A method, device, equipment and storage medium for repairing a flexible composite pipe

Abstract

The application discloses a flexible composite pipe repairing method, device, equipment and storage medium. The method comprises the following steps: based on a temperature difference-pressure-medium-humidity-illumination five-field coupling accelerated aging experiment field, analyzing and processing service environment, material combination and load characteristics of the flexible composite pipe to obtain monitoring data; based on the monitoring data and a failure process of the flexible composite pipe, establishing a damage evolution model, constructing a coupling constitutive model and verifying through finite element simulation; extracting features from the monitoring data to obtain feature parameters, combining the effective coupling constitutive model to construct a risk prediction model; based on the prediction result, determining a grading repair strategy and executing repair on the flexible composite pipe. The application can significantly improve the risk prediction accuracy and repair pertinence, effectively solve the cracking repair problem of the flexible composite pipe, reduce the maintenance cost and improve the service safety and reliability of the pipe material.

Description

Technical Field

[0001] This application relates to the field of material repair technology, and in particular to a repair method, apparatus, equipment and storage medium for flexible composite pipes. Background Technology

[0002] With the rapid development of deep-sea oil and gas gathering and transportation, desert oilfield exploitation, and polar resource development, flexible composite pipes are widely used as core transportation equipment. These pipes operate in extreme environments, facing not only drastic temperature fluctuations from -60℃ to 120℃, but also alternating internal pressures of 0.1 to 15 MPa and environments containing H2S / CO2 / Cl. - The effects of multiple coupled factors, including media corrosion, humidity variations ranging from 10% to 95% RH, and outdoor ultraviolet radiation, are significant. For example, a flexible riser in a deep-sea gas field in the South my country Sea is subjected to long-term conditions including a diurnal temperature range of 30°C, internal pressure fluctuations of 10 MPa, and high concentrations of Cl. - The complex corrosive conditions lead to frequent failures such as interface debonding and matrix cracking, with single repair costs exceeding ten million yuan, seriously affecting the safety and economy of oil and gas transportation.

[0003] Current technologies for repairing cracks in flexible composite pipes have significant limitations: First, the failure mechanism analysis is fragmented. Existing technologies mostly employ static or single-field coupling experiments, which cannot reveal the dynamic chain reaction of "thermal stress accumulation - interface phase reconstruction - matrix microcrack bifurcation - fiber-matrix debonding" during temperature cycling, and lack quantitative descriptions of the critical damage thresholds at each stage. Second, the accuracy of prediction models is insufficient. Traditional empirical formulas do not consider the nonlinear viscoelasticity of materials, damage anisotropy, and the correlation of multimodal data, resulting in prediction errors of stress cracking probability and remaining life often exceeding 30%. Third, the engineering applicability and repair specificity are poor. There is a lack of prediction frameworks for protecting data privacy across multiple oilfields, and repair methods are mostly single emergency treatments, failing to achieve graded and precise repair based on the degree of damage, making it difficult to support intelligent operation and maintenance of large-scale pipeline networks.

[0004] Therefore, how to accurately analyze the dynamic failure mechanism of stress cracking in flexible composite pipes under extreme multi-field coupling environments, construct a high-precision risk prediction model, and achieve targeted and highly adaptable hierarchical intelligent repair has become an urgent problem to be solved to ensure the safe service of flexible composite pipes throughout their entire life cycle and reduce maintenance costs. Summary of the Invention

[0005] In view of this, the present application provides a method, apparatus, device, and storage medium for repairing flexible composite pipes. This method analyzes and processes the service environment, material composition, and load characteristics of the flexible composite pipe based on a five-field coupled accelerated aging test field of temperature difference, pressure, medium, humidity, and light, obtaining monitoring data. Based on the monitoring data and the failure process of the flexible composite pipe, a damage evolution model is established, a coupled constitutive model is constructed, and verified through finite element simulation. Feature parameters are extracted from the monitoring data, and a risk prediction model is constructed by combining the effective coupled constitutive model. Based on the prediction results, a graded repair strategy is determined, and the flexible composite pipe is repaired. The method, apparatus, device, and storage medium for repairing flexible composite pipes provided in this application are implemented as follows:

[0006] This application provides a method for repairing a flexible composite pipe, comprising:

[0007] The service environment, material composition, and load characteristics of flexible composite pipes are analyzed and processed based on the coupled accelerated aging test field to obtain monitoring data. The coupled accelerated aging test field is a five-field coupled accelerated aging test field of temperature difference, pressure, medium, humidity, and light. The monitoring data includes temperature time series data, microscopic image data, acoustic emission signal data, and material property parameters.

[0008] Based on the monitoring data and the failure process of the flexible composite pipe, four consecutive failure periods are divided: thermal stress accumulation period, interface phase reconstruction period, matrix microcrack bifurcation period, and instability fracture period. The failure critical threshold and failure criteria corresponding to each failure period are determined. The damage evolution equation corresponding to each failure period is obtained by fitting based on energy dissipation theory, and a damage evolution model is established.

[0009] Based on the damage evolution model, a coupled constitutive model is constructed. The coupled constitutive model is then subjected to finite element embedding and simulation verification to obtain an effective coupled constitutive model.

[0010] Feature extraction processing is performed on the temperature time series data, the microscopic image data, the acoustic emission signal data, and the material property parameters to obtain feature parameters;

[0011] A dual-branch network model is constructed based on the effective coupling constitutive model as the global model. After integrating the dual-branch features in the dual-branch network model, the model is trained in combination with the feature parameters to obtain a risk prediction model for outputting stress cracking probability and remaining life. The first branch network is a graph neural network for extracting microscopic image features and material property features, and the second branch network is a Transformer network for extracting temperature time series features and acoustic emission signal features.

[0012] The flexible composite pipe is predicted based on the risk prediction model to obtain the prediction results, and a graded repair strategy is determined based on the prediction results.

[0013] The flexible composite pipe is repaired based on the graded repair strategy to obtain the repaired flexible composite pipe.

[0014] In some embodiments, based on the monitoring data and the failure process of the flexible composite pipe, four consecutive failure stages are divided: thermal stress accumulation stage, interface phase reconstruction stage, matrix microcrack bifurcation stage, and instability fracture stage. The failure critical threshold and failure criteria corresponding to each failure stage are determined. Based on energy dissipation theory, the damage evolution equation corresponding to each failure stage is fitted to establish a damage evolution model, including:

[0015] The monitoring data is analyzed and processed to obtain multiple failure periods, which include thermal stress accumulation period, interface phase reconstruction period, matrix microcrack bifurcation period and instability fracture period.

[0016] Acquire the failure critical thresholds corresponding to the temperature time series data, the microscopic image data, the acoustic emission signal data, and the material property parameters, respectively;

[0017] Based on the aforementioned failure threshold, multiple failure periods are quantified to obtain failure criteria corresponding to multiple failure periods.

[0018] The failure criteria, the temperature time series data, the microscopic image data, the acoustic emission signal data, and the material property parameters are fitted to obtain damage evolution models corresponding to multiple failure periods.

[0019] In some embodiments, the material property parameters include interfacial bond strength, difference in coefficients of thermal expansion, and viscoelastic relaxation modulus. The feature extraction processing of the temperature time-series data, the microscopic image data, the acoustic emission signal data, and the material property parameters to obtain feature parameters includes:

[0020] The temperature time series data is processed by wavelet packet decomposition and entropy calculation to extract energy entropy, sample entropy, approximate entropy, maximum Lyapunov exponent and adjacent temperature difference change rate to obtain temperature time series features.

[0021] The microscopic image data is analyzed and processed by a convolutional neural network to extract crack length, width ratio, interface debonding area ratio, and fiber-matrix contact point number to obtain microscopic image features.

[0022] The acoustic emission signal data is processed by short-time Fourier transform to extract the energy proportion of characteristic frequency bands, ringing count rate and rise time to obtain the acoustic emission signal characteristics;

[0023] The interfacial bonding strength, difference in thermal expansion coefficients, and viscoelastic relaxation modulus of the material properties are processed and analyzed to obtain the material property characteristics.

[0024] The temperature time-series features, the microscopic image features, the acoustic emission signal features, and the material property features are integrated and processed to obtain feature parameters.

