A data-driven material allowable boundary construction method, device, equipment and medium
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA NATIONAL OFFSHORE OIL (CHINA) CO LTD
- Filing Date
- 2026-04-08
- Publication Date
- 2026-06-19
Smart Images
Figure FT_1 
Figure FT_2 
Figure FT_3
Abstract
Description
Technical Field
[0001] This invention relates to the field of material service safety assessment, and more particularly to a data-driven method, apparatus, equipment, and medium for constructing material allowable boundaries. Background Technology
[0002] With the development of science and technology, the technology for assessing the safety of materials in service is constantly improving.
[0003] In harsh environments such as supercritical carbon dioxide, the corrosion behavior of stainless steel is influenced by the coupled effects of multiple factors, including carbon dioxide driving force parameters, hydrogen sulfide concentration, and temperature. In engineering applications, it is often necessary to determine the "usable / unusable" boundary or "allowable limit" of the material within a given operating condition range to guide material selection, operating parameter setting, and risk warning.
[0004] When determining the allowable boundaries of materials, related technologies often rely on empirical thresholds, extrapolation from single-factor experiments, or empirical formulas fitted based on a small number of discrete experimental points, which have low reliability. Summary of the Invention
[0005] This invention provides a data-driven method, apparatus, device, and medium for constructing material allowable boundaries, in order to address the shortcomings of low reliability in related technologies for determining material allowable boundaries and enhance the reliability of material allowable boundaries.
[0006] In a first aspect, the present invention provides a data-driven method for constructing allowable material boundaries, comprising:
[0007] The corrosion parameter value space of the target material under operating conditions is coarsely divided into multiple operating condition grids; wherein each operating condition grid includes multiple operating condition vertices; Based on each of the stated operating conditions, the set parameter perturbation range, and the trained corrosion rate prediction model, predict the robust corrosion rate of each of the stated operating conditions. Based on the corrosion rate threshold corresponding to at least one corrosion hazard level and the robust corrosion rate of each of the operating conditions, an initial parameter allowable boundary is generated for each of the corrosion rate thresholds. For any of the corrosion rate thresholds corresponding to the initial parameter allowable boundary, the plurality of working condition meshes are locally refined according to the corrosion rate thresholds to obtain the corresponding plurality of refined working condition meshes. The parameter allowable boundary is iteratively updated according to the plurality of refined working condition meshes to obtain the final parameter allowable boundary.
[0008] Optionally, the corrosion parameter value space consists of multiple corrosion parameter value ranges, and each working condition vertex includes the value of each corrosion parameter.
[0009] Optionally, the parameter perturbation range includes the value perturbation range of each corrosion parameter; The step of predicting the robust corrosion rate for each of the operating conditions based on each operating condition vertex, the set parameter perturbation range, and the trained corrosion rate prediction model includes: For any given working condition vertex, the value of each corrosion parameter in the parameter perturbation interval is used to perform uncertainty perturbation on the value of each corrosion parameter in the working condition vertex, resulting in multiple perturbation samples. The corrosion rate prediction model is used to predict the corrosion rate of each perturbation sample, resulting in a predicted corrosion rate for each perturbation sample. The predicted corrosion rates of each perturbation sample are aggregated to obtain the robust corrosion rate of the working condition vertex.
[0010] Optionally, the step of generating an initial parameter allowable boundary corresponding to each corrosion rate threshold based on at least one corrosion hazard level and the robust corrosion rate of each of the operating conditions includes: Based on the value of each corrosion parameter in each working condition vertex and the robust corrosion rate of each working condition vertex, a functional relationship between the corrosion rate and each corrosion parameter is fitted. For any of the corrosion rate thresholds, the value boundaries of each corrosion parameter are solved according to the corrosion rate threshold and the functional relationship, and the value boundaries of each corrosion parameter as a whole are used as the initial parameter allowable boundary corresponding to the corrosion rate threshold.
[0011] Optionally, the step of locally refining the plurality of working condition meshes according to the corrosion rate threshold to obtain corresponding plurality of refined working condition meshes includes: For any given working condition mesh, if among the robust corrosion rates of each working condition vertex of the working condition mesh, there exists a robust corrosion rate greater than the corrosion rate threshold and a robust corrosion rate less than the corrosion rate threshold, then the working condition mesh is determined as a boundary candidate mesh. Each of the boundary candidate meshes is further subdivided into finer meshes to obtain multiple finer condition meshes.
[0012] Optionally, the step of iteratively updating the allowable parameter boundaries based on the plurality of refined condition meshes to obtain the final allowable parameter boundaries includes: Based on each refined working condition vertex in each refined working condition mesh, the parameter perturbation range, and the corrosion rate prediction model, predict the robust corrosion rate of each refined working condition vertex. Based on the values of each corrosion parameter in each of the working condition vertices and each of the refined working condition vertices, and the robust corrosion rate of each of the working condition vertices and each of the refined working condition vertices, a new functional relationship between the corrosion rate and each corrosion parameter is fitted. Based on the corrosion rate threshold and the new functional relationship, the new value boundary of each corrosion parameter is solved, and the whole is used as the new parameter allowable boundary corresponding to the corrosion rate threshold; For any of the aforementioned working condition meshes or the refined working condition meshes, if there is a robust corrosion rate greater than the corrosion rate threshold and a robust corrosion rate less than the corrosion rate threshold among the robust corrosion rates of each working condition vertex of the target mesh, then the target mesh is determined as a boundary candidate mesh until the iteration termination condition is met, and the final allowable parameter boundary is obtained.
[0013] Optionally, the plurality of corrosion parameters may include multiple parameters such as temperature, hydrogen sulfide concentration, carbon dioxide fugacity, pressure, chloride ion concentration, and medium flow rate.
