between the two; Determine whether the prediction deviation exceeds a preset tolerance range; If the tolerance range is exceeded, the weight parameters of the physical information neural network model are updated in reverse based on the actual defect monitoring data to reduce the physical constraint loss in the subsequent prediction process.

[0014] This invention achieves accurate identification of defect echoes by acquiring multimodal signals and correcting ultrasonic A-scan signals based on low-temperature drift correction coefficients, thereby improving the efficiency of initial weld defect identification in low-temperature environments. By inputting the defect echo characteristic state vector into a physical information neural network model and combining it with the dual-constraint judgment output, it enables rapid and accurate prediction of defect evolution trajectories and remaining service life, improving the efficiency of weld defect development trend analysis. Furthermore, by combining real-time remaining pressure bearing capacity, safety design threshold, and defect propagation rate trend to determine the early warning level and generate graded control commands, it achieves targeted graded early warning and handling of low-temperature pipeline weld defects, improving the efficiency of weld defect early warning.

[0015] It should be understood that the description in the Summary of the Invention is not intended to limit the key or essential features of the embodiments of the present invention, nor is it intended to restrict the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0016] The above and other features, advantages, and aspects of the various embodiments of the present invention will become more apparent from the accompanying drawings and the following detailed description. The drawings are provided for a better understanding of the invention and are not intended to limit the scope of the invention. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein: Figure 1 A flowchart of a defect early warning method for cryogenic pipeline welds according to an embodiment of the present invention is shown; Figure 2 A block diagram of a defect early warning system for cryogenic pipeline welds according to an embodiment of the present invention is shown. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments 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, and 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.

[0018] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0019] Figure 1 A flowchart of a defect early warning method for cryogenic pipeline welds according to an embodiment of the present invention is shown, the method comprising: S1, acquire multimodal signals from the pipe weld area, determine whether there is low-temperature drift in the real-time temperature field data relative to the preset reference temperature, if so, determine the correction coefficient of the acoustic propagation characteristics to compensate and correct the synchronously acquired ultrasonic A-scan signal; determine whether there is a defect echo based on the corrected signal. S2, if the defect echo exists, determine the current characteristic state vector of the defect echo; input the current characteristic state vector into the physical information neural network model; determine whether the output of the physical information neural network model simultaneously satisfies the data fitting constraint and the physical law constraint; if satisfied, determine the evolution trajectory and remaining service life of the defect echo in the future service cycle. S3, determine the real-time remaining pressure bearing capacity of the pipeline based on the evolution trajectory; determine the safety warning level based on the real-time remaining pressure bearing capacity, the safety design threshold, and the trend of defect propagation rate; generate corresponding graded control instructions based on the determined safety warning level.

[0020] In some embodiments, S1 includes: calculating the absolute value of the temperature difference between the real-time temperature field data and the preset reference temperature; if the absolute value of the temperature difference is greater than a preset acoustic influence threshold, then determining that the low-temperature drift exists; determining the sound velocity compensation amount and the attenuation compensation factor based on the absolute value of the temperature difference and the thermoacoustic coupling equation of the material; using the sound velocity compensation amount to correct the flight time of the ultrasonic A-scan signal and using the attenuation compensation factor to correct the amplitude of the ultrasonic A-scan signal, so as to eliminate the deviation of the low-temperature environment in defect location and quantification. According to the embodiments of the present invention, by calculating the absolute value of the temperature difference and comparing it with the acoustic influence threshold, the accurate determination of the low-temperature drift state is achieved, improving the sensitivity of environmental perception; by determining the sound velocity compensation amount and the attenuation compensation factor based on the thermoacoustic coupling equation, the accurate mapping between temperature and acoustic parameters is achieved, improving the scientific nature of the compensation model; by correcting the flight time and amplitude of the ultrasonic A-scan signal, the deviation of defect location and quantification caused by the low-temperature environment is eliminated, improving the accuracy and reliability of defect detection.

[0021] For example, the preset reference temperature of an LNG pipeline made of 9% Ni steel is set to -162℃. The real-time temperature field data collected at the weld is -168.5℃, and the calculated absolute temperature difference is 6.5℃. Since a temperature difference exceeding 5℃ will cause the sound velocity in 9% Ni steel to change by more than 0.1%, the preset acoustic influence threshold is set to 5.0℃. Since 6.5℃ > 5.0℃, low-temperature drift is determined to exist. Based on the thermoacoustic coupling equation of the material, it is calculated that the longitudinal wave sound velocity increases by 12m / s under this temperature difference, that is, the sound velocity compensation is +12m / s, and the ultrasonic attenuation coefficient increases by 0.8dB / m, that is, the attenuation compensation factor corresponds to a gain compensation of 0.8dB / m. When processing the original A-scan signal, the flight time of one echo is corrected from 30.5μm to 30.42μm, and the echo amplitude is corrected from 32% full screen height compensation to 35% full screen height.