[0025] In some embodiments, the step of constructing a dual-branch network model based on the effectively coupled constitutive model as the global model foundation, integrating the dual-branch features in the dual-branch network model, and then combining the feature parameters to train the model to obtain a risk prediction model for outputting stress cracking probability and remaining life includes:

[0026] The effective coupled constitutive model is trained to obtain the trained model parameters;

[0027] The trained model parameters are aggregated and optimized to obtain the global model base.

[0028] A dual-branch network model is constructed, which includes a first branch network and a second branch network. The first branch network is used to extract microscopic image features and material property features, and the second branch network is used to extract temperature time series features and acoustic emission signal features.

[0029] The risk prediction model is obtained by configuring training parameters and training the dual-branch network model and the global model.

[0030] In some embodiments, the step of performing prediction processing on the flexible composite pipe based on the risk prediction model to obtain prediction results, and determining a graded repair strategy based on the prediction results, includes:

[0031] Acquire real-time environmental data and material status data of the flexible composite pipe;

[0032] The real-time environmental data and material state data are input into the risk prediction model for prediction processing to obtain the prediction result.

[0033] The prediction results are divided into categories to obtain the damage levels;

[0034] Based on the damage level, a repair method matching process is performed to obtain a graded repair strategy. The graded repair strategy includes pre-embedded microcapsule autonomous repair matching, shape memory polymer patch active sealing matching, and in-tube repair robot emergency sealing matching.

[0035] In some embodiments, after repairing the flexible composite pipe based on the graded repair strategy to obtain the repaired flexible composite pipe, the method further includes:

[0036] The repaired area of ​​the repaired flexible composite pipe is continuously monitored, and post-repair status data related to strain, acoustic emission, temperature, and leakage are collected.

[0037] The repaired flexible composite pipe was compared and analyzed with the preset repair success criteria to obtain the repair effect;

[0038] If the repair effect meets the target, the risk prediction model is iteratively optimized based on the repair effect;

[0039] Alternatively, if the repair effect is unsatisfactory and the maximum number of repair attempts has not been reached, the graded repair strategy is adjusted and the repair operation is re-executed; if the repair effect is unsatisfactory and the maximum number of repair attempts has been reached, the flexible composite pipe is marked as high-risk to obtain a marked flexible composite pipe.

[0040] In some embodiments, the maximum number of repair attempts is 3.

[0041] This application provides a repair device for a flexible composite tube, comprising:

[0042] The analysis module is used to analyze and process the service environment, material combination and load characteristics of flexible composite pipe based on the coupled accelerated aging test field to obtain monitoring data. The coupled accelerated aging test field is a five-field coupled accelerated aging test field of temperature difference-pressure-medium-humidity-light. The monitoring data includes temperature time series data, microscopic image data, acoustic emission signal data and material property parameters.

[0043] The module is used to divide the failure process of the flexible composite pipe into four consecutive failure periods: thermal stress accumulation period, interface phase reconstruction period, matrix microcrack bifurcation period, and instability fracture period, based on the monitoring data and the failure process of the flexible composite pipe. It determines the failure critical threshold and failure criteria corresponding to each failure period, and obtains the damage evolution equation corresponding to each failure period based on energy dissipation theory, and establishes a damage evolution model.

[0044] The construction module is also used to construct a coupled constitutive model based on the damage evolution model, and to perform finite element embedding and simulation verification on the coupled constitutive model to obtain an effective coupled constitutive model;

[0045] The processing module is used to perform feature extraction processing on the temperature time series data, the microscopic image data, the acoustic emission signal data, and the material property parameters to obtain feature parameters;

[0046] The processing module is further configured to construct a dual-branch network model based on the effective coupling constitutive model as the global model basis, integrate the dual-branch features in the dual-branch network model and combine them with the feature parameters to train the model, thereby obtaining a risk prediction model for outputting stress cracking probability and remaining life, wherein the first branch network is a graph neural network for extracting microscopic image features and material property features, and the second branch network is a Transformer network for extracting temperature time series features and acoustic emission signal features;

[0047] The processing module is also used to perform prediction processing on the flexible composite pipe based on the risk prediction model, obtain prediction results, and determine a graded repair strategy based on the prediction results;

[0048] The processing module is further configured to repair the flexible composite tube based on the graded repair strategy to obtain a repaired flexible composite tube. The computer device provided in this application embodiment includes a memory and a processor. The memory stores a computer program that can run on the processor, and the processor executes the program to implement the method described in this application embodiment.

[0049] The computer-readable storage medium provided in this application embodiment stores a computer program thereon, which, when executed by a processor, implements the method described in this application embodiment.

[0050] This application provides a method, apparatus, equipment, and storage medium for repairing flexible composite pipes. It analyzes and processes the service environment, material composition, and load characteristics of the flexible composite pipe using a five-field coupled accelerated aging test field based on temperature difference, pressure, medium, humidity, and light, obtaining monitoring data. Based on the monitoring data and the failure process of the flexible composite pipe, a damage evolution model is established, a coupled constitutive model is constructed, and verified through finite element simulation. Feature parameters are extracted from the monitoring data, and a risk prediction model is constructed by combining the effectively coupled constitutive model. Based on the prediction results, a graded repair strategy is determined, and repair is performed on the flexible composite pipe. This significantly improves the accuracy of risk prediction and the targeted nature of repair, effectively solves the problem of crack repair in flexible composite pipes, reduces maintenance costs, improves the service safety and reliability of the pipe, and addresses the technical problems mentioned in the background art. Attached Figure Description

[0051] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application 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 based on these drawings without creative effort.

[0052] Figure 1A schematic diagram illustrating the implementation process of a repair method for a flexible composite pipe provided in an embodiment of this application;

[0053] Figure 2 A schematic diagram illustrating the implementation process of establishing a damage evolution model, provided in an embodiment of this application;

[0054] Figure 3 This is a schematic diagram of a repair device for a flexible composite pipe provided in an embodiment of this application. Detailed Implementation

[0055] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0056] The following description of some technologies involved in the embodiments of this application is provided to aid understanding and should be considered merely exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. Similarly, for clarity and brevity, some descriptions of well-known functions and structures are omitted in the following description.

[0057] Figure 1 This is a schematic flowchart illustrating the implementation of a repair method for a flexible composite pipe provided in an embodiment of this application, including steps 101 to 107. Wherein, Figure 1 This is merely one execution order shown in the embodiments of this application and does not represent the only execution order of a method for repairing flexible composite pipes. Where the final result can be achieved, Figure 1 The steps shown can be performed in parallel or in reverse order.

[0058] Step 101: Based on the coupled accelerated aging test field, the service environment, material combination and load characteristics of the flexible composite pipe are analyzed and processed to obtain monitoring data.

[0059] In this embodiment, a multi-field coupling accelerated aging test system was designed and built based on the actual service environment of the flexible composite pipe to collect monitoring data.

[0060] The experimental system is based on the coupling of five fields: temperature difference, pressure, medium, humidity, and light. The environmental parameter spectrum, sample system, and load spectrum are determined with reference to the actual application scenarios of flexible composite pipes. The environmental parameter spectrum covers extreme temperature ranges, pressure fluctuation ranges, corrosive medium composition and concentration, humidity gradient, and light intensity, comprehensively simulating extreme working conditions such as deep sea and desert. The sample system combines different types of reinforcing phases, matrix, and interface phases. Samples with typical prefabrication defects are processed and pretreated according to standards. Sufficient parallel samples are set to ensure data reliability. The load spectrum is compiled based on long-term oilfield service data, including transient, steady-state, and impact conditions, ensuring consistency with actual service load characteristics.

[0061] During the experiment, cross-scale monitoring was conducted simultaneously. At the microscopic level, in-situ dynamic characterization techniques were used to observe the evolution of the interface and internal microstructure of the flexible composite pipe, collecting data on interface bonding state, molecular chain changes, and other relevant information. At the mesoscopic level, multi-parameter synchronous monitoring methods were employed to capture strain field distribution, crack propagation signals, and local thermal effects, establishing correlations between damage-related physical quantities. At the macroscopic level, performance testing and analysis were conducted to obtain macroscopic performance parameters of the pipe, including mechanical, thermal, and dielectric properties, as well as performance degradation patterns. Through experiments and monitoring, monitoring data including temperature time-series data, microscopic image data, acoustic emission signal data, and material property parameters were ultimately obtained.

