A method for predicting the corrosion life of thermal pipeline networks

Abstract

This invention relates to a method for predicting the corrosion lifespan of thermal pipeline networks, belonging to the field of computer processing technology. The method includes the following steps: synchronously acquiring acoustic wave reflection signals from the thermal pipeline wall using a detection device, and forming three-dimensional acoustic field data based on the combination of multiple acquired acoustic wave reflection signals; constructing depth coordinates of the circumferential distribution of the thermal pipeline wall based on the pipe wall feedback signals; constructing a geometric topology structure on the inner surface of the thermal pipeline wall based on the three-dimensional acoustic field data to obtain a continuous thermal pipeline wall map; and constructing a multi-scale graph neural network model specific to corrosion and crack propagation in the thermal pipeline network based on the continuous thermal pipeline wall map, including: dividing the pipeline into several segments, using the average corrosion area, maximum corrosion depth, and axial corrosion diffusion probability of each segment as node features and edge weights. This invention provides predictions for the pipeline network's operation and maintenance lifespan.

Description

Technical Field

[0001] This invention relates to the field of computer processing technology, and in particular to a method for predicting the corrosion life of thermal pipeline networks. Background Technology

[0002] As the core carriers of urban energy, heating and gas pipelines transport high-temperature and high-pressure media over long periods. The inner walls of their carbon steel or alloy steel pipes are prone to localized corrosion defects due to electrochemical corrosion, erosion, and stress coupling. Therefore, regular inspections of heating pipelines are necessary for proactive maintenance to prevent damage caused by pipelines being unable to withstand high pressure, thus avoiding disruptions to normal heating needs. Traditional inspection methods include, but are not limited to, ultrasonic thickness measurement and radiographic testing.

[0003] Acoustic wave detection (such as the Chinese authorized patent, publication number CN106352243B, publication date 2018-06-26, disclosing a gas pipeline leak detection system based on acoustic wave method) has gradually become the mainstream method for pipeline corrosion monitoring due to its non-contact and long-distance propagation characteristics. However, existing acoustic wave analysis has two major drawbacks:

[0004] First, signal processing remains at the level of time-domain attenuation analysis, failing to fully utilize the correlation between frequency-domain features (such as the frequency shift and energy distribution of high-frequency reflection peaks) and the microscopic morphology of corrosion pits;

[0005] Secondly, ignoring the topological structure of the sound wave propagation path makes it impossible to identify the spatial continuity (such as chain corrosion along the weld) or isolation (such as corrosion depth) of the corrosion area, resulting in a high misjudgment rate for complex modes such as stress corrosion cracking.

[0006] Furthermore, current corrosion life prediction models are mostly based on machine learning or finite element simulation, but they suffer from significant scale separation: the lack of data interaction between microscopic and macroscopic models leads to a large deviation between the predicted results and the actual failure time. For example, traditional convolutional neural networks (CNNs) can only process two-dimensional corrosion images and cannot capture the depth gradient and neighborhood association features of corrosion pits in the three-dimensional space of the pipeline; while graph neural networks (GNNs) can model spatial relationships, they do not incorporate microscopic physical parameters such as the roughness of corrosion pit edges and lattice distortion.

[0007] In summary, existing technologies cannot simultaneously address the challenges in corrosion detection, such as a lack of topological awareness in space, difficulty in achieving dynamic evolution tracking in time, and an inability to integrate microscopic morphology and macroscopic distribution characteristics at the scale. Summary of the Invention

[0008] To address the aforementioned technical problems, the present invention provides a method for predicting the corrosion life of thermal pipeline networks, comprising the following steps:

[0009] The detection device synchronously collects acoustic wave reflection signals from the wall of the heat pipe, and combines multiple collected acoustic wave reflection signals to form three-dimensional sound field data. Based on the feedback signals from the pipe wall, the depth coordinates of the circumferential distribution of the heat pipe wall are constructed.

[0010] Based on the three-dimensional sound field data, a geometric topology is constructed on the inner surface of the heat pipe wall to obtain a continuous heat pipe wall atlas.