[0022] In some embodiments, determining the current characteristic state vector of the defect echo in S2 and inputting the current characteristic state vector into the physical information neural network model includes: extracting the energy amplitude and geometric feature size of the defect echo; obtaining the micro-strain monitoring value of the pipe weld area and calculating the dynamic change rate of the geometric feature size under the action of the strain field; fusing the energy amplitude, geometric feature size and dynamic change rate to determine the current characteristic state vector; when inputting the current characteristic state vector into the physical information neural network model, the real-time temperature field data and the low-temperature toughness index of the pipe base material are input as environmental boundary parameters to constrain the prediction space of the model. According to embodiments of the present invention, by calculating the dynamic rate of change of geometric feature dimensions under strain field, the dynamic response of defects under stress is captured, improving the dynamics of defect state tracking; by fusing energy amplitude, geometric feature dimensions, and dynamic rate of change to determine feature state vectors, a multi-dimensional comprehensive characterization of defect information is achieved, improving the completeness of feature description; by using real-time temperature field data and low-temperature toughness indicators as environmental boundary parameters, physical constraints are imposed on the prediction space of the physical information neural network model, improving the physical interpretability and accuracy of model prediction.

[0023] For example, the energy amplitude of the defect echo extracted from the corrected A-scan signal is 35% FSH, and the length of the defect in terms of geometry is measured to be 4.2 mm using the 6 dB method; the microstrain monitoring value of the weld area is 450 με, and under continuous pressure fluctuations, the dynamic change rate of the crack length is calculated to be 0.05 μm / s, indicating that the crack is in an active opening state at the micro level; the energy amplitude of 35%, the geometric size of 4.2 mm, and the dynamic change rate of 0.05 μm / s are fused to construct the current characteristic state vector V. t = [0.35, 4.2, 0.05]; When inputting the physical information neural network model, simultaneously input the real-time temperature of -168.5℃ and the fracture toughness of 9% Ni steel at this temperature as 165 MPa·m. 0.5 These environmental parameters limit the model from predicting physically impossible instantaneous brittle fracture results, ensuring that the prediction results are within the range allowed by the material's toughness.

[0024] In some embodiments, determining whether the output of the physical information neural network model simultaneously satisfies data fitting constraints and physical law constraints in step S2 includes: calculating the data-driven loss between the predicted value of the physical information neural network model and historical measured data; calculating the physical constraint loss based on fracture mechanics theory; determining whether the data-driven loss is less than a preset data convergence threshold and whether the physical constraint loss is less than a preset physical consistency threshold; if both of the above determination conditions are met, the output of the physical information neural network model is determined to be valid; wherein, the calculation of the physical constraint loss is based on the Paris fatigue crack propagation formula, and determining whether the propagation rate predicted by the physical information neural network model matches the theoretically calculated propagation rate. According to the embodiments of the present invention, by calculating the data-driven loss and the physical constraint loss based on fracture mechanics respectively, model evaluation from both data fitting and physical law dimensions is achieved, improving the comprehensiveness of model evaluation; by setting dual threshold determination conditions, strict screening of model output results is achieved, ensuring that they simultaneously satisfy data accuracy and physical consistency, improving the credibility of prediction results; by calculating the physical constraint loss based on the Paris fatigue crack propagation formula, the evaluation process of embedding fracture mechanics theory into the neural network is realized, improving the theoretical support and reliability of defect propagation prediction.

[0025] For example, the physical information neural network model output predicted that the length of the defect would reach 4.28 mm in the next 24 hours; the mean square error of the model fitting the past 10 sampling points was 1.2 × 10⁻⁶. -4 The data convergence threshold is less than the preset threshold of 5.0 × 10⁻⁶. -4 This indicates a good data fit; simultaneously, based on the Paris formula da / dN = C(∆K) m Calculate the theoretical propagation rate, assuming the stress intensity factor amplitude ∆K = 15 MPa·m 0.5 The theoretically calculated expansion rate is 3.5 × 10⁻⁶. -5 mm / cycle; the equivalent rate derived from the model output is 3.8 × 10⁻⁶. -5 mm / cycle; the difference in physical constraint loss between the two is calculated to be 0.3 × 10⁻⁶. -5 It is less than the preset physical consistency threshold of 1.0 × 10⁻⁶. -5 Since both conditions are met simultaneously, the model's prediction that the crack will extend to 4.28 mm in the next 24 hours is valid and can be used as a basis for early warning.