[0062] Step 102: Establish a damage evolution model based on monitoring data and the failure process of the flexible composite pipe.

[0063] In this embodiment of the application, a damage evolution model is established based on the collected monitoring data and the failure characteristics and evolution law of the flexible composite pipe.

[0064] First, a comprehensive analysis of the monitoring data was conducted to identify the key stages in the failure process of the flexible composite pipe, dividing it into four consecutive failure stages: thermal stress accumulation period, interface phase reconstruction period, matrix microcrack bifurcation period, and instability fracture period. For each failure stage, corresponding characteristic parameters were extracted, and the critical failure threshold for each stage was determined through quantitative analysis, forming a clear four-stage failure criterion to accurately define the failure process.

[0065] Subsequently, based on the energy dissipation theory and combined with monitoring data of each failure stage, including temperature change patterns, stress state parameters, crack evolution characteristics, and inherent material properties, the relevant parameters of the model were fitted and determined, and a damage evolution model corresponding to each failure stage was established to quantify the change of damage degree with service time or number of cycles.

[0066] Step 103: Construct a coupled constitutive model based on the damage evolution model, and perform finite element embedding and simulation verification on the coupled constitutive model to obtain an effective coupled constitutive model.

[0067] In this embodiment, a coupled constitutive model is constructed based on the damage evolution model and combined with the mechanical properties of materials and the law of multi-field coupling, and the optimization is verified by finite element simulation.

[0068] Based on the generalized continuum damage mechanics theory, a damage-related tensor is introduced, the influence of the interface relative to the stiffness of the flexible composite tube is considered, and the nonlinear viscoelastic properties of the material are combined with the instantaneous elastic stiffness and various strains (thermal strain, plastic strain, damage strain, etc.) to construct an incremental coupled constitutive model, thereby realizing the quantitative correlation between the microscopic damage mechanism and the macroscopic mechanical response.

[0069] The constructed coupled constitutive model was embedded into finite element analysis software, and adaptive mesh generation technology was used to optimize the mesh in key areas to improve the accuracy of simulation calculations. Using measured data from monitoring data, simulation calculations were performed on the stress distribution at the pipe interface and areas of concentrated damage. The differences between the simulation results and the measured data were compared, and the model parameters were iteratively optimized to finally obtain an effective coupled constitutive model that highly matches the actual working conditions.

[0070] Step 104: Perform feature extraction processing on temperature time series data, microscopic image data, acoustic emission signal data, and material property parameters to obtain feature parameters.

[0071] In this embodiment of the application, various basic data in the monitoring data are subjected to targeted feature extraction and integration to form a feature parameter set for model construction.

[0072] For temperature time-series data, signal processing methods are used for decomposition and entropy calculation to extract relevant features reflecting temperature change patterns and fluctuation characteristics. For microscopic image data, image analysis techniques are used to extract crack geometric parameters, interfacial debonding degree, and material microstructure features. For acoustic emission signal data, after signal transformation processing, features such as characteristic frequency band energy proportions, signal count rate, and rise time related to crack propagation and interfacial debonding are extracted. Material property parameters are analyzed and processed to extract interfacial bonding strength, differences in thermal expansion coefficients, and viscoelasticity-related features. Finally, all features are integrated to form a complete feature parameter set.

[0073] Step 105: Perform model construction processing on the effectively coupled constitutive model and feature parameters to obtain the risk prediction model.

[0074] In this embodiment of the application, by combining the constitutive model and the feature parameter set effectively, a suitable model architecture is built to construct a risk prediction model that can accurately predict the risk status of pipe materials.

[0075] First, a federated learning framework is built to balance the protection of privacy of multi-source data with the generalization ability of the model. Each edge node trains the basic model based on the simulation results of the local feature parameters and the effective coupling of the constitutive model. Only the model parameters are uploaded to the central server, which optimizes the global model base through the parameter aggregation algorithm.

[0076] Subsequently, a dual-branch network model was constructed. The first branch network was used to extract microscopic topological features of the material, while the second branch network was used to capture the dependencies in time-series data. The features extracted by the two branches were weighted and integrated through a gated fusion unit to obtain comprehensive features that integrate multi-dimensional information. Finally, training parameters and loss functions were set, and the model was trained and optimized to obtain a risk prediction model that can output the stress cracking probability and remaining life of flexible composite pipes.

[0077] Step 106: Perform prediction processing on the flexible composite pipe based on the risk prediction model to obtain the prediction results, and determine the graded repair strategy based on the prediction results.

[0078] In this embodiment of the application, real-time environmental data and material condition data from on-site monitoring of the flexible composite pipe are input into the established risk prediction model to obtain the stress cracking probability and remaining life prediction results for each part.

[0079] Based on the predicted stress cracking probability and the four-stage failure criteria, the corresponding damage levels are classified as follows: when the stress cracking probability is low, it is judged as minor damage, corresponding to the thermal stress accumulation period or the interface phase reconstruction period; when the stress cracking probability is at a medium level, it is judged as a large crack, corresponding to the matrix microcrack bifurcation period; when the stress cracking probability is high, it is judged as severe damage, corresponding to the unstable fracture period.

[0080] For different damage levels, corresponding repair methods are matched to determine a graded repair strategy: micro-damage is repaired by pre-embedded microcapsules, which utilize the characteristic of microcapsules to automatically rupture and release repair agents under specific stress conditions to achieve autonomous repair of nanoscale cracks; larger cracks are sealed by shape memory polymer patches, which apply stable pressure to the cracks by activating the shape recovery characteristics of shape memory polymers to achieve effective sealing; severe damage is repaired by in-pipe repair robots for emergency sealing, which achieves emergency repair of macroscopic fractures or leaks by precise robot positioning and rapid curing material injection.

[0081] Step 107: Repair the flexible composite pipe based on the graded repair strategy to obtain the repaired flexible composite pipe.

[0082] In this embodiment, targeted repair operations are performed on the damaged areas of the flexible composite pipe according to a determined graded repair strategy. After the repair is completed, the repaired area is continuously monitored, and data on strain, acoustic emission, temperature, and leakage-related conditions are collected to evaluate the repair effect.

[0083] If the repaired status data meets the preset repair success criteria, namely uniform strain distribution, no abnormal acoustic emission signals, no leakage, and no new cracks initiation during subsequent temperature cycling, the repair is deemed successful. The pipe health record is updated, and the repair-related data is added to the model training dataset for iterative optimization of the risk prediction model. If the repair effect does not meet the criteria and the maximum number of repair attempts has not been reached, the repair strategy is upgraded and the repair operation is re-executed. If the maximum number of repair attempts has been reached but the criteria are still not met, a manual intervention alarm is triggered, the flexible composite pipe is marked as high-risk, and the subsequent monitoring frequency is increased to ensure the safe operation of the flexible composite pipe.

[0084] This application's embodiments construct a complete technical system encompassing five-field coupled experiments, model building, predictive decision-making, and graded repair. This system accurately reproduces the service conditions of flexible composite pipes under extreme environments involving multiple coupled fields such as temperature difference, pressure, medium, humidity, and light, solving the problem that traditional single-field / static experiments cannot simulate real failure scenarios. Through multi-model collaboration, it achieves full-link quantification from failure mechanism to risk prediction to repair execution, overcoming the fragmented limitations of existing technologies and significantly improving the timeliness of stress cracking risk identification and the effectiveness of repair for flexible composite pipes.

[0085] In the above Figure 1 Based on the above, this application embodiment also provides a schematic diagram of the implementation process for establishing a damage evolution model, as shown below. Figure 2 As shown, steps 201 to 204 are included:

[0086] Step 201: Analyze and process the monitoring data to obtain multiple failure periods, including the thermal stress accumulation period, the interface phase reconstruction period, the matrix microcrack bifurcation period, and the instability fracture period.