[0014] In a second aspect, the present invention provides a data-driven material allowable boundary construction apparatus, comprising: The coarse meshing unit is used to coarsely divide the corrosion parameter value space of the target material under operating conditions to obtain multiple operating condition meshes; wherein, each operating condition mesh includes multiple operating condition vertices; The prediction unit is used to predict the robust corrosion rate of each of the said working conditions based on each working condition vertex, the set parameter perturbation range and the trained corrosion rate prediction model. The generation unit is used to generate an initial parameter allowable boundary corresponding to each corrosion rate threshold based on a corrosion rate threshold corresponding to at least one corrosion hazard level and a robust corrosion rate for each of the operating conditions. A local refinement unit is used to locally refine the plurality of working condition meshes according to the corrosion rate threshold for any initial parameter allowable boundary corresponding to the corrosion rate threshold, so as to obtain the corresponding plurality of refined working condition meshes. The update unit is used to iteratively update the allowable parameter boundaries based on the multiple refined condition meshes to obtain the final allowable parameter boundaries.
[0015] Thirdly, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to perform the data-driven material allowable boundary construction method of the first aspect or any corresponding embodiment described above.
[0016] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to perform the data-driven material allowable boundary construction method described in the first aspect or any corresponding embodiment thereof.
[0017] This invention provides a data-driven method, apparatus, device, and medium for constructing allowable material boundaries. It can coarsely divide the corrosion parameter value space of a target material under operating conditions into multiple operating condition meshes, each including multiple operating condition vertices. Based on each operating condition vertex, a set parameter perturbation range, and a trained corrosion rate prediction model, the robust corrosion rate of each operating condition vertex is predicted. Based on a corrosion rate threshold corresponding to at least one corrosion hazard level and the robust corrosion rate of each operating condition vertex, an initial allowable parameter boundary corresponding to each corrosion rate threshold is generated. For any initial allowable parameter boundary corresponding to a corrosion rate threshold, the multiple operating condition meshes are locally refined according to the corrosion rate threshold, resulting in multiple refined operating condition meshes. The allowable parameter boundary is iteratively updated based on the multiple refined operating condition meshes to obtain the final allowable parameter boundary. This invention can improve the efficiency and accuracy of material allowable boundary construction and enhance the reliability of the material allowable boundary.
[0018] Compared with related technologies, the present invention has at least the following beneficial effects: (1) The corrosion rate prediction output is transformed into a robust corrosion rate field in the variable space, and contour lines / contour surfaces are automatically extracted based on a multi-threshold set to realize the automatic construction of allowable material boundaries and multi-level risk zoning; (2) By introducing uncertainty disturbances and using mean or quantile aggregation, desired boundaries and conservative boundaries can be formed, making the boundaries more robust to measurement errors and operating condition fluctuations. (3) Adaptive encrypted search is adopted, and the calculation is only refined for the candidate boundary regions, which improves the efficiency of boundary construction and enhances the accuracy of the boundary. (4) Combining detection devices and update triggering mechanisms, it enables boundary version release and dynamic reconstruction, which can continuously evolve with new data and working conditions, and is suitable for online risk warning and closed-loop optimization scenarios. Attached Figure Description
[0019] To more clearly illustrate the technical solutions in this invention or related technologies, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0020] Figure 1A flowchart of a first data-driven method for constructing allowable material boundaries provided in an embodiment of the present invention; Figure 2 A flowchart of a second data-driven method for constructing allowable material boundaries provided in an embodiment of the present invention; Figure 3 A flowchart of training and verification of a corrosion rate prediction model provided in an embodiment of the present invention; Figure 4 A flowchart of three-variable spatial grid construction and uncertainty perturbation reconstruction provided in an embodiment of the present invention; Figure 5 A flowchart for multi-threshold risk zoning and material allowable boundary extraction provided in this embodiment of the invention; Figure 6 A flowchart of adaptive encryption search and boundary refinement provided in an embodiment of the present invention; Figure 7 This is a flowchart of an online detection and dynamic update process provided in an embodiment of the present invention; Figure 8 This is a schematic diagram of a data-driven material allowable boundary construction device provided in an embodiment of the present invention; Figure 9 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. Detailed Implementation
[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention 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.
[0022] When determining the allowable boundary of materials, related technologies often rely on empirical thresholds, extrapolation from single-factor experiments, or empirical formulas fitted based on a small number of discrete experimental points. This makes it difficult to form a continuous boundary surface in a multivariable space. Furthermore, under conditions of measurement error, operating condition fluctuation, and sparse samples, the boundary results are prone to instability, affecting the consistency of safety margin judgment and decision-making.
[0023] With the development of data-driven technologies, machine learning models can improve the ability to fit corrosion rates to some extent. However, these technologies often use model outputs for single-point predictions or only for static plotting, lacking a general construction mechanism to automatically transform discrete predictions into multi-level risk partitions and boundary surfaces. Furthermore, these technologies do not incorporate sensor errors and operating condition fluctuations into the boundary generation process, making the boundaries sensitive to input disturbances. On the other hand, in online applications, operating conditions change over time and may experience distributional drift. Without a closed-loop mechanism for model updates and boundary reconstruction, the boundaries cannot continuously evolve with new data, thus limiting engineering usability and long-term reliability.
[0024] Therefore, there is an urgent need for a data-driven method that can automatically construct the allowable boundary of materials under target working conditions and can be dynamically updated by combining online monitoring, so as to improve the continuity, robustness and engineering applicability of material safety assessment.
[0025] The following is combined Figures 1-7 This invention describes a data-driven method for constructing allowable material boundaries.
[0026] like Figure 1 As shown, this embodiment proposes a first data-driven method for constructing material allowable boundaries, which may include the following steps: S101. Perform coarse meshing on the corrosion parameter value space of the target material under operating conditions to obtain multiple operating condition meshes; wherein, each operating condition mesh includes multiple operating condition vertices.
[0027] The target material can be stainless steel or other types of materials that work in harsh environments such as supercritical carbon dioxide.
[0028] Specifically, the operating conditions can be the operating conditions under a specific service environment or evaluation environment of the target material. These conditions can be operating conditions under real service scenarios or laboratory simulation conditions used for material evaluation. This embodiment does not limit the specific conditions.