[0026] In some embodiments, determining the safety warning level in S3 based on the real-time remaining bearing capacity, the safety design threshold, and the trend of defect propagation rate includes: determining the effective bearing cross-sectional area of ​​the pipeline weld region based on the predicted defect depth and length in the evolution trajectory; calculating the real-time remaining bearing capacity based on the effective bearing cross-sectional area and the yield strength of the material; calculating the ratio of the real-time remaining bearing capacity to the pipeline design pressure, denoted as the strength safety factor; and comparing the strength safety factor with a first safety design threshold and a second safety design threshold, wherein the first safety design threshold is greater than the second safety design threshold. According to embodiments of the present invention, by determining the effective bearing cross-sectional area based on the predicted defect size, a quantitative characterization of the structural weakening effect caused by defects is achieved, improving the accuracy of structural assessment; by calculating the ratio of the real-time remaining bearing capacity to the design pressure to obtain the strength safety factor, a standardized measurement of pipeline safety margin is achieved, improving the intuitiveness and comparability of safety assessment; and by comparing the strength safety factor with the graded safety design threshold, a quantitative judgment basis is provided for subsequent graded warnings, improving the logic and operability of the warning system.

[0027] For example, based on the evolution trajectory predicted by the model, for a pipe with a current defect depth of 2.5 mm, a length of 4.2 mm, and a wall thickness of 20 mm, the effective load-bearing cross-sectional area remaining rate of the weld area is calculated to be 87.5%. Given that the yield strength of 9% Ni steel at low temperatures is 680 MPa, the real-time remaining bearing capacity of the defective part is calculated to be 680 × 87.5% = 595 MPa, corresponding to an allowable internal pressure limit of approximately 23.8 MPa. The design operating pressure of the pipe is 10 MPa; the strength safety factor S = 23.8 / 10 = 2.38; setting the first safety design threshold as 2.5 and the second safety design threshold as 1.8, the currently calculated 2.38 is compared with these two thresholds.

[0028] In some embodiments, S3 determines the safety warning level based on the real-time remaining pressure bearing capacity, the safety design threshold, and the trend of defect expansion rate, including: if the strength safety factor is greater than the first safety design threshold, and the trend of defect expansion rate is stable fluctuation, then the safety warning level is determined to be a Level 1 observation warning; based on the Level 1 observation warning, an instruction is generated to maintain the current operating state but include it in the regular inspection plan; if the strength safety factor is between the second safety design threshold and the first safety design threshold, or the trend of defect expansion rate shows an accelerating characteristic, then the safety warning level is determined to be a Level 2 attention warning; based on the Level 2 attention warning, an instruction is generated to shorten the detection cycle and prepare a maintenance plan. According to the embodiments of the present invention, by identifying a state with a high safety factor and stable expansion and confirming it as a Level 1 observation warning, over-maintenance can be avoided while maintaining necessary monitoring, thus improving the economy of operation and maintenance management; by identifying a state with a decreasing safety factor or accelerated expansion and confirming it as a Level 2 attention warning, early attention to potential risks and maintenance preparation can be achieved, thus improving the predictability and initiative of maintenance work; by generating differentiated control instructions for different warning levels, hierarchical management and resource optimization of pipeline risks can be achieved, thus improving the pertinence and efficiency of on-site handling.

[0029] For example, the current strength safety factor is 2.38, which is between the second safety design threshold of 1.8 and the first safety design threshold of 2.5. In addition, the previously calculated dynamic change rate shows that the crack has a slight expansion, but it is not an explosive growth. Although it has not fallen below the critical value of 1.8, it is already lower than the ideal value of 2.5, and the safety warning level is determined to be a level two warning. The generated control instruction is to automatically shorten the ultrasonic inspection cycle of the weld area from once a week to once every 24 hours, and send an instruction to the maintenance department to generate a Class B preparatory repair plan for the weld, requiring technicians to prepare the repair welding process procedure.