[0011] Based on the continuous atlas of thermal pipe walls, a multi-scale graph neural network model specific to corrosion and crack propagation in thermal pipe networks is constructed, including:

[0012] The pipeline is divided into several segments, and the mean corrosion area, maximum corrosion depth and corrosion diffusion probability along the axial direction of each segment are used as node features and edge weights.

[0013] Obtain the coverage area of ​​a single corrosion pit within the corresponding segment, and then take the center of the coverage area of ​​the single corrosion pit as the geometric center point. The geometric center point includes the following structural corrosion features:

[0014] Based on the set of acoustic reflection signals corresponding to the coverage area of ​​the single corrosion pit, the aspect ratio of the single corrosion pit is determined. The aspect ratio is used to distinguish between the coverage area of ​​the corrosion pit and the corrosion depth. When the aspect ratio is higher than a preset aspect ratio threshold, it indicates the ratio of corrosion coverage to the coverage area within the segment.

[0015] Preferably, the continuous thermal pipe wall atlas includes analyzing the coverage ratio of the corrosion area around the weld of the thermal pipe and the number of rust points in the covered area, and determining whether the number of rust points in the covered area is concentrated along the influence area of ​​the corrosion area around the weld. When the coverage ratio exceeds a preset coverage threshold, it is determined to be a stress corrosion feature.

[0016] Preferably, the construction of a multi-scale graph neural network model specific to corrosion and crack propagation in thermal pipelines to generate a joint feature vector includes, based on the joint feature vector, outputting a pipeline health assessment report containing corrosion level determination, remaining life prediction, and leakage risk warning.

[0017] Preferably, the pipeline health assessment report generates a corrosion level determination based on the following rules:

[0018] If the corrosion accumulation index in the circumferential weld zone exceeds the preset coverage threshold and the grain boundary distortion rate exceeds the preset distortion threshold, it is judged as high-risk stress corrosion.

[0019] If the corrosion is randomly distributed and the grain boundary distortion rate is lower than the seventh preset threshold, it is determined to be the area covered by uniform corrosion pits.

[0020] The assessment report combines the thermal diagram of the pipe wall corrosion with the three-dimensional reconstruction of the grain boundary structure to present the analysis results, and marks the cloud map of the remaining wall thickness distribution.

[0021] Preferably, the calculation of the concentration of the corrosion-affected area along the weld seam includes:

[0022] The three-dimensional acoustic field data of the weld area of ​​the heat pipe are segmented, and the proportion of the corrosion area coverage length to the total length of the weld area is calculated.

[0023] The topological connectivity of the corrosion boundary was analyzed using the Vietoris-Rips complex algorithm. When the number of connected components was 1 and the coverage ratio exceeded the preset coverage threshold, it was determined to be a stress corrosion cracking feature.

[0024] It also includes calculating the distortion rate of the grain boundary structure of the heat pipe wall using the phase offset of the acoustic wave scattering signal. If the distortion rate exceeds a preset distortion threshold, it is marked as a typical feature of intergranular coverage.

[0025] Preferably, when the axial diffusion probability of the corrosion pit coverage area in a series of segments exceeds a first preset threshold, it is determined to be continuous corrosion.

[0026] When the axial diffusion rate of the corrosion pit coverage area is lower than the second preset threshold and the number of connected components exceeds the third preset threshold, it is determined to be a local corrosion depth.

[0027] The edge weight is determined by the correlation between the minimum curvature similarity between corrosion zones and the intergranular distortion rate. If the correlation exceeds the fourth preset threshold, the stress corrosion determination is strengthened.

[0028] Preferably, the product of the number of corrosion pit coverage areas in consecutive segments and the edge weights in consecutive segments is calculated to amplify the corrosion impact of high-temperature zones or elbow sections.