[0029] Optionally, the corrosion parameter value space consists of multiple corrosion parameter value ranges, with each working condition vertex including the value of each corrosion parameter.
[0030] Optionally, multiple corrosion parameters may be selected from temperature, hydrogen sulfide concentration, carbon dioxide fugacity, pressure, chloride ion concentration, and medium flow rate.
[0031] Among them, the corrosion parameters of the target material under operating conditions are the relevant parameters that affect the corrosion rate of the target material, namely corrosion rate-related parameters, such as temperature, pressure and carbon dioxide fugacity.
[0032] Specifically, the corrosion parameter value space can include the value ranges of multiple corrosion parameters of the target material. For example, when the corrosion parameters include carbon dioxide fugacity, hydrogen sulfide concentration, and temperature, the value ranges of carbon dioxide fugacity, hydrogen sulfide concentration, and temperature can be mapped to the same coordinate system. The corrosion parameter value space can be a cuboid value range formed by the value ranges of carbon dioxide fugacity, hydrogen sulfide concentration, and temperature in this coordinate system.
[0033] Specifically, in this embodiment, for the value range of each corrosion parameter in the corrosion parameter value space, the value range of the corrosion parameter can be divided into multiple continuous sub-ranges of equal length according to a set step size. For example, according to a set step size of 3, the value range of 6 to 20 can be divided into multiple continuous sub-ranges of equal length. It should be noted that this embodiment can also divide the value range of the corrosion parameter into multiple sub-ranges according to the discretization rules set by the technician.
[0034] Each working condition grid includes sub-ranges of values for various corrosion parameters. For example, when the corrosion parameter value space is the aforementioned cuboid value range, the working condition grid can be a cuboid grid obtained by dividing each edge of the cuboid value range.
[0035] Specifically, the operating condition vertex includes the combination of values for various corrosion parameters. For example, when the corrosion parameters include carbon dioxide fugacity, hydrogen sulfide concentration, and temperature, a certain operating condition vertex can be represented as "carbon dioxide fugacity is 2.0 MPa, hydrogen sulfide concentration is 1500 ppm, and temperature is 80℃".
[0036] S102. Based on each working condition vertex, the set parameter perturbation range, and the trained corrosion rate prediction model, predict the robust corrosion rate of each working condition vertex.
[0037] Specifically, the parameter perturbation range is the range in which uncertain perturbations are applied to the values of corrosion parameters at the peak of the operating condition.
[0038] Specifically, in this embodiment, a trained corrosion rate prediction model can be used to predict the robust corrosion rate of each vertex in each working condition grid.
[0039] Optionally, the parameter perturbation range includes the perturbation range for the value of each corrosion parameter. Step S102 includes: For any working condition vertex, the value of each corrosion parameter in the parameter perturbation interval is used to perturb the value of each corrosion parameter in the working condition vertex with uncertainty, resulting in multiple perturbation samples. The corrosion rate prediction model is used to predict the corrosion rate of each perturbation sample, resulting in the predicted corrosion rate of each perturbation sample. The predicted corrosion rates of each perturbation sample are aggregated to obtain the robust corrosion rate of the working condition vertex.
[0040] Specifically, the corrosion rate prediction model can be obtained by training the prediction model to be trained using training data and corresponding labels in this embodiment. The training data can include the value of each corrosion parameter, and the label can be the actual corrosion rate corresponding to the training data.
[0041] The robust corrosion rate of the operating condition vertex is obtained by aggregating the predicted corrosion rates of each perturbation sample of that operating condition vertex.
[0042] Specifically, the aggregation method can be mean aggregation or quantile aggregation, or other aggregation methods; this embodiment does not limit it.
[0043] S103. Based on the corrosion rate threshold corresponding to at least one corrosion hazard level and the robust corrosion rate of each operating condition peak, generate the allowable boundary of initial parameters corresponding to each corrosion rate threshold.
[0044] Specifically, in this embodiment, technicians can set corrosion rate thresholds corresponding to different corrosion hazard levels according to actual needs.
[0045] Specifically, in this embodiment, the allowable boundaries of initial parameters corresponding to each corrosion rate threshold can be generated based on each corrosion rate threshold and the robust corrosion rate at each operating condition peak.
[0046] Optionally, step S103 includes: Based on the value of each corrosion parameter in each working condition vertex and the robust corrosion rate of each working condition vertex, a functional relationship between the corrosion rate and each corrosion parameter is fitted. For any corrosion rate threshold, the value boundary of each corrosion parameter is solved based on the corrosion rate threshold and the functional relationship. The value boundary of each corrosion parameter as a whole is used as the initial parameter allowable boundary corresponding to the corrosion rate threshold.
[0047] Specifically, this embodiment can use the corrosion parameter values and robust corrosion rate at each working condition vertex to fit a functional relationship between the corrosion rate and the corrosion parameters. Then, this embodiment can use this functional relationship to solve for the value boundaries of each corrosion parameter, given a determined corrosion rate threshold, and use this as the initial allowable parameter boundary corresponding to the corrosion rate threshold.
[0048] S104. For any allowable boundary of initial parameters corresponding to a corrosion rate threshold, locally refine multiple working condition meshes according to the corrosion rate threshold to obtain multiple refined working condition meshes.
[0049] Specifically, in this embodiment, multiple working condition grids in the corrosion parameter value space can be locally refined according to the allowable boundary of the initial parameters corresponding to the corrosion rate threshold, so as to obtain multiple refined working condition grids.
[0050] Optionally, step S104 includes: For any given working condition mesh, if there exists a robust corrosion rate greater than the corrosion rate threshold and a robust corrosion rate less than the corrosion rate threshold among the robust corrosion rates of each working condition vertex of the working condition mesh, then the working condition mesh is determined as a boundary candidate mesh. Each boundary candidate mesh is further subdivided into finer meshes to obtain multiple finer condition meshes.