[0030] In some embodiments, determining the safety warning level in S3 based on the changing trend of the real-time remaining pressure bearing capacity, the safety design threshold, and the defect propagation rate further includes: if the strength safety factor is less than the second safety design threshold, then determining the safety warning level as a Level 3 action warning; based on the Level 3 action warning, generating an automatic maintenance work order and triggering a pressure reduction operation prompt; determining whether the evolution trajectory indicates that the defect will undergo unstable propagation within a preset short period of time, or whether the real-time remaining pressure bearing capacity is close to the burst pressure limit; if any of the determination conditions are met, then determining the safety warning level as a Level 4 emergency warning; based on the Level 4 emergency warning, generating an instruction to link an emergency shut-off valve and send the highest level alarm. According to embodiments of the present invention, by identifying a low safety factor state and confirming it as a Level 3 action warning, automatic maintenance and pressure reduction operations are triggered to promptly block the deterioration of risks and improve the timeliness of risk management; by determining that a defect is unstable and expanding or approaching the blast limit and confirming it as a Level 4 emergency warning, emergency shutdown and the highest alarm are triggered to effectively prevent catastrophic accidents and improve the system's ultimate safety assurance capability; by generating automatic maintenance work orders and linked emergency shutdown commands, automated rapid response in critical situations is achieved, reducing the lag of manual intervention and improving the response speed and reliability of emergency response.

[0031] For example, if the defect rapidly expands to a depth of 8mm after 72 hours, causing a decrease in the real-time remaining pressure bearing strength and the calculated strength safety factor to drop to 1.65, since 1.65 < the second safety design threshold of 1.8, a Level 3 action warning is directly triggered, automatically generating a maintenance work order and triggering an operation prompt suggesting immediate pressure reduction to 8.0 MPa. At the same time, the physical information neural network model predicts that, according to the current expansion rate, the stress intensity factor at the crack tip will exceed the material fracture toughness after 4 hours, which meets the preset condition for unstable expansion within a short period of time. In this case, the warning is immediately upgraded to a Level 4 emergency warning, issuing an instruction to shut down the emergency shut-off valve upstream of the pipeline and sending the highest-level red audible and visual alarm to the entire plant.

[0032] Figure 2 A block diagram of a defect early warning system for cryogenic pipeline welds according to an embodiment of the present invention is shown. The system includes: Edge computing gateway 201 is used to perform signal acquisition and low temperature drift determination in S1; the edge computing gateway has a built-in data buffer module to determine whether the communication connection with the upper-level server is interrupted. If interrupted, the acquired data is temporarily stored and resumed after the connection is restored. Central analysis server 202 is used to perform model predictions in S2 and early warning level determinations in S3; The early warning execution terminal 203 is used to receive and execute the hierarchical control instructions; when it receives the instruction corresponding to the fourth-level emergency early warning, the early warning execution terminal directly locks the pipeline operation permission and starts the emergency procedure.

[0033] In some embodiments, the central analysis server is further configured with a dynamic adaptive optimization module, used to: acquire actual defect monitoring data and prediction results at historical moments; calculate the prediction deviation between the two; determine whether the prediction deviation exceeds a preset tolerance range; if it exceeds the tolerance range, then update the weight parameters of the physical information neural network model in reverse based on the actual defect monitoring data to reduce physical constraint loss in subsequent prediction processes.

[0034] It should be understood that the various processes described above can be used to rearrange, add, or delete steps. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0035] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A defect early warning method for cryogenic pipeline welds, characterized in that, include: S1, acquire multimodal signals from the pipe weld area, determine whether there is low-temperature drift in the real-time temperature field data relative to the preset reference temperature, if so, determine the correction coefficient of the acoustic propagation characteristics to compensate and correct the synchronously acquired ultrasonic A-scan signal; determine whether there is a defect echo based on the corrected signal. S2, If the defect echo exists, determine the current characteristic state vector of the defect echo; The current feature state vector is input into the physical information neural network model; it is determined whether the output of the physical information neural network model simultaneously satisfies the data fitting constraints and physical law constraints; if so, the evolution trajectory and remaining service life of the defect echo in the future service cycle are determined. S3, determine the real-time remaining pressure bearing capacity of the pipeline based on the evolution trajectory; The safety warning level is determined based on the real-time remaining bearing capacity, the safety design threshold, and the changing trend of the defect propagation rate. Based on the determined safety warning level, corresponding hierarchical control instructions are generated.

2. The defect early warning method for cryogenic pipeline welds according to claim 1, characterized in that, S1 includes: Calculate the absolute value of the temperature difference between the real-time temperature field data and the preset reference temperature; If the absolute value of the temperature difference is greater than the preset acoustic influence threshold, then the low temperature drift is determined to exist. Based on the absolute value of the temperature difference and the thermo-acoustic coupling equation of the material, the sound velocity compensation amount and attenuation compensation factor are determined. The flight time of the ultrasonic A-scan signal is corrected by using the sound velocity compensation amount, and the amplitude of the ultrasonic A-scan signal is corrected by using the attenuation compensation factor, so as to eliminate the deviation of low temperature environment in defect location and quantification.