[0087] In this embodiment, the microscopic characterization data, mesoscopic monitoring data, and macroscopic performance data in the monitoring data are comprehensively analyzed. Combined with the damage evolution characteristics of the flexible composite pipe under the multi-field coupling of temperature difference, pressure, medium, humidity, and light, the core failure signals of each stage are identified: In the early stage of temperature difference cycling, the interfacial shear stress is monitored to accumulate with the increase of temperature difference cycle number, the acoustic emission signal activity is low, and the strain field distribution is relatively uniform, corresponding to the thermal stress accumulation period; subsequently, the interfacial chemical bonds are reorganized, and the interfacial peeling force decreases significantly, corresponding to the interfacial phase reconstruction period; then, the number of silver crazing in the matrix increases sharply and branches and expands along the fiber direction, and the energy proportion of the acoustic emission characteristic frequency band increases sharply, corresponding to the matrix microcrack branching period; finally, the fiber-matrix debonding area increases significantly, and the remaining strength of the flexible composite pipe decreases sharply, corresponding to the instability fracture period. Thus, four failure periods are divided: thermal stress accumulation period, interfacial phase reconstruction period, matrix microcrack branching period, and instability fracture period.

[0088] Step 202: Obtain the temperature time series data, microscopic image data, acoustic emission signal data, and the failure critical threshold corresponding to the material property parameters.

[0089] In this embodiment, based on temperature time-series data, the critical number of temperature difference cycles and the critical value of interfacial shear stress during the thermal stress accumulation period are determined by analyzing the correlation between temperature difference amplitude, cycle number, and damage. Based on microscopic image data, the changes in interfacial nanocrack size and interfacial phase morphology are observed. Combined with the detection results, the critical characteristics of chemical bond recombination during the interfacial phase reconstruction period and the critical range of interfacial peeling force are obtained. Wavelet packet decomposition and feature extraction are performed on acoustic emission signal data to determine the critical value of acoustic emission ringing count, the critical value of characteristic frequency band energy ratio, and the critical range of crack propagation rate during the matrix microcrack bifurcation period. Based on the difference in thermal expansion coefficients, initial interfacial bonding strength, and residual strength in the material property parameters, the critical value of debonding area ratio and the critical ratio of residual strength during the unstable fracture period are obtained, thereby obtaining the failure critical thresholds corresponding to various data.

[0090] Step 203: Quantify multiple failure periods based on the failure critical threshold to obtain failure criteria corresponding to multiple failure periods.

[0091] In this embodiment, a fuzzy comprehensive evaluation method is used to establish a membership function. Integrating acoustic emission, strain monitoring, and microscopic characterization data, the critical thresholds for each failure stage are quantified and integrated: the quantification criteria for the thermal stress accumulation stage are: interfacial shear stress ≥ 30% of the initial interfacial bond strength, acoustic emission ringing count ≤ 10 times / min, and strain concentration factor ≤ 1.5; the quantification criteria for the interfacial phase reconstruction stage are: detection of a ≥ 20% increase in Si-O peak area and a decrease in interfacial peeling force to 40%~60% of the initial value; and the quantification criteria for the matrix microcrack bifurcation stage are: matrix crazing density ≥ 10... 4The crack length / mm³, the increase in the full width at half maximum (FWHM) of the CC peak in the Raman spectrum is ≥15%, the crack bifurcation angle is between 30° and 75°, and the crack propagation rate is 10. -9 ~10 -7 m / cycle; the quantitative criteria for the unstable fracture period are: fiber-matrix debonding area ≥ 50%, acoustic emission b value drops below 0.8, and residual strength ≤ 50% of initial strength, thus forming clear failure criteria for each failure period.

[0092] Step 204 involves fitting the failure criteria, temperature time series data, microscopic image data, acoustic emission signal data, and material property parameters to obtain damage evolution models corresponding to multiple failure periods.

[0093] In this embodiment, based on energy dissipation theory and using failure criteria for each failure stage as constraints, the data is combined with temperature change patterns, stress state parameters, crack evolution characteristics, and inherent material properties (such as thermal expansion coefficient difference, interfacial bond strength, and viscoelastic relaxation modulus) from monitoring data to fit the damage variation with the number of cycles. For the thermal stress accumulation stage, an elastic damage evolution equation including parameters such as interfacial shear stress, activation energy, and temperature is fitted; for the interfacial phase reconstruction stage, an interfacial phase reconstruction damage evolution equation involving parameters such as thermal expansion coefficient difference, temperature difference, and holding time is fitted; for the matrix microcrack bifurcation stage, a matrix damage evolution equation including parameters such as creasing density, crack propagation rate, and activation energy is fitted; and for the unstable fracture stage, an unstable fracture damage evolution equation involving parameters such as residual strength and initial strength is fitted, ultimately yielding damage evolution models corresponding to the four failure stages.

[0094] This application addresses the problems of ambiguous analysis and lack of quantitative description in existing failure mechanisms by quantifying the critical thresholds and criteria for failure at each stage, thus achieving precise definition of the failure process. Based on multi-source monitoring data, it fits specific damage evolution models for each failure stage, quantifying the change in damage with the number of cycles, improving the accuracy of predicting damage evolution trends, and avoiding delays in repair decisions due to unclear mechanisms. By integrating multi-parameter monitoring data through fuzzy comprehensive evaluation to determine failure criteria, it ensures the scientific validity and reliability of the transition thresholds at each stage, providing a clear basis for the formulation of graded repair strategies and making repair operations more targeted.

[0095] In some embodiments, the material property parameters include interfacial bonding strength, difference in thermal expansion coefficients, and viscoelastic relaxation modulus. Feature extraction processing is performed on temperature time series data, microscopic image data, acoustic emission signal data, and material property parameters to obtain feature parameters, including: performing wavelet packet decomposition and entropy calculation processing on temperature time series data to extract energy entropy, sample entropy, approximate entropy, maximum Lyapunov exponent, and adjacent temperature difference change rate to obtain temperature time series features.

[0096] Specifically, after preprocessing the acquired temperature time-series data, a five-layer wavelet packet decomposition process is performed to decompose the temperature signal into different frequency sub-bands. The wavelet packet energy entropy is obtained by calculating the energy distribution of each sub-band. Based on the information entropy theory, the sample entropy and approximate entropy of the temperature time-series data are calculated to quantify the complexity and irregularity of the data sequence. The maximum Lyapunov exponent is solved using the small data quantity method to characterize the chaotic characteristics of temperature changes. The rate of change of adjacent temperature differences is obtained by calculating the ratio of the temperature difference difference to the time interval between two adjacent monitoring periods. The extracted wavelet packet energy entropy, sample entropy, approximate entropy, maximum Lyapunov exponent, and rate of change of adjacent temperature differences are combined to form the temperature time-series characteristics.

[0097] Furthermore, convolutional neural network analysis was performed on the microscopic image data to extract crack length, width ratio, interface debonding area ratio, and fiber-matrix contact point number, thus obtaining microscopic image features.

[0098] Specifically, microscopic image data of flexible composite pipes are collected, including fiber-matrix interface morphology images and crack initiation and propagation images. After preprocessing such as grayscale correction and noise suppression, the images are input into a convolutional neural network for feature analysis. Deep features of the images are extracted step-by-step through the convolutional and pooling layers of the convolutional neural network. The ratio of crack length to width is accurately identified and calculated. Image segmentation techniques are used to quantify the area ratio of the interface debonding region to the entire observation area. The number of effective contact points between the fiber and the matrix is ​​counted. Finally, microscopic image features containing the crack length-to-width ratio, the proportion of interface debonding area, and the number of fiber-matrix contact points are obtained.

[0099] Furthermore, the acoustic emission signal data is processed by short-time Fourier transform to extract the energy proportion of characteristic frequency bands, ringing count rate, and rise time, thereby obtaining the acoustic emission signal characteristics.

[0100] Specifically, the acoustic emission signal data acquired by the acoustic emission sensor array is filtered and denoised to remove environmental interference signals and retain effective signals related to crack propagation and interface debonding. Short-time Fourier transform (SFT) technology is used to perform time-frequency analysis on the effective acoustic emission signals, decomposing the signals into a three-dimensional time-frequency-amplitude spectrum. The focus is on the 150–250 kHz characteristic frequency band, calculating the proportion of energy in this band to the total signal energy. The ringing count of the acoustic emission signal per unit time is statistically analyzed to obtain the ringing count rate. The time interval from the start point to the peak point of each acoustic emission pulse signal is extracted as the rise time. The characteristic frequency band energy proportion, ringing count rate, and rise time are integrated to form the acoustic emission signal characteristics.