[0051] Specifically, for an initial parameter allowable boundary corresponding to a certain corrosion rate threshold, for any operating condition mesh, if there are operating condition vertices in the operating condition mesh with a robust corrosion rate greater than the corrosion rate threshold, and also operating condition vertices with a robust corrosion rate less than the corrosion rate threshold, then the operating condition mesh is determined as a boundary candidate mesh, that is, a boundary candidate mesh corresponding to the initial parameter allowable boundary. In this embodiment, after determining each boundary candidate mesh corresponding to the initial parameter allowable boundary, each boundary candidate mesh can be further subdivided into multiple refined operating condition meshes.
[0052] S105. Iteratively update the allowable parameter boundary based on multiple refined process condition meshes to obtain the final allowable parameter boundary.
[0053] Specifically, in this embodiment, for any initial parameter allowable boundary corresponding to any corrosion rate threshold, after dividing each boundary candidate grid corresponding to the initial parameter allowable boundary into multiple refined working condition grids, the initial parameter allowable boundary is iteratively updated for each refined working condition grid to construct the final parameter allowable boundary.
[0054] Optionally, step S105 includes: Based on each refined working condition vertex, parameter perturbation range, and corrosion rate prediction model in each refined working condition mesh, predict the robust corrosion rate of each refined working condition vertex. Based on the values of each corrosion parameter in each working condition vertex and each refined working condition vertex, as well as the robust corrosion rate of each working condition vertex and each refined working condition vertex, a new functional relationship between the corrosion rate and each corrosion parameter is fitted. Based on the corrosion rate threshold and the new functional relationship, the new value boundary of each corrosion parameter is solved, and the whole is used as the new parameter allowable boundary corresponding to the corrosion rate threshold; For any working condition mesh or refined working condition mesh, if there is a robust corrosion rate greater than the corrosion rate threshold and a robust corrosion rate less than the corrosion rate threshold among the robust corrosion rates of each working condition vertex of the target mesh, then the target mesh is determined as a boundary candidate mesh until the iteration termination condition is met, and the final allowable boundary of parameters is obtained.
[0055] Specifically, in this embodiment, for any initial parameter allowable boundary corresponding to a corrosion rate threshold, each candidate boundary grid corresponding to the initial parameter allowable boundary can be divided into multiple refined working condition grids. Then, based on each refined working condition vertex, parameter perturbation interval, and corrosion rate prediction model in each refined working condition grid corresponding to the initial parameter allowable boundary, this embodiment can predict the robust corrosion rate of each refined working condition vertex. Then, referring to the aforementioned processing flow, a new functional relationship between the corrosion rate and each corrosion parameter is fitted, and a new parameter allowable boundary is determined based on this new functional relationship, thus updating the initial parameter allowable boundary. Subsequently, this embodiment can continue to update the new parameter allowable boundary until the set iteration stopping condition is met, and the final parameter allowable boundary is obtained.
[0056] The data-driven material allowable boundary construction method proposed in this embodiment can coarsely divide the corrosion parameter value space of the target material under operating conditions into multiple operating condition meshes, each of which includes multiple operating condition vertices. Based on each operating condition vertex, a set parameter perturbation range, and a trained corrosion rate prediction model, the robust corrosion rate of each operating condition vertex is predicted. Based on a corrosion rate threshold corresponding to at least one corrosion hazard level and the robust corrosion rate of each operating condition vertex, an initial parameter allowable boundary corresponding to each corrosion rate threshold is generated. For any initial parameter allowable boundary corresponding to a corrosion rate threshold, the multiple operating condition meshes are locally refined according to the corrosion rate threshold, resulting in multiple refined operating condition meshes. The parameter allowable boundary is iteratively updated based on the multiple refined operating condition meshes to obtain the final parameter allowable boundary. This embodiment can improve the efficiency and accuracy of material allowable boundary construction and enhance the reliability of the material allowable boundary.
[0057] based on Figure 1 This embodiment proposes a second data-driven method for constructing allowable material boundaries, which includes the following steps: S1. Obtain the sample dataset under the target operating condition. The target operating condition is the specific service environment or evaluation environment corresponding to the allowable boundary of the material to be constructed. It can be the operating condition under the actual service scenario or the laboratory simulation condition used for material evaluation. It is characterized by at least carbon dioxide fugacity or its equivalent, hydrogen sulfide concentration and temperature, and may further include at least one of pressure, medium composition, salt content, pH value, flow state, water content and material state, depending on the application.
[0058] S2. Train a corrosion rate prediction model based on training data. The corrosion rate prediction model is used to output corrosion rate prediction values based on an input feature vector that includes at least carbon dioxide fugacity or its equivalent, hydrogen sulfide concentration, and temperature.
[0059] S3. Construct a grid point set within a variable space consisting of carbon dioxide fugacity or its equivalent, hydrogen sulfide concentration, and temperature. The variable space refers to the continuous operating condition parameter domain determined by the aforementioned operating condition variables, and the range of values for each variable can be determined based on the on-site operating range, experimental setting range, or sample data coverage range corresponding to the target operating condition. The grid point set is a discrete set of points formed by dividing the continuous operating condition parameter domain according to a preset discrete step size or discrete rule. A grid point is any point in the grid point set, corresponding to a set of determined operating condition parameter values, and can be represented as a ternary feature vector containing carbon dioxide fugacity or its equivalent, hydrogen sulfide concentration, and temperature.
[0060] For example, in one embodiment, a grid point can be represented as a discrete operating condition point with "carbon dioxide fugacity of 2.0 MPa, hydrogen sulfide concentration of 1500 ppm, and temperature of 80°C", which is used to characterize a specific combination of parameters within the target operating condition range.
[0061] S4. For each grid point in the grid point set, introduce uncertainty perturbation to generate multiple sets of perturbation samples. Input the multiple sets of perturbation samples into the corrosion rate prediction model to obtain multiple sets of corrosion rate prediction values. Then, aggregate the multiple sets of corrosion rate prediction values to obtain the robust corrosion rate value of the grid point, thereby forming a robust corrosion rate field. Among them, the multiple sets of perturbation samples are multiple perturbed feature vectors generated around the operating parameter values of the corresponding grid point, which are used to characterize the corrosion response changes of the grid point under parameter fluctuations, measurement errors or environmental perturbation conditions.