[0029] The edge roughness of the corrosion pit coverage area is the spectral entropy value of the acoustic wave reflection signal at the corrosion boundary. When the spectral entropy value exceeds the fifth preset threshold, it is marked as an active corrosion lesion. When the spectral entropy value is lower than the sixth preset threshold, it indicates a stable corrosion morphology.

[0030] Preferably, the detection device includes a base and an extension fixedly disposed on the side wall of the forward end of the base. An acoustic wave sensor array is fixedly installed on the side of the extension facing the heat pipe. A walking wheel driven by a motor is fixedly installed on the base. The number of bases is not less than four, and a connecting part is fixedly installed between two adjacent bases.

[0031] The present invention has at least the following beneficial effects:

[0032] By constructing three-dimensional acoustic field data and reconstructing geometric topology, precise spatial perception of pipeline corrosion morphology was achieved. This method utilizes acoustic wave reflection signals to generate circumferential depth coordinates of the pipe wall, forming a continuous atlas of thermal pipe walls, overcoming the shortcomings of traditional two-dimensional detection methods in characterizing three-dimensional corrosion morphology. It can accurately identify the distribution patterns, depth gradients, and spatial correlation characteristics of corrosion pits, providing a reliable data foundation for subsequent corrosion mode discrimination.

[0033] Based on a feature extraction mechanism using a multi-scale graph neural network model, this study effectively integrates macroscopic corrosion distribution and microscopic morphology features. By segmenting the pipeline and using the mean corrosion area, maximum depth, and diffusion probability as node features, combined with microscopic parameters such as aspect ratio and edge roughness, multi-scale feature correlation analysis is achieved, significantly improving the identification accuracy of complex corrosion modes such as stress corrosion cracking and intergranular corrosion.

[0034] By employing a dynamic evolution tracking and risk warning mechanism, this method enables time-series prediction of corrosion development. Utilizing axial diffusion probability and boundary topological connectivity analysis, it can determine the continuous development trend of corrosion and quantify the degree of corrosion activity through spectral entropy. By combining multiple parameters such as grain boundary distortion rate and coverage ratio, an assessment report is generated that includes remaining lifetime prediction and leakage risk level. Attached Figure Description

[0035] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0036] Figure 1 Here is a flowchart of a method for predicting the corrosion life of a thermal pipeline network according to Embodiment 1 of the present invention;

[0037] Figure 2 This is a structural diagram provided in Embodiment 1 of the present invention;

[0038] Figure 3 This is a structural diagram of the acoustic wave sensing array and extension provided in Embodiment 1 of the present invention.

[0039] Explanation of reference numerals in the attached figures:

[0040] 1. Base; 2. Extension; 3. Acoustic wave sensor array; 4. Wheels; 5. Connecting part. Detailed Implementation

[0041] 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.

[0042] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0043] Example 1

[0044] This embodiment provides a method for predicting the corrosion life of a thermal pipeline network. The method includes the following steps: Figure 1 As shown:

[0045] The sound wave reflection signals of the heat pipe wall are collected synchronously by the detection device, and three-dimensional sound field data are formed by combining multiple collected sound wave reflection signals. The depth coordinates of the heat pipe wall circumferential distribution are constructed based on the pipe wall feedback signal.

[0046] Based on three-dimensional sound field data, a geometric topology is constructed on the inner surface of the heat pipe wall to obtain a continuous heat pipe wall atlas.

[0047] Based on continuous atlases of thermal pipe walls, a multi-scale graph neural network model specific to corrosion and crack propagation in thermal pipe networks is constructed, including:

[0048] The pipeline is divided into several segments, and the mean corrosion area, maximum corrosion depth and corrosion diffusion probability along the axial direction of each segment are used as node features and edge weights.

[0049] Obtain the coverage area of ​​a single corrosion pit within the corresponding segment, and then take the center of the coverage area of ​​a single corrosion pit as the geometric center point. The geometric center point includes the following structural corrosion features:

[0050] Based on the set of acoustic reflection signals corresponding to the coverage area of ​​a single corrosion pit, the depth-to-width ratio of the single corrosion pit coverage area is determined. The depth-to-width ratio is used to distinguish between the coverage area and corrosion depth of the corrosion pit. When the depth-to-width ratio is higher than the preset depth-to-width ratio threshold, it indicates the coverage ratio within the corrosion coverage segment.