[0101] Furthermore, the interfacial bonding strength, difference in thermal expansion coefficients, and viscoelastic relaxation modulus in the material property parameters are processed and analyzed to obtain the material property characteristics.

[0102] Specifically, a systematic compilation and analysis were conducted on the interfacial bond strength, the difference in thermal expansion coefficients, and the viscoelastic relaxation modulus, which are key material property parameters. Interfacial bond strength was determined using curve measurement data, with the average of multiple measurements taken as the characteristic parameter. The difference in thermal expansion coefficients was obtained by separately testing the thermal expansion coefficients of the reinforcing phase material and the matrix material, calculating the difference between their absolute values. The viscoelastic relaxation modulus was determined by fitting stress relaxation experimental data at different temperatures and times to establish model parameters, thereby obtaining the viscoelastic relaxation modulus characteristics. The comprehensive analysis of these three types of parameters resulted in the material property characteristics.

[0103] Furthermore, the temperature time series characteristics, microscopic image characteristics, acoustic emission signal characteristics, and material property characteristics are integrated and processed to obtain characteristic parameters.

[0104] Specifically, the extracted temperature time-series features, microscopic image features, acoustic emission signal features, and material property features are standardized to unify data dimensions and units. The multi-dimensional features are then integrated into a complete feature vector set using a feature splicing method. This feature vector set covers various aspects such as temperature change patterns, microstructure evolution, damage signal response, and inherent material properties, ultimately forming feature parameters for building risk prediction models.

[0105] This application's embodiments employ a dedicated feature extraction scheme designed for temperature time series, microscopic images, acoustic emission signals, and material property parameters. This comprehensively captures multi-dimensional key information related to stress cracking, solving the problems of single feature extraction and incomplete information coverage in traditional methods, and improving the correlation between features and damage states. The extracted core features, such as energy entropy, crack geometric parameters, and energy proportions of characteristic frequency bands, accurately reflect temperature fluctuation patterns, microstructural evolution, damage signal response, and inherent material properties, providing high-quality, highly discriminative data support for risk prediction models. By integrating features to form a complete feature parameter set, effective fusion of multimodal data is achieved, avoiding the limitations of single features. This enables subsequent risk prediction models to comprehensively learn damage-related information, improving the model's ability to identify stress cracking risks under complex working conditions.

[0106] In some embodiments, a risk prediction model is obtained by performing model construction processing on the effectively coupled constitutive model and feature parameters, including: training the effectively coupled constitutive model to obtain trained model parameters.

[0107] Specifically, measured data acquired through cross-scale monitoring is used as training input. This measured data includes microscopic interface crack propagation data, mesoscopic strain field and acoustic emission signal data, and macroscopic stress-strain and performance degradation data. This data is input into a coupled constitutive model of nonlinear viscoelasticity-anisotropic damage-interfacial phase evolution. Through data fitting methods, the material constants in the model are optimized, including nonlinear viscoelastic parameters, anisotropic damage tensor correlation parameters, and the influence coefficient of the interface phase volume fraction, ensuring that the stress response and damage evolution laws output by the model are consistent with the measured data. Finally, the trained model parameters are obtained.

[0108] Furthermore, the trained model parameters are aggregated and optimized to obtain the global model foundation.

[0109] Specifically, parameter aggregation optimization is performed based on a federated learning architecture of a central server and edge nodes. Each edge node uses locally collected monitoring data and trained coupled constitutive model parameters to train a local basic model, extracting only the core model parameters and uploading them to the central server to ensure the privacy and security of multi-source data. After receiving the model parameters uploaded by each edge node, the central server uses a federated averaging algorithm to perform weighted aggregation of the parameters, eliminating model bias caused by differences in data distribution across different nodes.

[0110] Furthermore, a dual-branch network model is constructed, which includes a first branch network and a second branch network. The first branch network is used to extract microscopic image features and material property features, while the second branch network is used to extract temperature time series features and acoustic emission signal features.

[0111] Specifically, a dual-branch network model adapted to multimodal features is constructed to achieve accurate extraction and adaptation of different types of features: The first branch network adopts a graph neural network architecture to extract microscopic image features and material property features. Information such as fiber arrangement, interface phase distribution, and crack morphology reflected in the microscopic image is transformed into a graph structure. Combined with inherent properties in material property parameters such as interfacial bonding strength, difference in thermal expansion coefficients, and viscoelastic relaxation modulus, topologically related features such as node degree distribution and clustering coefficient are captured through graph convolutional layers to achieve a deep fusion representation of microstructure and material properties.

[0112] The second branch network employs a Transformer architecture to extract temperature time-series features and acoustic emission signal features. Temperature time-series data and acoustic emission signal data are encoded into sequence vectors along the time dimension. A multi-head self-attention mechanism is used to focus on key time-series nodes, capturing the time-dependent relationship between temperature changes and acoustic emission signals, thus enhancing the extraction of key time-series information on damage evolution.

[0113] Furthermore, the training parameters of the dual-branch network model and the global model are configured and the model is trained to obtain the risk prediction model.

[0114] Specifically, the extracted feature parameters are fused with the simulation data output by the effectively coupled constitutive model to form a training dataset that combines measured features with theoretical simulation support.

[0115] A multi-task weighted loss function is set up, with weight coefficients determined through Bayesian optimization. This function includes classification loss for predicting stress cracking probability, regression loss for predicting remaining life, and KL divergence loss to constrain the consistency of damage variable distribution. The AdamW optimizer is selected, with a weight decay coefficient of 0.02 configured to suppress overfitting. The OneCycleLR learning rate scheduling strategy is adopted, with a maximum learning rate of 0.001 to adapt to gradient changes during model training.

[0116] The fused training data is input into the dual-branch network and iteratively trained in conjunction with the global model. The features extracted from the two branches are weighted and integrated through the gating fusion unit, so that the model can learn the correlation between microstructure and material properties and the evolution of time series data at the same time. Finally, a risk prediction model that can accurately output stress cracking probability and remaining life is obtained.

[0117] This application's embodiments, by combining the theoretical support of an effectively coupled constitutive model with measured information on multimodal characteristic parameters, construct a risk prediction model that combines theoretical consistency with engineering practicality, addressing the problems of traditional models either lacking theoretical basis or being detached from actual working conditions. The dual-branch network model accurately extracts microscopic topological features and temporal dependency features respectively, and after training and optimization, it reduces the prediction error of stress cracking probability and the relative error of remaining lifetime, improving prediction accuracy compared to traditional models.

[0118] In some embodiments, a risk prediction model is used to predict the flexible composite pipe to obtain prediction results, and a graded repair strategy is determined based on the prediction results, including: acquiring real-time environmental data and material state data of the flexible composite pipe.

[0119] Specifically, a multi-parameter sensor network deployed at key locations of the flexible composite pipe (such as elbows, welds, and stress concentration areas) collects real-time environmental and material condition data. The real-time environmental data covers the temperature (accuracy ±0.5℃), internal pressure (range 0.1~15MPa), and medium composition and concentration (including H2S, CO2, Cl) for the current service environment. -The data includes thermal strain data collected by fiber optic grating sensors, crack-related acoustic signals captured by acoustic emission sensor arrays, acoustic characteristics monitored by distributed acoustic wave sensing systems, and media corrosion correlation data fed back by micro electrochemical sensors, comprehensively reflecting the current service environment and condition of the flexible composite pipe.

[0120] Furthermore, real-time environmental data and material condition data are input into the risk prediction model for prediction processing to obtain prediction results.

[0121] Specifically, the collected real-time environmental data and material condition data are standardized and preprocessed to unify data dimensions and units before being input into the constructed bi-branch risk prediction model. The model extracts the microscopic topological features of the pipe material and captures the long-term temporal dependence of temperature time series and acoustic emission signals. After weighted integration of multi-dimensional features by a gated fusion unit, it outputs two prediction results: stress cracking probability and remaining life.