[0062] S5. Obtain the corrosion rate threshold set, divide the robust corrosion rate field into dangerous ranges based on the corrosion rate threshold set, and extract the contour lines or contour surfaces that satisfy the robust corrosion rate field being equal to each corrosion rate threshold, as the allowable material boundary under the corresponding threshold. S6. Identify candidate regions of the allowable material boundary and perform adaptive refinement on the candidate regions to make the mesh resolution of the candidate regions higher than that of the non-candidate regions. Iteratively update the robust corrosion rate field and reconstruct the allowable material boundary until the preset stopping condition is met. S7. Output the allowable boundaries of multi-level materials and the corresponding hazard level range results.
[0063] Specifically, other data-driven methods for constructing material allowable boundaries may also include: S8. Connect the detection device to collect carbon dioxide fugacity or its equivalent characterization, hydrogen sulfide concentration and temperature, and transmit the collected data to the prediction terminal; S9. Based on the collected data, call the corrosion rate prediction model to make inferences, obtain the current corrosion rate prediction value, and combine it with the current version of the multi-level material allowable boundary to determine the corresponding danger level range and output early warning information; S10. When the update trigger condition is met, the corrosion rate prediction model is incrementally updated or retrained, and the multi-level material allowable boundary is automatically reconstructed, a new boundary version is generated and released synchronously.
[0064] Compared with related technologies, this embodiment has at least the following beneficial effects: (1) The corrosion rate prediction output is transformed into a robust corrosion rate field in the variable space, and contour lines / contour surfaces are automatically extracted based on a multi-threshold set to realize the automatic construction of allowable material boundaries and multi-level risk zoning; (2) By introducing uncertainty disturbances and using mean or quantile aggregation, desired boundaries and conservative boundaries can be formed, making the boundaries more robust to measurement errors and operating condition fluctuations. (3) Adaptive encrypted search is adopted, and the calculation is only refined for the candidate boundary regions, which improves the efficiency of boundary construction and enhances the accuracy of the boundary. (4) Combining detection devices and update triggering mechanisms, it enables boundary version release and dynamic reconstruction, which can continuously evolve with new data and working conditions, and is suitable for online risk warning and closed-loop optimization scenarios.
[0065] Explanation of terms and symbols: Carbon dioxide fugacity or its equivalent characterization: a parameter used to characterize the chemical driving force of carbon dioxide under target operating conditions, which can be the fugacity itself or a driving force characterization equivalent to the fugacity; it should be consistent between the training and inference phases.
[0066] Hydrogen sulfide concentration: The concentration of H2S in the target medium environment can be represented as volume fraction, mole fraction, partial pressure, etc. The dimensions should be kept consistent during the training and inference phases.
[0067] Corrosion rate: denoted as CR, representing the corrosion rate of the target material under the target working conditions; the unit is consistent with the threshold set.
[0068] Single-shot corrosion rate prediction: denoted as CR_pred, representing the model's output on a single input feature vector.
[0069] Robust corrosion rate: denoted as CR_robust, representing the robust output of multiple predictions obtained after introducing a disturbance to the same grid point and then aggregating them.
[0070] Grid point: denoted as x_g, representing a discrete sampling point within the variable space (CO2 fugacity or equivalent characterization, H2S concentration, temperature). Perturbation samples can be denoted as x_g(m), m=1…M.
[0071] Threshold set: denoted as {Li}, i=1…n; when i=1…4, it can correspond to L1=0.025, L2=0.055, L3=0.076, L4=0.086 (units are consistent with CR).
[0072] Material allowable boundary: For any threshold Li, the contour lines (slices) or contour surfaces (spatial boundaries) that satisfy “CR_robust = Li”.
[0073] Optimal range and setting recommendations for key parameters: The number of perturbation samples M: To balance robustness and computational cost, M ≥ 10 is preferred; when computational resources allow, it can be 30 to 200; when updating the online boundary, a larger M can be used for points near the boundary and a smaller M can be used for points far from the boundary.
[0074] Disturbance source and cutoff: The disturbance amplitude can be determined based on sensor measurement error, historical fluctuation statistics, experimental repeatability error, etc. After the disturbance is generated, a cutoff constraint can be added to ensure that the CO2 fugacity / equivalent amount and H2S concentration are not negative and fall within the reasonable range of engineering, and the temperature falls within the allowable range of engineering. If there are physical correlation constraints between variables, joint constraints can be introduced.
[0075] Aggregation method and quantile parameter q: Robust output can be achieved by mean aggregation or quantile aggregation; when using quantile aggregation to construct conservative boundaries, q is between 0.8 and 0.95, preferably 0.90.
[0076] Threshold set: The threshold set contains at least two thresholds; the thresholds may be derived from standard specifications, lifetime index conversions, engineering experience, or user configurations, and are recorded together with boundary version metadata.
[0077] (1) Applicable domain determination and rollback strategy: To ensure the engineering reliability of model inference and boundary determination, this embodiment introduces an applicable domain determination and backoff strategy in online inference and boundary update, specifically including: Domain of applicability determination: Using the input feature distribution of the training data as a reference, the range and distribution consistency of the real-time input feature vector (including CO2 fugacity or its equivalent representation, H2S concentration, and temperature) are verified. Domain of applicability determination includes at least one of the following methods: Range determination: When any input variable exceeds the minimum and maximum value range of the corresponding variable in the training data, or exceeds the preset reasonable range, it is determined to be outside the applicable domain; Quantile range determination: When any input variable exceeds the preset quantile range of the variable in the training data; for example, when it is between the 5th and 95th percentiles, it is determined to be outside the applicable domain. Distance determination: Calculate the distance between the input vector and the training samples based on the distance metric of the training data feature space; for example, the distance to the nearest neighbor or the distance to the sample center. When the distance is greater than a preset threshold, it is determined to be outside the applicable domain.