[0051] Specifically, firstly, sound wave reflection signals are synchronously collected by a detection device and combined into three-dimensional sound field data. This data is then used to construct the depth coordinates of the circumferential distribution of the pipe wall, achieving data dimensionality upgrade from one-dimensional signals to three-dimensional space, thus solving the problem of "insufficient spatial topology perception capability" in the background. Secondly, a continuous atlas of thermal pipe walls is constructed based on the three-dimensional sound field data, mapping the physical pipe into a computable geometric topology model, providing structured input for subsequent graph neural networks. Finally, a multi-scale graph neural network model is constructed, dividing the pipe into several segments. The average corrosion area, maximum corrosion depth, and axial diffusion probability are used as node features and edge weights, and the aspect ratio of corrosion pits is introduced as a discrimination index to distinguish between uniform corrosion and localized corrosion. This initially achieves the fusion of macroscopic corrosion distribution and microscopic morphological features (aspect ratio), addressing the problem of "scale fragmentation" in the background.

[0052] Furthermore, the continuous thermal pipe wall atlas in the above embodiments includes analyzing the coverage ratio of the corrosion area around the weld of the thermal pipe and the number of rust points in the covered area. Based on whether the number of rust points in the covered area is concentrated along the influence area of ​​the corrosion area around the weld, the coverage ratio is determined to be stress corrosion characteristics when it exceeds a preset coverage threshold.

[0053] Specifically, by calculating the corrosion coverage ratio and the concentration of rust points, and combining the preset threshold to determine stress corrosion characteristics, the ability to identify stress corrosion cracking, a specific corrosion mode, is enhanced. The depth-to-width ratio discrimination is complementary: the depth-to-width ratio focuses on the morphology of a single corrosion pit, while this embodiment focuses on the spatial distribution pattern of regional corrosion (such as concentration along the weld). The two together improve the model's discrimination accuracy for complex corrosion modes.

[0054] Furthermore, the calculation of the concentration of the corrosion-affected zone around the weld includes:

[0055] The three-dimensional acoustic field data of the weld zone of the segmented heat pipe were used to calculate the proportion of the corrosion zone coverage length to the total length of the weld zone.

[0056] The topological connectivity of the corrosion boundary was analyzed using the Vietoris-Rips complex algorithm. When the number of connected components was 1 and the coverage ratio exceeded the preset coverage threshold, it was determined to be a stress corrosion cracking feature.

[0057] It also includes calculating the distortion rate of the grain boundary structure of the heat pipe wall using the phase offset of the acoustic wave scattering signal. If the distortion rate exceeds a preset distortion threshold, it is marked as a typical feature of intergranular coverage.

[0058] Specifically, it provides a concrete algorithm for calculating the concentration of the corrosion influence zone around the weld. By analyzing the topological connectivity of the corrosion boundary using the Vietoris-Rips complex algorithm, it combines the mathematical topological characteristics (number of connected components) of the corrosion morphology with the coverage ratio, providing a more rigorous computational geometric basis for stress corrosion assessment. Simultaneously, it adds a method for calculating the grain boundary distortion rate through the phase shift of acoustic signals, directly linking acoustic signals with changes in the material's microstructure, providing a concrete measurement means for the "grain boundary distortion rate," and demonstrating the bridging role of acoustic data between micro and macro scales.

[0059] Furthermore, the multi-scale graph neural network model in the above embodiments generates a joint feature vector, which outputs a pipeline health assessment report based on the joint feature vector, including corrosion level determination, remaining life prediction, and leakage risk warning.

[0060] The pipeline health assessment report generates a corrosion level determination based on the following rules:

[0061] If the corrosion accumulation index in the circumferential weld zone exceeds the preset coverage threshold and the grain boundary distortion rate exceeds the preset distortion threshold, it is judged as high-risk stress corrosion.