[0122] Furthermore, the prediction results are divided into categories to obtain the damage level.

[0123] Specifically, based on the stress cracking probability in the prediction results and referring to the four-stage failure criterion, the damage level of the flexible composite pipe is classified. When the stress cracking probability is less than 30%, it is judged as minor damage, corresponding to the thermal stress accumulation period or the interface phase reconstruction period. At this time, the flexible composite pipe mainly exhibits interface shear stress accumulation or early interface phase reconstruction, without obvious crack propagation. When the stress cracking probability is between 30% and 70%, it is judged as a large crack, corresponding to the microcrack bifurcation period of the matrix. The flexible composite pipe matrix has formed a certain density of silver streaks and bifurcates along the fiber direction. When the stress cracking probability is greater than or equal to 70%, it is judged as severe damage, corresponding to the unstable fracture period. The fiber-matrix debonding area accounts for more than 50%, and there is a risk of macroscopic fracture or leakage.

[0124] Repair methods are matched based on damage level to obtain a graded repair strategy, which includes pre-embedded microcapsules for autonomous repair matching, shape memory polymer patches for active sealing matching, and in-tube repair robots for emergency sealing matching.

[0125] Specifically, based on the classified damage levels, corresponding repair methods are matched to form a graded repair strategy. For minor damage, a pre-embedded microcapsule self-repair method is matched. Microcapsules (50~100μm in diameter) pre-embedded at the interface between the inner liner and the reinforcement layer automatically rupture when the interfacial shear stress reaches 30% of the initial bond strength, releasing a repair agent to fill the nanoscale cracks. For larger cracks, a shape memory polymer patch active sealing method is matched. By activating epoxy resin-based patches near the damage point and heating them above the glass transition temperature to cause them to shrink, a compressive force of not less than 2MPa is applied to the crack to achieve active sealing. For severe damage, an in-tube repair robot emergency sealing method is matched. An in-tube repair robot is precisely moved to the damage point and injects a fast-curing material to fill the fracture area to achieve emergency sealing. Finally, a complete graded repair strategy including the matching of three repair methods is formed.

[0126] This application's embodiments utilize dynamic prediction based on real-time environmental and material condition data to ensure the timeliness and accuracy of damage level classification. This addresses the problems of traditional repair methods relying on manual inspection and experiencing response delays, enabling early warning and rapid response to stress cracking risks. Three repair methods—pre-embedded microcapsules, shape memory polymer patches, and in-pipe repair robots—are matched according to damage level, avoiding a one-size-fits-all approach. This allows for precise adaptation to autonomous repair of minor damage, active sealing of larger cracks, and emergency plugging of severe damage, improving repair efficiency and effectiveness. The tiered repair strategy balances autonomy and specificity, reducing unnecessary manual intervention and over-repair, lowering maintenance costs, and ensuring the safety and reliability of repair operations, adapting to repair needs at different failure stages.

[0127] In some embodiments, after repairing the flexible composite pipe based on a graded repair strategy and obtaining the repaired flexible composite pipe, the method further includes: continuously monitoring the repaired area of ​​the repaired flexible composite pipe and collecting post-repair status data related to strain, acoustic emission, temperature, and leakage.

[0128] Specifically, for the repaired flexible composite pipe, a multi-parameter sensor network is deployed to continuously monitor the repaired area and surrounding stress-sensitive zones to ensure comprehensive capture of post-repair changes. Fiber optic strain sensors collect strain distribution data in the repaired area, focusing on changes in strain concentration coefficient; acoustic emission sensor arrays capture crack propagation-related acoustic emission signals, focusing on parameters such as energy percentage and ringing count rate in the 150-250kHz characteristic frequency band; miniature temperature sensors record temperature fluctuation data to ensure stability under temperature cycling; and a distributed acoustic wave sensing system monitors leak-related acoustic characteristics to determine if media leakage exists. Ultimately, complete post-repair status data, including strain, acoustic emission, temperature, and leak-related information, is collected.

[0129] Furthermore, the repaired flexible composite pipe was compared and analyzed with the preset successful repair criteria to obtain the repair effect.

[0130] Specifically, clear criteria for successful repair are pre-defined, based on the four-stage failure mechanism of flexible composite pipes and repair objectives: strain concentration factor ≤ 1.5, acoustic emission energy proportion in the 150~250kHz characteristic frequency band ≤ 20%, signal-to-noise ratio of leakage signal detected by distributed acoustic wave sensing system < 5dB, and no signs of new crack propagation in the subsequent 100 temperature differential cycles. The collected post-repair status data are compared and analyzed against the above pre-defined criteria for successful repair: if all indicators meet the criteria, the repair effect is deemed satisfactory; if any indicator fails to meet the criteria, the repair effect is deemed unsatisfactory.

[0131] Furthermore, if the repair results meet the standards, the risk prediction model is iteratively optimized based on the repair results.

[0132] Specifically, when the repair effect meets the standards, the iterative optimization process of the risk prediction model is initiated. First, all repair process data is compiled, including original damage state data, repair method parameters, post-repair state data, and effect evaluation results, forming a complete repair log and updating the health record of the flexible composite pipe segment, supplementing corrected values ​​for key parameters such as interface bonding strength and remaining life. Then, this repair-related data is incorporated into the model training dataset. Using a federated learning framework, edge nodes use the updated local data for fine-tuning the basic model, uploading only the optimized model parameters to the central server. The central server aggregates and optimizes the parameters of each node, updating the material constants, feature weights, and damage evolution equation parameters in the model, enabling the risk prediction model to adapt to changes in the pipe's performance after repair, further improving the prediction accuracy of stress cracking probability and remaining life.

[0133] Furthermore, if the repair effect is not up to standard and the maximum number of repair attempts has not been reached, the graded repair strategy is adjusted and the repair operation is repeated; if the repair effect is not up to standard and the maximum number of repair attempts has been reached, the flexible composite pipe is marked as high-risk to obtain the marked flexible composite pipe.

[0134] Specifically, based on the maximum number of repair attempts (preset to 3), cases of substandard repair results are handled in a tiered manner: if the repair results are substandard and the maximum number of repair attempts (3) has not been reached, the tiered repair strategy is upgraded and adjusted. That is, the original Level 1 pre-embedded microcapsule self-repair is upgraded to Level 2 shape memory polymer patch active sealing repair; the original Level 2 repair is upgraded to Level 3 in-pipe repair robot emergency sealing repair. After the adjustment, the repair operation and subsequent monitoring and effect evaluation process are re-executed.

[0135] If the repair effect is not up to standard and the maximum number of repair attempts has been reached (3), it indicates that the existing graded repair method cannot meet the repair needs of the pipe section. The flexible composite pipe will be marked as high-risk and included in the high-risk pipe section management list. The subsequent monitoring frequency will be increased from 1Hz to 5Hz. At the same time, a manual intervention alarm will be triggered to notify relevant operation and maintenance personnel to carry out on-site inspection, pipe section replacement and other disposal measures in a timely manner to ensure the safe operation of the pipeline.

[0136] This application's embodiments employ a post-repair continuous monitoring and effect evaluation mechanism to ensure verifiable repair results, addressing the problem of traditional repair methods that emphasize execution but neglect verification. This avoids secondary failures due to substandard repairs and improves the repair success rate. After successful repairs, iterative model optimization supplements the training dataset with repair data, allowing the risk prediction model to continuously adapt to changes in pipe performance, gradually improving prediction accuracy and forming a virtuous cycle of repair-verification-optimization. Upgrading repair strategies and imposing maximum attempt constraints when results are not met prevents ineffective repeated repairs. Simultaneously, high-risk labeling and increased monitoring frequency mechanisms minimize pipeline operational safety hazards, achieving closed-loop management throughout the flexible composite pipe's lifecycle and further ensuring service safety in extreme environments.

[0137] While this application provides method operation steps as shown in the embodiments or flowcharts, more or fewer operation steps may be included based on conventional or non-inventive labor. The order of steps listed in this embodiment is merely one possible execution order among many and does not represent the only execution order. In actual device or client product execution, the method can be executed sequentially according to this embodiment or the accompanying drawings, or in parallel (e.g., in a parallel processor or multi-threaded processing environment).