[0078] (2) Rollback strategy: When it is determined that the application domain is exceeded, the system shall execute at least one of the following rollback strategies: Output a status message "Requires review / Unavailable" and record the input and timestamp; A more conservative and robust output method is adopted for risk assessment, namely, switching the robust output from mean aggregation to quantile aggregation, or increasing the quantile parameter q; The sampling / calibration suggestion is triggered, indicating that new corrosion rate calibration data is needed to update the model and boundary versions.
[0079] By employing the aforementioned applicable domain determination and backoff strategies, the cumulative error caused by input distribution drift or abnormal inputs in risk determination can be reduced, thereby improving the availability and security of the system engineering.
[0080] like Figure 2 As shown, other data-driven material allowable boundary construction methods proposed in this embodiment may include the following steps: S201. Data Acquisition and Preprocessing. Acquire a sample dataset under the target operating conditions, which should include at least CO2 fugacity or its equivalent, H2S concentration, temperature, and corresponding corrosion rate (CR). Perform unit unification, missing value processing, outlier processing, and necessary feature filtering on the samples to obtain training data.
[0081] S202, Training and Validation of Corrosion Rate Prediction Model. For example... Figure 3 As shown, a corrosion rate prediction model is trained based on training data, enabling the model to output a single predicted value CR_pred for the input feature vector. The training process employs a training set-validation set partitioning, and optionally uses cross-validation to evaluate the model's generalization performance and parameter stability. Specifically, this includes sample data import, data pre-cleaning and unit unification, feature construction and selection, training statistics generation, training-validation, model training, hyperparameter search / tuning, model performance evaluation, and model and metadata storage.
[0082] S203, Multivariable Spatial Mesh Construction. For example... Figure 4 As shown, in this embodiment, a grid point set {x_g} can be constructed in a variable space consisting of (CO2 fugacity or equivalent characterization, H2S concentration, and temperature), and the grid adopts a coarse-to-fine strategy.
[0083] S204. Uncertainty perturbation reconstruction and aggregation (B1). As Figure 4 shown, for each grid point x_g, M perturbation samples {x_g(m)} (m = 1...M) are generated according to measurement errors and operating condition fluctuations, and are respectively input into the prediction model to obtain a prediction set {CR_pred(m)}; the prediction set is aggregated to obtain a robust corrosion rate value CR_robust. The aggregation method is as follows: a) Mean aggregation: CR_robust = mean(CR_pred(1...M)); b) Quantile aggregation: CR_robust = quantile_q(CR_pred(1...M)), where q ranges from 0.8 to 0.95, preferably 0.90.
[0084] S205. Multi-threshold risk zoning and extraction of material allowable boundaries. As Figure 5 shown, a threshold set {Li} is obtained, the robust corrosion rate field is divided into danger level intervals according to the thresholds, and the contour line or surface that satisfies "CR_robust = Li" is extracted for each threshold as the material allowable boundary corresponding to the threshold.
[0085] In a preferred embodiment, the threshold set takes L1 = 0.025, L2 = 0.055, L3 = 0.076, L4 = 0.086, and five danger level intervals are divided: Interval A: CR_robust ≤ 0.025; it is determined as the first danger level interval when the robust corrosion rate value is not greater than 0.025; Interval B: 0.025 < CR_robust ≤ 0.055; it is determined as the second danger level interval when the robust corrosion rate value is greater than 0.025 and not greater than 0.055; Interval C: 0.055 < CR_robust ≤ 0.076; it is determined as the third danger level interval when the robust corrosion rate value is greater than 0.055 and not greater than 0.076; Interval D: 0.076 < CR_robust ≤ 0.086; it is determined as the fourth danger level interval when the robust corrosion rate value is greater than 0.076 and not greater than 0.086; Interval E: CR_robust > 0.086. It is determined as the fifth danger level interval when the robust corrosion rate value is greater than 0.086.
[0086] S206. Adaptive encryption and boundary refinement. As Figure 6As shown, candidate boundary regions are identified under a coarse mesh, and local mesh refinement is performed. The robust erosion rate field is iteratively updated, and the boundary is reconstructed until a stopping condition is met. One method for candidate region identification is: if at the vertex of a voxel, at least one robust erosion rate value is greater than a threshold and at least one is less than a threshold, then that voxel is a candidate region corresponding to that threshold. The stopping condition may include the change in boundary geometry being less than a threshold, the rate of change of boundary area / volume being less than a threshold, an upper limit on the number of iterations, or an upper limit on resources.
[0087] S207, Output the results of multi-level material allowable boundaries and corresponding hazard level ranges.
[0088] like Figure 7 As shown, this embodiment can also automatically construct material allowable boundaries through the system. This embodiment can include a detection device, communication and edge gateway, and may also include the following processes: Data cleaning and feature construction; online model + boundary inference service; data storage, including raw data / features / logs; drift detection and update triggering; new calibration data, model update / retraining; boundary reconstruction; version control update; risk assessment and early warning output; interactive segments / interfaces, queries / reports / alarms.
[0089] This embodiment trains a corrosion rate prediction model based on sample data under target operating conditions. Carbon dioxide fugacity, hydrogen sulfide concentration, and temperature are used as boundary variables to construct a grid of points within the variable space. For each grid point, measurement errors and uncertainties related to operating condition fluctuations are introduced to generate multiple sets of perturbation samples, which are then input into the prediction model to obtain a set of corrosion rate predictions. These samples are aggregated using the mean or quantiles to obtain a robust corrosion rate field. Based on a preset corrosion rate threshold set, the method automatically divides the danger level range and extracts contour lines or isosurfaces where the corrosion rate equals the threshold as multi-level allowable material boundaries. Furthermore, adaptive densification search iteratively refines the grid in the boundary candidate region, improving boundary construction efficiency and accuracy. In online applications, the method can connect to a detection device to collect and transmit operating condition variables. When new calibrated corrosion rate data is acquired or an input distribution drift is detected, model updates and boundary reconstruction are triggered, enabling versioned release and synchronization of boundaries. This embodiment can solve the problems in related technologies, such as the reliance on discrete experimental points for boundary construction, the difficulty in forming continuous boundaries in multivariable space, the sensitivity to input disturbances, and the lack of online update closed loops. It can automatically construct multi-threshold and multi-level allowable material boundaries under target working conditions, and can output conservative boundaries to enhance the robustness of safety margin determination.