[0062] If the corrosion is randomly distributed and the grain boundary distortion rate is lower than the seventh preset threshold, it is determined to be the area covered by uniform corrosion pits.

[0063] The assessment report combines the thermal map of the pipe wall corrosion with the three-dimensional reconstruction map of the grain boundary structure to present the analysis results, and marks the cloud map of the remaining wall thickness distribution.

[0064] Specifically, the "corrosion accumulation index in the circumferential weld zone" and "grain boundary distortion rate" were introduced as criteria, and high-risk stress corrosion and uniform corrosion were distinguished. Furthermore, the assessment report was required to provide a visual representation combining thermal maps, 3D reconstructions of the grain boundary structure, and residual wall thickness cloud maps. In addition to stress corrosion identification, grain boundary distortion, a microscopic material characteristic, was added and combined with macroscopic corrosion thermal maps and wall thickness distribution.

[0065] Furthermore, in this embodiment, for the corrosion pit coverage area in consecutive pipe segments, when the axial diffusion probability of the corrosion pit coverage area exceeds a first preset threshold (when the axial diffusion probability of the corrosion pit coverage area in multiple consecutive pipe segments exceeds this threshold, it indicates that corrosion is continuously and stably expanding along the axial direction of the pipe. This is a very dangerous trend, meaning that long groove-like corrosion or cracks may have formed, which can easily lead to a sharp decrease in the overall strength of the pipe or a through-thrust leak. Setting a high threshold is to accurately capture this highly continuous corrosion pattern and avoid misjudging occasional adjacent sporadic corrosion as continuous corrosion. It is usually set between 70% and 90%, which is used as the lower limit of the axial diffusion probability for determining continuous corrosion), then it is determined to be continuous corrosion.

[0066] When the axial diffusion rate of the corrosion pit coverage area is lower than the second preset threshold (when the axial diffusion rate of the corrosion pit is lower than this threshold, it indicates that the current corrosion pit is isolated and does not show a significant tendency to extend to adjacent pipe sections. This suggests that the corrosion may be caused by local factors (such as corrosion depth, local coating damage) rather than axial stress or flow field dominance. Combined with the third preset threshold (high-throughput threshold) to confirm "local corrosion depth") and the number of connected components (the number of small, independent, and unconnected corrosion areas within a single pipe section) exceeds the third preset threshold (the lower limit of the number of connected components used to determine local corrosion depth, which includes 3-5 connected components), it is determined to be local corrosion depth.

[0067] The edge weight is determined by the correlation between the minimum curvature similarity between corrosions and the intergranular distortion rate. If the correlation exceeds the fourth preset threshold (the correlation threshold used to strengthen the stress corrosion determination), then the stress corrosion determination is strengthened.

[0068] Specifically, when the number of connected components exceeds this threshold, it indicates the presence of numerous scattered, independent pitting pits within that segment, rather than a large, interconnected area of ​​corrosion. This is a typical characteristic of localized deep corrosion. Although the diffusion of a single pit is not strong (meeting the second preset threshold), the presence of a large number of pitting pits will significantly weaken the overall strength of the pipe wall, requiring close attention to its maximum corrosion depth.

[0069] The fourth preset threshold is used to determine the correlation strength between the two indicators, "minimum curvature similarity" and "intergranular distortion rate." Minimum curvature similarity describes the sharpness of the macroscopic geometry of the corrosion pit, while intergranular distortion rate describes the degree of damage to the material's microscopic lattice. When the correlation between the two exceeds this threshold, it indicates that the macroscopic sharp corrosion and the microscopic grain boundary corrosion have a high degree of synchronicity and causality.

[0070] The calculation of edge weights was specifically defined. It stipulated how to determine continuous or localized corrosion based on the axial diffusion probability and connectivity of corrosion pits, and clarified that edge weights are jointly determined by the minimum curvature similarity (reflecting geometric morphology) and the correlation between intergranular distortion rate (reflecting the material's microstructure). Linking microscopic intergranular distortion with macroscopic corrosion spatial distribution through correlation calculation directly drives the edge strength in the graph neural network, which is a key mechanism for effectively fusing multi-scale information in the model.