[0138] like Figure 3 As shown in the illustration, this application also provides a repair device 300 for flexible composite pipes. The device includes:

[0139] Analysis module 301 is used to analyze and process the service environment, material combination and load characteristics of flexible composite pipe based on the coupled accelerated aging test field to obtain monitoring data. The coupled accelerated aging test field is a five-field coupled accelerated aging test field of temperature difference, pressure, medium, humidity and light. The monitoring data includes temperature time series data, microscopic image data, acoustic emission signal data and material property parameters.

[0140] Module 302 is used to divide the failure process of flexible composite pipe into four consecutive failure stages based on monitoring data: thermal stress accumulation stage, interface phase reconstruction stage, matrix microcrack bifurcation stage, and instability fracture stage. It determines the failure critical threshold and failure criteria corresponding to each failure stage, and obtains the damage evolution equation corresponding to each failure stage based on energy dissipation theory, and establishes a damage evolution model.

[0141] The construction module 302 is also used to construct a coupled constitutive model based on the damage evolution model, and to perform finite element embedding and simulation verification on the coupled constitutive model to obtain an effective coupled constitutive model.

[0142] The processing module 303 is used to perform feature extraction processing on temperature time series data, microscopic image data, acoustic emission signal data and material property parameters to obtain feature parameters.

[0143] The processing module 303 is also used to construct a dual-branch network model based on the effective coupling constitutive model as the global model, integrate the dual-branch features in the dual-branch network model and combine the feature parameters to train the model, and obtain a risk prediction model for outputting stress cracking probability and remaining life. The first branch network is a graph neural network used to extract microscopic image features and material property features, and the second branch network is a Transformer network used to extract temperature time series features and acoustic emission signal features.

[0144] The processing module 303 is also used to perform predictive processing on the flexible composite pipe based on the risk prediction model, obtain the prediction results, and determine the graded repair strategy based on the prediction results.

[0145] The processing module 303 is also used to repair the flexible composite pipe based on a graded repair strategy to obtain the repaired flexible composite pipe.

[0146] In some embodiments, the analysis module 301 is further configured to analyze and process the monitoring data to obtain multiple failure periods, including the thermal stress accumulation period, the interface phase reconstruction period, the matrix microcrack bifurcation period, and the instability fracture period.

[0147] The analysis module 301 is also used to acquire temperature time series data, microscopic image data, acoustic emission signal data, and failure critical thresholds corresponding to material property parameters.

[0148] The processing module 303 is also used to quantify multiple failure periods based on the failure critical threshold to obtain failure criteria corresponding to multiple failure periods.

[0149] The processing module 303 is also used to fit the failure criteria, temperature time series data, microscopic image data, acoustic emission signal data and material property parameters to obtain damage evolution models corresponding to multiple failure periods.

[0150] In some embodiments, the processing module 303 is further configured to perform wavelet packet decomposition and entropy calculation on the temperature time series data, and extract energy entropy, sample entropy, approximate entropy, maximum Lyapunov exponent and adjacent temperature difference change rate to obtain temperature time series features.

[0151] The processing module 303 is also used to perform convolutional neural network analysis on the microscopic image data to extract crack length, width ratio, interface debonding area ratio and fiber-matrix contact point number to obtain microscopic image features.

[0152] The processing module 303 is also used to perform short-time Fourier transform processing on the acoustic emission signal data, extract the energy ratio of the characteristic frequency band, the ringing count rate and the rise time, and obtain the acoustic emission signal characteristics.

[0153] The processing module 303 is also used to organize and analyze the interfacial bonding strength, thermal expansion coefficient difference, and viscoelastic relaxation modulus in the material property parameters to obtain the material property characteristics.

[0154] The processing module 303 is also used to integrate and process temperature time-series features, microscopic image features, acoustic emission signal features, and material property features to obtain feature parameters.

[0155] In some embodiments, the processing module 303 is further configured to perform training processing on the effectively coupled constitutive model to obtain trained model parameters.

[0156] The processing module 303 is also used to perform aggregation and optimization processing on the trained model parameters to obtain the global model base.

[0157] The construction module 302 is also used to construct a dual-branch network model, which includes a first branch network and a second branch network. The first branch network is used to extract microscopic image features and material property features, and the second branch network is used to extract temperature time series features and acoustic emission signal features.

[0158] The processing module 303 is also used to configure training parameters and perform model training processing on the dual-branch network model and the global model base to obtain the risk prediction model.

[0159] The analysis module 301 is also used to acquire real-time environmental data and material status data of the flexible composite pipe.

[0160] The processing module 303 is also used to input real-time environmental data and material state data into the risk prediction model for prediction processing to obtain prediction results.

[0161] The processing module 303 is also used to classify the prediction results to obtain the damage level.

[0162] The processing module 303 is also used to perform repair method matching based on the damage level to obtain a graded repair strategy. The graded repair strategy includes pre-embedded microcapsule autonomous repair matching, shape memory polymer patch active sealing matching, and in-tube repair robot emergency sealing matching.

[0163] In some embodiments, the processing module 303 is further configured to continuously monitor the repaired area of ​​the repaired flexible composite pipe and collect post-repair status data related to strain, acoustic emission, temperature, and leakage.

[0164] The processing module 303 is also used to compare and analyze the repaired flexible composite pipe with the preset repair success criteria to obtain the repair effect.

[0165] The processing module 303 is also used to iteratively optimize the risk prediction model based on the repair effect when the repair effect meets the standard.

[0166] The processing module 303 is also used to adjust the graded repair strategy and re-execute the repair operation when the repair effect is unsatisfactory and the maximum number of repair attempts has not been reached. When the repair effect is unsatisfactory and the maximum number of repair attempts has been reached, the flexible composite pipe is marked as high-risk to obtain a marked flexible composite pipe.

[0167] Some modules in the apparatus described in this application can be described in the general context of computer-executable instructions that are executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, classes, etc., that perform a specific task or implement a specific abstract data type. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0168] The apparatus or module described in the above embodiments can be implemented by a computer chip or physical entity, or by a product with a certain function. For ease of description, the above apparatus is described by dividing it into various modules according to their functions. When implementing the embodiments of this application, the functions of each module can be implemented in one or more software and / or hardware. Of course, a module that implements a certain function can also be implemented by combining multiple sub-modules or sub-units.

[0169] The methods, apparatus, or modules described in this application can be implemented in a computer-readable program code manner. The controller can be implemented in any suitable manner, such as a microprocessor or processor and a computer-readable medium storing computer-readable program code (e.g., software or firmware) executable by the (micro)processor, logic gates, switches, application-specific integrated circuits (ASICs), programmable logic controllers, and embedded microcontrollers. Examples of controllers include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicon Labs C8051F320. A memory controller can also be implemented as part of the control logic of a memory. Those skilled in the art will also recognize that, in addition to implementing the controller in purely computer-readable program code manner, the same functionality can be achieved by logically programming the method steps to make the controller take the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, such a controller can be considered a hardware component, and the means included within it for implementing various functions can also be considered as structures within the hardware component. Alternatively, the device used to implement various functions can be viewed as either a software module that implements the method or a structure within a hardware component.

[0170] This application also provides an apparatus, the apparatus comprising: a processor; a memory for storing processor-executable instructions; wherein, when the processor executes the executable instructions, it implements the method described in this application.

[0171] This application also provides a non-volatile computer-readable storage medium storing a computer program or instructions thereon, which, when executed, enables the method described in this application embodiment to be implemented.

[0172] Furthermore, in the various embodiments of the present invention, each functional module can be integrated into a processing module, or each module can exist independently, or two or more modules can be integrated into a single module.

[0173] The aforementioned storage media include, but are not limited to, Random Access Memory (RAM), Read-Only Memory (ROM), Cache, Hard Disk Drive (HDD), or Memory Card. The memory can be used to store computer program instructions.

[0174] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary hardware. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product, or it can be embodied in the process of data migration. The computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, mobile terminal, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0175] The various embodiments described in this specification are presented in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on its differences from other embodiments. All or part of this application can be used in numerous general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, mobile communication terminals, multiprocessor systems, microprocessor-based systems, programmable electronic devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices, etc.