[0090] like Figure 8 As shown, this embodiment proposes a data-driven material allowable boundary construction apparatus, which includes: The coarse meshing element 801 is used to coarsely mesh the corrosion parameter value space of the target material under operating conditions to obtain multiple operating condition meshes; wherein, each operating condition mesh includes multiple operating condition vertices; The prediction unit 802 is used to predict the robust corrosion rate of each working condition vertex based on each working condition vertex, the set parameter perturbation range and the trained corrosion rate prediction model. The generation unit 803 is used to generate the allowable boundary of initial parameters corresponding to each corrosion rate threshold based on the corrosion rate threshold corresponding to at least one corrosion hazard level and the robust corrosion rate of each working condition peak. Local refinement element 804 is used to locally refine multiple working condition meshes according to the corrosion rate threshold for any initial parameter allowable boundary corresponding to any corrosion rate threshold, so as to obtain multiple refined working condition meshes. Update unit 805 is used to iteratively update the allowable parameter boundary based on multiple refined condition meshes to obtain the final allowable parameter boundary.
[0091] It should be noted that the processing procedures and beneficial effects of the coarse division unit 801, prediction unit 802, generation unit 803, local refinement unit 804, and update unit 805 can be referred to respectively. Figure 1 Steps S101 to S105 are not described in detail here.
[0092] Optionally, the corrosion parameter value space consists of multiple corrosion parameter value ranges, with each working condition vertex including the value of each corrosion parameter.
[0093] Optionally, the parameter perturbation range includes the value perturbation range of each corrosion parameter; Prediction unit 802 is also used for: For any working condition vertex, the value of each corrosion parameter in the parameter perturbation interval is used to perturb the value of each corrosion parameter in the working condition vertex with uncertainty, resulting in multiple perturbation samples. The corrosion rate prediction model is used to predict the corrosion rate of each perturbation sample, resulting in the predicted corrosion rate of each perturbation sample. The predicted corrosion rates of each perturbation sample are aggregated to obtain the robust corrosion rate of the working condition vertex.
[0094] Optionally, the generating unit 803 is also used for: Based on the value of each corrosion parameter in each working condition vertex and the robust corrosion rate of each working condition vertex, a functional relationship between the corrosion rate and each corrosion parameter is fitted. For any corrosion rate threshold, the value boundary of each corrosion parameter is solved based on the corrosion rate threshold and the functional relationship. The value boundary of each corrosion parameter as a whole is used as the initial parameter allowable boundary corresponding to the corrosion rate threshold.
[0095] Optionally, the local refinement unit 804 is also used for: For any given working condition mesh, if there exists a robust corrosion rate greater than the corrosion rate threshold and a robust corrosion rate less than the corrosion rate threshold among the robust corrosion rates of each working condition vertex of the working condition mesh, then the working condition mesh is determined as a boundary candidate mesh. Each boundary candidate mesh is further subdivided into finer meshes to obtain multiple finer condition meshes.
[0096] Optionally, update unit 805 is also used for: Based on each refined working condition vertex, parameter perturbation range, and corrosion rate prediction model in each refined working condition mesh, predict the robust corrosion rate of each refined working condition vertex. Based on the values of each corrosion parameter in each working condition vertex and each refined working condition vertex, as well as the robust corrosion rate of each working condition vertex and each refined working condition vertex, a new functional relationship between the corrosion rate and each corrosion parameter is fitted. Based on the corrosion rate threshold and the new functional relationship, the new value boundary of each corrosion parameter is solved, and the whole is used as the new parameter allowable boundary corresponding to the corrosion rate threshold; For any working condition mesh or refined working condition mesh, if there is a robust corrosion rate greater than the corrosion rate threshold and a robust corrosion rate less than the corrosion rate threshold among the robust corrosion rates of each working condition vertex of the target mesh, then the target mesh is determined as a boundary candidate mesh until the iteration termination condition is met, and the final allowable boundary of parameters is obtained.
[0097] Optionally, multiple corrosion parameters may be selected from temperature, hydrogen sulfide concentration, carbon dioxide fugacity, pressure, chloride ion concentration, and medium flow rate.
[0098] The data-driven material allowable boundary construction device proposed in this embodiment can coarsely divide the corrosion parameter value space of the target material under operating conditions into multiple operating condition meshes, each of which includes multiple operating condition vertices. Based on each operating condition vertex, a set parameter perturbation range, and a trained corrosion rate prediction model, the robust corrosion rate of each operating condition vertex is predicted. Based on a corrosion rate threshold corresponding to at least one corrosion hazard level and the robust corrosion rate of each operating condition vertex, an initial parameter allowable boundary corresponding to each corrosion rate threshold is generated. For any initial parameter allowable boundary corresponding to a corrosion rate threshold, the multiple operating condition meshes are locally refined according to the corrosion rate threshold, resulting in multiple refined operating condition meshes. The parameter allowable boundary is iteratively updated based on the multiple refined operating condition meshes to obtain the final parameter allowable boundary. This embodiment can improve the efficiency and accuracy of material allowable boundary construction and enhance the reliability of the material allowable boundary.
[0099] In this embodiment, the data-driven material allowable boundary construction device is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.
[0100] This invention also provides a computer device having the above-described features. Figure 8 The data-driven material allowable boundary construction device shown.
[0101] Please see Figure 9 The present invention provides a schematic diagram of the structure of a computer device according to an optional embodiment. The computer device includes one or more processors 10, a memory 20, and interfaces for connecting the various components, including high-speed interfaces and low-speed interfaces. The various components are interconnected via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, multiple processors and / or multiple buses can be used with multiple memories, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 9 Take a processor 10 as an example.
[0102] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.
[0103] The memory 20 stores instructions executable by at least one processor 10 to cause at least one processor 10 to perform the method shown in the above embodiments.