3. The defect early warning method for cryogenic pipeline welds according to claim 2, characterized in that, The current characteristic state vector of the defect echo is determined as described in S2; Inputting the current feature state vector into the physical information neural network model includes: Extract the energy amplitude and geometric feature dimensions of the defect echo; Obtain the micro-strain monitoring value of the pipeline weld area and calculate the dynamic change rate of the geometric feature size under the action of strain field; The energy amplitude, geometric feature size, and dynamic change rate are fused to determine the current feature state vector; When the current feature state vector is input into the physical information neural network model, the real-time temperature field data and the low-temperature toughness index of the pipeline base material are input as environmental boundary parameters to constrain the prediction space of the model.

4. The defect early warning method for cryogenic pipeline welds according to claim 3, characterized in that, The determination in S2 of whether the output of the physical information neural network model simultaneously satisfies data fitting constraints and physical law constraints includes: Compute the data-driven loss between the predicted values ​​of the physical information neural network model and historical measured data; Calculation of physical constraint loss based on fracture mechanics theory; Determine whether the data-driven loss is less than a preset data convergence threshold and whether the physical constraint loss is less than a preset physical consistency threshold; If both of the above conditions are met, then the output of the physical information neural network model is determined to be valid. The calculation of the physical constraint loss is based on the Paris fatigue crack propagation formula, which determines whether the propagation rate predicted by the physical information neural network model matches the theoretically calculated propagation rate.

5. The defect early warning method for cryogenic pipeline welds according to claim 4, characterized in that, S3 determines the safety warning level based on the real-time remaining bearing capacity, the safety design threshold, and the trend of defect propagation rate, including: Based on the predicted defect depth and length in the evolution trajectory, the effective load-bearing cross-sectional area of ​​the pipe weld region is determined; The real-time residual bearing strength is calculated based on the effective bearing cross-sectional area and the yield strength of the material. The ratio of the real-time remaining pressure bearing capacity to the pipeline design pressure is calculated and denoted as the strength safety factor. The strength safety factor is compared with a first safety design threshold and a second safety design threshold, wherein the first safety design threshold is greater than the second safety design threshold.

6. The defect early warning method for cryogenic pipeline welds according to claim 5, characterized in that, S3 determines the safety warning level based on the real-time remaining bearing capacity, the safety design threshold, and the trend of defect propagation rate, including: If the strength safety factor is greater than the first safety design threshold, and the trend of the defect propagation rate is stable fluctuation, then the safety warning level is determined to be a Level 1 observation warning; based on the Level 1 observation warning, an instruction to maintain the current operating status but be included in the periodic inspection plan is generated; If the strength safety factor is between the second safety design threshold and the first safety design threshold, or if the defect propagation rate shows an accelerating trend, then the safety warning level is determined to be a Level 2 attention warning; based on the Level 2 attention warning, instructions to shorten the detection cycle and prepare maintenance plans are generated.

7. The defect early warning method for cryogenic pipeline welds according to claim 6, characterized in that, S3, which determines the safety warning level based on the real-time remaining bearing capacity, the safety design threshold, and the trend of defect propagation rate, also includes: If the strength safety factor is less than the second safety design threshold, the safety warning level is determined to be a Level 3 action warning; based on the Level 3 action warning, an automatic maintenance work order is generated and a voltage reduction operation prompt is triggered; Determine whether the evolution trajectory indicates that the defect will become unstable and expand within a preset time, or whether the real-time remaining pressure bearing capacity is close to the burst pressure limit; if either determination condition is met, then determine the safety warning level as a level four emergency warning; based on the level four emergency warning, generate a command to link the emergency shut-off valve and send the highest level alarm.

8. A system for performing the defect early warning method for cryogenic pipeline welds as described in claim 7, characterized in that, include: The edge computing gateway is used to perform signal acquisition and cryogenic drift determination in S1; The edge computing gateway has a built-in data buffer module to determine whether the communication connection with the upper-level server is interrupted. If interrupted, the collected data is temporarily stored and resumed after the connection is restored. The central analysis server is used to perform model predictions in S2 and early warning level determinations in S3. The early warning execution terminal is used to receive and execute the hierarchical control instructions; when it receives the instruction corresponding to the fourth-level emergency early warning, the early warning execution terminal directly locks the pipeline operation permission and starts the emergency procedure.

9. The system for defect early warning method of cryogenic pipeline welds according to claim 8, characterized in that, The central analysis server is also configured with a dynamic adaptive optimization module, used for: Obtain actual defect monitoring data and historical prediction results; Calculate the prediction deviation between the two; Determine whether the prediction deviation exceeds a preset tolerance range; If the tolerance range is exceeded, the weight parameters of the physical information neural network model are updated in reverse based on the actual defect monitoring data to reduce the physical constraint loss in the subsequent prediction process.

Images