[0071] In the above embodiments, the product of the number of corrosion pit coverage areas in consecutive segments and the edge weights in consecutive segments is calculated to amplify the corrosion impact of high-temperature zones or elbow sections.

[0072] The edge roughness of the corrosion pit coverage area is the spectral entropy value of the acoustic wave reflection signal at the corrosion boundary. When the spectral entropy value exceeds the fifth preset threshold (the lower limit of the spectral entropy used to mark active corrosion lesions), it is marked as an active corrosion lesion. When the spectral entropy value is lower than the sixth preset threshold (set between 3.0 and 4.0 (with the same dimension as the fifth threshold)), it indicates a stable corrosion morphology.

[0073] Specifically, a high spectral entropy value indicates that the distribution of the reflected sound wave signal in the frequency domain is highly chaotic and disordered, containing a large number of broadband components. This typically corresponds to the rough, irregular, and rapidly changing edges of corrosion pits, as this morphology scatters sound waves of multiple frequencies. When the spectral entropy exceeds this high threshold, the system marks it as an "active corrosion lesion," meaning that corrosion is developing rapidly and it is a high-risk point requiring priority treatment and close monitoring.

[0074] A low spectral entropy value indicates that the sound wave reflection signal is concentrated and orderly distributed in the frequency domain. This typically corresponds to smooth, regular, and slowly changing corrosion pit edges, such as old corrosion pits that have been partially covered by corrosion products or are in a stable state. When the spectral entropy is below this low threshold, the system indicates a "stable corrosion morphology." Although corrosion still exists, its risk of short-term expansion is relatively low.

[0075] Spectral entropy is introduced as an indicator to quantify the roughness of corrosion edges, and it is labeled as "active" or "stable" corrosion lesions. The frequency domain characteristics (spectral entropy) of the acoustic signal are dynamically correlated with the evolutionary state (active / stable) of corrosion, aiming to capture the temporal changes in corrosion and directly address the challenge of "difficulty in dynamically tracking evolution over time" in the background. Simultaneously, weighted calculations amplify the corrosion impact in high-temperature zones / elbow sections, reflecting a focus on high-risk areas in the actual service environment of pipelines, making the model predictions closer to engineering reality.

[0076] This embodiment achieves precise spatial perception of pipeline corrosion morphology through the construction of three-dimensional sound field data and reconstruction of geometric topology. The method utilizes sound wave reflection signals to generate circumferential depth coordinates of the pipe wall, forming a continuous atlas of thermal pipe walls, overcoming the shortcomings of traditional two-dimensional detection methods in representing three-dimensional corrosion morphology. It can accurately identify the distribution pattern, depth gradient, and spatial correlation characteristics of corrosion pits, providing a reliable data foundation for subsequent corrosion mode discrimination. Secondly, based on a multi-scale graph neural network model feature extraction mechanism, macroscopic corrosion distribution and microscopic morphological features are effectively integrated. By segmenting the pipeline, using the average corrosion area, maximum depth, and diffusion probability as node features, and combining microscopic parameters such as aspect ratio and edge roughness, multi-scale feature correlation analysis is achieved, significantly improving the identification accuracy of complex corrosion modes such as stress corrosion cracking and intergranular corrosion. Furthermore, through dynamic evolution tracking and risk warning mechanisms, the method achieves the temporal prediction function of corrosion development. This method utilizes corrosion diffusion probability along the axial direction and boundary topological connectivity analysis to determine the continuous development trend of corrosion and quantifies the degree of corrosion activity through spectral entropy values. By combining multiple parameters such as grain boundary distortion rate and coverage ratio, an assessment report is generated that includes remaining lifetime prediction and leakage risk level.