[0104] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function. The data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, which can be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.
[0105] Memory 20 may include volatile memory, such as random access memory. Memory may also include non-volatile memory, such as flash memory, hard disk, or solid-state drive. Memory 20 may also include combinations of the above types of memory.
[0106] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.
[0107] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.
[0108] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention 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 of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A data-driven method for constructing allowable material boundaries, characterized in that, include: The corrosion parameter value space of the target material under operating conditions is coarsely divided into multiple operating condition grids; wherein each operating condition grid includes multiple operating condition vertices; Based on each of the stated operating conditions, the set parameter perturbation range, and the trained corrosion rate prediction model, predict the robust corrosion rate of each of the stated operating conditions. Based on the corrosion rate threshold corresponding to at least one corrosion hazard level and the robust corrosion rate of each of the operating conditions, an initial parameter allowable boundary is generated for each of the corrosion rate thresholds. For any of the corrosion rate thresholds corresponding to the initial parameter allowable boundary, the plurality of working condition meshes are locally refined according to the corrosion rate thresholds to obtain the corresponding plurality of refined working condition meshes. The parameter allowable boundary is iteratively updated according to the plurality of refined working condition meshes to obtain the final parameter allowable boundary.
2. The method according to claim 1, characterized in that, The corrosion parameter value space consists of multiple corrosion parameter value ranges, and each working condition vertex includes the value of each corrosion parameter.
3. The method according to claim 2, characterized in that, The parameter disturbance range includes the value disturbance range of each corrosion parameter; The step of predicting the robust corrosion rate for each of the operating conditions based on each operating condition vertex, the set parameter perturbation range, and the trained corrosion rate prediction model includes: For any given working condition vertex, the value of each corrosion parameter in the parameter perturbation interval is used to perform uncertainty perturbation on the value of each corrosion parameter in the working condition vertex, resulting in multiple perturbation samples. The corrosion rate prediction model is used to predict the corrosion rate of each perturbation sample, resulting in a predicted corrosion rate for each perturbation sample. The predicted corrosion rates of each perturbation sample are aggregated to obtain the robust corrosion rate of the working condition vertex.
4. The method according to claim 2, characterized in that, The process of generating an initial parameter allowable boundary corresponding to each corrosion rate threshold based on at least one corrosion hazard level and the robust corrosion rate of each operating condition peak includes: Based on the value of each corrosion parameter in each working condition vertex and the robust corrosion rate of each working condition vertex, a functional relationship between the corrosion rate and each corrosion parameter is fitted. For any of the corrosion rate thresholds, the value boundaries of each corrosion parameter are solved according to the corrosion rate threshold and the functional relationship, and the value boundaries of each corrosion parameter as a whole are used as the initial parameter allowable boundary corresponding to the corrosion rate threshold.
5. The method according to claim 4, characterized in that, The step of locally refining the plurality of working condition meshes according to the corrosion rate threshold to obtain corresponding plurality of refined working condition meshes includes: For any given working condition mesh, if among the robust corrosion rates of each working condition vertex of the working condition mesh, there exists a robust corrosion rate greater than the corrosion rate threshold and a robust corrosion rate less than the corrosion rate threshold, then the working condition mesh is determined as a boundary candidate mesh. Each of the boundary candidate meshes is further subdivided into finer meshes to obtain multiple finer condition meshes.
6. The method according to claim 5, characterized in that, The step of iteratively updating the allowable parameter boundaries based on the multiple refined condition meshes to obtain the final allowable parameter boundaries includes: Based on each refined working condition vertex in each refined working condition mesh, the parameter perturbation range, and the corrosion rate prediction model, predict the robust corrosion rate of each refined working condition vertex. Based on the values of each corrosion parameter in each of the working condition vertices and each of the refined working condition vertices, and the robust corrosion rate of each of the working condition vertices and each of the refined working condition vertices, a new functional relationship between the corrosion rate and each corrosion parameter is fitted. Based on the corrosion rate threshold and the new functional relationship, the new value boundary of each corrosion parameter is solved, and the whole is used as the new parameter allowable boundary corresponding to the corrosion rate threshold; For any of the aforementioned working condition meshes or the refined working condition meshes, if there is a robust corrosion rate greater than the corrosion rate threshold and a robust corrosion rate less than the corrosion rate threshold among the robust corrosion rates of each working condition vertex of the target mesh, then the target mesh is determined as a boundary candidate mesh until the iteration termination condition is met, and the final allowable parameter boundary is obtained.
7. The method according to any one of claims 2 to 6, characterized in that, The multiple corrosion parameters include several of the following: temperature, hydrogen sulfide concentration, carbon dioxide fugacity, pressure, chloride ion concentration, and medium flow rate.
8. A data-driven material allowable boundary construction device, characterized in that, include: The coarse meshing unit is used to coarsely divide the corrosion parameter value space of the target material under operating conditions to obtain multiple operating condition meshes; wherein, each operating condition mesh includes multiple operating condition vertices; The prediction unit is used to predict the robust corrosion rate of each of the said working conditions based on each working condition vertex, the set parameter perturbation range and the trained corrosion rate prediction model. The generation unit is used to generate an initial parameter allowable boundary corresponding to each corrosion rate threshold based on a corrosion rate threshold corresponding to at least one corrosion hazard level and a robust corrosion rate for each of the operating conditions. A local refinement unit is used to locally refine the plurality of working condition meshes according to the corrosion rate threshold for any initial parameter allowable boundary corresponding to the corrosion rate threshold, so as to obtain the corresponding plurality of refined working condition meshes. The update unit is used to iteratively update the allowable parameter boundaries based on the multiple refined condition meshes to obtain the final allowable parameter boundaries.
9. A computer device, characterized in that, include: A memory and a processor are communicatively connected, the memory storing computer instructions, and the processor executing the computer instructions to perform the data-driven material allowable boundary construction method of any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing a computer to perform the data-driven material allowable boundary construction method of any one of claims 1 to 7.