[0176] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of this application.

Claims

1. A method for repairing a flexible composite pipe, characterized in that, include: The service environment, material composition, and load characteristics of flexible composite pipes are analyzed and processed based on the coupled accelerated aging test field to obtain monitoring data. The coupled accelerated aging test field is a five-field coupled accelerated aging test field of temperature difference, pressure, medium, humidity, and light. The monitoring data includes temperature time series data, microscopic image data, acoustic emission signal data, and material property parameters. Based on the monitoring data and the failure process of the flexible composite pipe, four consecutive failure periods are divided: thermal stress accumulation period, interface phase reconstruction period, matrix microcrack bifurcation period, and instability fracture period. The failure critical threshold and failure criteria corresponding to each failure period are determined. The damage evolution equation corresponding to each failure period is obtained by fitting based on energy dissipation theory, and a damage evolution model is established. Based on the damage evolution model, a coupled constitutive model is constructed. The coupled constitutive model is then subjected to finite element embedding and simulation verification to obtain an effective coupled constitutive model. Feature extraction processing is performed on the temperature time series data, the microscopic image data, the acoustic emission signal data, and the material property parameters to obtain feature parameters; A dual-branch network model is constructed based on the effective coupling constitutive model as the global model. After integrating the dual-branch features in the dual-branch network model, the model is trained in combination with the feature parameters to obtain a risk prediction model for outputting stress cracking probability and remaining life. The first branch network is a graph neural network for extracting microscopic image features and material property features, and the second branch network is a Transformer network for extracting temperature time series features and acoustic emission signal features. The flexible composite pipe is predicted based on the risk prediction model to obtain the prediction results, and a graded repair strategy is determined based on the prediction results. The flexible composite pipe is repaired based on the graded repair strategy to obtain the repaired flexible composite pipe.

2. The method according to claim 1, characterized in that, Based on the monitoring data and the failure process of the flexible composite pipe, four consecutive failure stages are divided: thermal stress accumulation stage, interface phase reconstruction stage, matrix microcrack bifurcation stage, and instability fracture stage. The failure critical threshold and failure criteria corresponding to each failure stage are determined. Based on energy dissipation theory, the damage evolution equations corresponding to each failure stage are fitted, and a damage evolution model is established, including: The monitoring data is analyzed and processed to obtain multiple failure periods, which include thermal stress accumulation period, interface phase reconstruction period, matrix microcrack bifurcation period and instability fracture period. Acquire the failure critical thresholds corresponding to the temperature time series data, the microscopic image data, the acoustic emission signal data, and the material property parameters, respectively; Based on the aforementioned failure threshold, multiple failure periods are quantified to obtain failure criteria corresponding to multiple failure periods. The failure criteria, the temperature time series data, the microscopic image data, the acoustic emission signal data, and the material property parameters are fitted to obtain damage evolution models corresponding to multiple failure periods.

3. The method according to claim 1, characterized in that, The material property parameters include interfacial bond strength, difference in thermal expansion coefficients, and viscoelastic relaxation modulus. Feature extraction processing is performed on the temperature time-series data, the microscopic image data, the acoustic emission signal data, and the material property parameters. The feature parameters are obtained, including: The temperature time series data is processed by wavelet packet decomposition and entropy calculation to extract energy entropy, sample entropy, approximate entropy, maximum Lyapunov exponent and adjacent temperature difference change rate to obtain temperature time series features. The microscopic image data is analyzed and processed by a convolutional neural network to extract crack length, width ratio, interface debonding area ratio, and fiber-matrix contact point number to obtain microscopic image features. The acoustic emission signal data is processed by short-time Fourier transform to extract the energy proportion of characteristic frequency bands, ringing count rate and rise time to obtain the acoustic emission signal characteristics; The interfacial bonding strength, difference in thermal expansion coefficients, and viscoelastic relaxation modulus of the material properties are processed and analyzed to obtain the material property characteristics. The temperature time-series features, the microscopic image features, the acoustic emission signal features, and the material property features are integrated and processed to obtain feature parameters.

4. The method according to claim 1, characterized in that, The process involves constructing a dual-branch network model based on the effectively coupled constitutive model as the global model foundation, integrating the dual-branch features from the dual-branch network model, and then training the model using the feature parameters to obtain a risk prediction model for outputting stress cracking probability and remaining life, including: The effective coupled constitutive model is trained to obtain the trained model parameters; The trained model parameters are aggregated and optimized to obtain the global model base. A dual-branch network model is constructed, which includes a first branch network and a second branch network. The first branch network is used to extract microscopic image features and material property features, and the second branch network is used to extract temperature time series features and acoustic emission signal features. The risk prediction model is obtained by configuring training parameters and training the dual-branch network model and the global model.

5. The method according to claim 1, characterized in that, The process of performing prediction processing on the flexible composite pipe based on the risk prediction model to obtain prediction results, and determining a graded repair strategy based on the prediction results, includes: Acquire real-time environmental data and material status data of the flexible composite pipe; The real-time environmental data and material state data are input into the risk prediction model for prediction processing to obtain the prediction result. The prediction results are divided into categories to obtain the damage levels; Based on the damage level, a repair method matching process is performed to obtain a graded repair strategy. The graded repair strategy includes pre-embedded microcapsule autonomous repair matching, shape memory polymer patch active sealing matching, and in-tube repair robot emergency sealing matching.

6. The method according to claim 1, characterized in that, After repairing the flexible composite pipe based on the graded repair strategy to obtain the repaired flexible composite pipe, the process further includes: The repaired area of ​​the repaired flexible composite pipe is continuously monitored, and post-repair status data related to strain, acoustic emission, temperature, and leakage are collected. The repaired flexible composite pipe was compared and analyzed with the preset repair success criteria to obtain the repair effect; If the repair effect meets the target, the risk prediction model is iteratively optimized based on the repair effect; Alternatively, if the repair effect is unsatisfactory and the maximum number of repair attempts has not been reached, the graded repair strategy is adjusted and the repair operation is re-executed; if the repair effect is unsatisfactory and the maximum number of repair attempts has been reached, the flexible composite pipe is marked as high-risk to obtain a marked flexible composite pipe.

7. The method according to claim 6, characterized in that, The maximum number of repair attempts is 3.

8. A repair device for a flexible composite pipe, characterized in that, include: The analysis module is used to analyze and process the service environment, material combination and load characteristics of flexible composite pipe based on the coupled accelerated aging test field to obtain monitoring data. The coupled accelerated aging test field is a five-field coupled accelerated aging test field of temperature difference-pressure-medium-humidity-light. The monitoring data includes temperature time series data, microscopic image data, acoustic emission signal data and material property parameters. The module is used to divide the failure process of the flexible composite pipe into four consecutive failure periods: thermal stress accumulation period, interface phase reconstruction period, matrix microcrack bifurcation period, and instability fracture period, based on the monitoring data and the failure process of the flexible composite pipe. It determines the failure critical threshold and failure criteria corresponding to each failure period, and obtains the damage evolution equation corresponding to each failure period based on energy dissipation theory, and establishes a damage evolution model. The construction module is also used to construct a coupled constitutive model based on the damage evolution model, and to perform finite element embedding and simulation verification on the coupled constitutive model to obtain an effective coupled constitutive model; The processing module is used to perform feature extraction processing on the temperature time series data, the microscopic image data, the acoustic emission signal data, and the material property parameters to obtain feature parameters; The processing module is further configured to construct a dual-branch network model based on the effective coupling constitutive model as the global model basis, integrate the dual-branch features in the dual-branch network model and combine them with the feature parameters to train the model, thereby obtaining a risk prediction model for outputting stress cracking probability and remaining life, wherein the first branch network is a graph neural network for extracting microscopic image features and material property features, and the second branch network is a Transformer network for extracting temperature time series features and acoustic emission signal features; The processing module is also used to perform prediction processing on the flexible composite pipe based on the risk prediction model, obtain prediction results, and determine a graded repair strategy based on the prediction results; The processing module is also used to repair the flexible composite pipe based on the graded repair strategy to obtain the repaired flexible composite pipe.

9. A computer device comprising a memory and a processor, the memory storing a computer program executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 7.

Images