[0077] Example 2

[0078] This example aims to provide a detection device for collecting acoustic wave reflection signals from the wall of a heat pipe, such as... Figure 2 and Figure 3 The structure includes a base 1 and an extension 2 fixedly disposed on the side wall of the forward end of the base 1. An acoustic wave sensor array 3 is fixedly installed on the side of the extension 2 facing the heat pipe. A walking wheel 4 driven by a motor is fixedly installed on the base 1. The number of bases 1 is not less than four. A connecting part 5 is fixedly installed between two adjacent bases 1.

[0079] Specifically, sound wave reflection signals are collected using a four-directionally distributed acoustic wave sensor array 3. These multiple collected sound wave reflection signals are then combined to form three-dimensional sound field data. Furthermore, depth coordinates distributed circumferentially around the pipe wall are constructed based on feedback signals from the pipe wall. Power is supplied by a rechargeable lithium battery in the base 1, which is replenished by an external charger. The collected data is received by an external computer (transmitted via a local area network).

[0080] Example 3

[0081] This invention provides a non-transitory computer-readable storage medium storing at least one instruction or at least one program segment, which is loaded and executed by a processor to implement the following steps:

[0082] The sound wave reflection signals of the heat pipe wall are collected synchronously by the detection device, and three-dimensional sound field data are formed by combining multiple collected sound wave reflection signals. The depth coordinates of the heat pipe wall circumferential distribution are constructed based on the pipe wall feedback signal.

[0083] Based on three-dimensional sound field data, a geometric topology is constructed on the inner surface of the heat pipe wall to obtain a continuous heat pipe wall atlas.

[0084] Based on continuous atlases of thermal pipe walls, a multi-scale graph neural network model specific to corrosion and crack propagation in thermal pipe networks is constructed, outputting a pipe network health assessment report that includes corrosion level determination, remaining life prediction, and leakage risk warning.

[0085] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0086] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0087] Example 4

[0088] This invention provides an electronic device, including a processor and a memory, wherein the memory stores at least one instruction or at least one program segment, and the at least one instruction or the at least one program segment is loaded and executed by the processor to implement the following steps:

[0089] The sound wave reflection signals of the heat pipe wall are collected synchronously by the detection device, and three-dimensional sound field data are formed by combining multiple collected sound wave reflection signals. The depth coordinates of the heat pipe wall circumferential distribution are constructed based on the pipe wall feedback signal.

[0090] Based on three-dimensional sound field data, a geometric topology is constructed on the inner surface of the heat pipe wall to obtain a continuous heat pipe wall atlas.

[0091] Based on continuous atlases of thermal pipe walls, a multi-scale graph neural network model specific to corrosion and crack propagation in thermal pipe networks is constructed, outputting a pipe network health assessment report that includes corrosion level determination, remaining life prediction, and leakage risk warning.

[0092] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for predicting the corrosion life of a heat distribution pipe network, characterized by, The method comprises the following steps: Synchronously collecting the sound wave reflection signals of the heat pipe wall through the detection device, combining the three-dimensional sound field data based on the collected multiple sound wave reflection signals, and constructing the depth coordinates of the circumferential distribution of the heat pipe wall based on the pipe wall feedback signals; Based on the three-dimensional sound field data, a geometric topology structure is constructed on the inner surface of the heat pipe wall to obtain a continuous heat pipe wall atlas; Based on the continuous heat pipe wall atlas, a multi-scale graph neural network model specific to the corrosion and crack propagation of the heat pipe network is constructed, comprising: The pipeline is divided into several segments, and the corrosion area average, maximum corrosion depth and corrosion axial diffusion probability of each segment are taken as node features and edge weights; The single corrosion pit coverage area in the segment is obtained, and then the geometric center point of the single corrosion pit coverage area is taken as the center point, which includes the following structural corrosion features: Based on the set of sound wave reflection signals corresponding to the single corrosion pit coverage area obtained, the depth-width ratio under the single corrosion pit coverage area is judged, which is used to distinguish the corrosion pit coverage area and the corrosion depth. When the depth-width ratio is higher than a preset depth-width ratio threshold, it indicates that the corrosion coverage occupies a larger coverage range in the segment; The multi-scale graph neural network model specific to the corrosion and crack propagation of the heat pipe network generates a joint feature vector, which includes outputting a pipe network health assessment report containing corrosion grade determination, remaining life prediction and leakage risk warning based on the joint feature vector; The continuous heat pipe wall atlas includes analyzing the coverage ratio and rust point number of the corrosion area around the heat pipe weld, judging whether the rust point number in the corrosion area around the weld affects the central area, calculating the corrosion coverage ratio and the concentration degree of the rust points, and combining the preset threshold to determine the stress corrosion feature; The corrosion grade determination of the pipe network health assessment report is generated by the following rules: If the corrosion aggregation index of the girth weld area exceeds the preset coverage threshold and the grain boundary distortion rate exceeds the preset distortion threshold, it is determined as high-risk stress corrosion; If the corrosion is randomly distributed and the grain boundary distortion rate is lower than the seventh preset threshold, it is determined as the average corrosion pit coverage area; The evaluation report combines the heat pipe wall corrosion heat map and the three-dimensional reconstruction map of the grain boundary structure to display the analysis results and mark the remaining wall thickness distribution cloud map; The calculation of the central area of the corrosion area around the weld includes: Segmenting the three-dimensional sound field data of the weld area of the heat pipe to calculate the proportion of the corrosion area coverage length to the total length of the weld area; By Vietoris-Rips complex algorithm, the corrosion boundary topological connectivity is analyzed. When the number of connected components is 1 and the coverage ratio exceeds the preset coverage threshold, it is determined as stress corrosion cracking feature; The distortion rate of the grain boundary structure of the heat pipe wall is calculated by the phase shift amount of the sound wave scattering signal. If the distortion rate exceeds the preset distortion threshold, it is marked as the intergranular coverage typical feature; For the corrosion pit coverage area in the continuous segment, when the corrosion pit coverage area axial diffusion probability exceeds the first preset threshold, it is determined as continuous corrosion. When the corrosion pit coverage area axial diffusion rate is lower than the second preset threshold and the number of connected components exceeds the third preset threshold, the local corrosion depth is determined; The edge weight is determined by the inter-corrosion minimum curvature similarity and the inter-crystal distortion rate correlation. If the correlation exceeds the fourth preset threshold, the stress corrosion is strengthened.

2. The method for predicting the corrosion life of a heat distribution pipe network according to claim 1, characterized by, The product of the number of corrosion pit coverage areas in consecutive segments and the edge weight in consecutive segments is calculated to amplify the corrosion impact of high-temperature zones or elbow segments. The edge roughness of the corrosion pit coverage area is the spectral entropy value of the corrosion boundary acoustic reflection signal. When the spectral entropy value exceeds the fifth preset threshold, it is marked as an active corrosion lesion. When the spectral entropy value is lower than the sixth preset threshold, it prompts a stable corrosion pattern.

3. The method for predicting the corrosion life of a heat distribution pipe network according to claim 1, characterized by, The detection device comprises a base (1) and an extension (2) fixedly arranged on the front end side wall of the base (1). The extension (2) is fixedly installed with an acoustic sensor array (3) towards the heat pipe side. The base (1) is fixedly installed with walking wheels (4) driven by a motor. The number of bases (1) is not less than four. Adjacent two bases (1) are fixedly installed with a connecting part (5). 4.A non-transitory computer-readable storage medium having stored therein at least one instruction or at least one piece of program, characterized in that, The at least one instruction or the at least one program is loaded and executed by the processor to realize the steps of the heat pipe network corrosion life prediction method according to any one of claims 1-3.

5. An electronic device, comprising: The processor and the memory are included. The memory stores at least one instruction or at least one program. The at least one instruction or the at least one program is loaded and executed by the processor to realize the steps of the heat pipe network corrosion life prediction method according to any one of claims 1-3.

Images