A method for predicting the corrosion damage of weathering steel in marine atmosphere and evaluating the service life

Abstract

This invention provides a method for predicting marine atmospheric corrosion damage and assessing the service life of weathering steel. The method includes constructing a corrosion dataset, converting corrosion damage results, determining model input features, training a corrosion loss prediction model, outputting a corrosion loss prediction sequence, determining the predicted service life, conducting outdoor exposure verification, and conducting rust layer characterization verification. The corrosion loss prediction results are further correlated with allowable corrosion loss thresholds to obtain the predicted service life of weathering steel under a target marine atmospheric environment. This method combines corrosion loss prediction, service life threshold determination, and verification and correction processes, and uses outdoor exposure data and rust layer characterization results to support the prediction conclusions. This forms a complete technical solution suitable for assessing the marine atmospheric corrosion service life of weathering steel, offering advantages such as strong engineering applicability of the prediction results, clear service life determination criteria, and high model evaluation reliability.

Description

Technical Field

[0001] This invention relates to the field of weathering steel corrosion evaluation technology, specifically to a method for predicting marine atmospheric corrosion damage and assessing the service life of weathering steel. Background Technology

[0002] Weathering steel is widely used in bridges, transmission towers, port facilities, coastal engineering equipment, and other long-term service structures because it can form a protective rust layer in the atmospheric environment. The corrosion resistance of weathering steel is closely related to the regulation of rust layer structure and phase evolution by alloying elements such as Cr, Cu, Ni, P, and Si. Especially during long-term service, alloying elements have a significant impact on the density, stability, and protective phase formation of the rust layer. However, in the marine atmospheric environment, factors such as Cl- deposition, high humidity, rainfall, condensation, wet-dry cycles, pH changes, and SO2 pollution can significantly affect the stability of the rust layer. These environmental factors do not act alone, but rather influence the corrosion reaction process on the surface of weathering steel together with regional differences, seasonal changes, and climate fluctuations. This makes the corrosion damage evolution of weathering steel in the marine atmospheric environment exhibit obvious multi-factor coupling, long-term accumulation, and nonlinear characteristics.

[0003] Current methods for evaluating the corrosion life of weathering steel typically rely on natural exposure tests, empirical corrosion models, dose response functions, and accelerated laboratory tests. While natural exposure tests can realistically reflect the actual service behavior of materials, they have long testing cycles and are highly dependent on geographical location. It usually takes a long time to obtain stable evaluation conclusions, which is difficult to meet the needs of rapid material screening, marine atmospheric environment adaptability evaluation, and service life prediction. Therefore, further optimization is needed.

[0004] To shorten the evaluation cycle and improve prediction efficiency, existing technologies have developed schemes that use empirical models, dose response functions, and data-driven models to predict the corrosion rate, corrosion loss, or weather resistance of weathering steel. These schemes typically use material composition, environmental parameters, and exposure time as inputs and output the predicted corrosion rate, corrosion level, or corrosion loss through preset functional relationships or machine learning models. However, existing schemes are still easily limited by the model output format, data integrity, and engineering judgment rules.

[0005] First, empirical models and dose-response functions typically require pre-defined function forms, which are insufficient in describing the nonlinear coupling effects between material composition, marine chloride environmental load, humid and hot conditions, pollutant levels, and exposure time. Second, some data-driven models primarily focus on prediction accuracy or feature importance analysis, with most methods only outputting corrosion rate or corrosion loss values. They lack engineering conversion rules from corrosion damage prediction results to service life output. Furthermore, they lack a unified approach to address the incomplete environmental records in historical outdoor exposure data, which is suitable for predicting long-term cumulative corrosion loss. They also lack a closed-loop mechanism to verify or correct prediction results using independent outdoor exposure samples and corrosion evidence such as rust morphology, elemental distribution, phase composition, and valence state. Therefore, they cannot fully meet the actual needs of weathering steel service life assessment in marine atmospheric environments.

[0006] There is an urgent need for a method to predict marine atmospheric corrosion damage and assess service life of weathering steel, in order to solve the problems in the existing corrosion evaluation of weathering steel, such as long natural exposure period, insufficient characterization of multi-factor nonlinear coupling effect by empirical models, disconnect between corrosion rate or corrosion loss prediction results and service life determination, insufficient use of historical outdoor exposure data, and lack of corrosion verification support for model prediction results. Summary of the Invention

[0007] This invention discloses a method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel. It belongs to the category of weathering steel corrosion evaluation and solves the technical problems in existing weathering steel corrosion evaluation, such as long natural exposure period, insufficient characterization of multi-factor nonlinear coupling effect by empirical models, disconnect between corrosion rate or corrosion loss prediction results and service life determination, insufficient utilization of historical outdoor exposure data, and lack of corrosion-related verification support for model prediction results.

[0008] To achieve the above-mentioned technical objectives, the specific technical solution adopted by the present invention is as follows: A method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel includes the following steps: S1. Construct a corrosion dataset, which includes weathering steel material composition, marine atmospheric environment parameters, exposure time, and corrosion damage results; S2. Convert the corrosion damage result into corrosion loss value; when the corrosion damage result is corrosion weight loss, convert the corrosion weight loss into an equivalent corrosion loss value according to the exposed area of ​​the sample and the material density, and use the converted corrosion loss value as the target variable of the model. S3. Based on the constraints of marine atmospheric corrosion mechanism and statistical analysis, determine the model input characteristics, so that the model input characteristics include material composition characteristics, environmental load characteristics and exposure time characteristics related to the marine atmospheric corrosion process of weathering steel; S4. Train a machine learning regression model based on the model input features and the model target variable to obtain a corrosion loss prediction model. The corrosion loss prediction model is used to establish the mapping relationship between weathering steel material composition, marine atmospheric environment parameters, exposure time and corrosion loss value. S5. Input the material composition of the weathering steel to be evaluated, the target marine atmospheric environment parameters, and multiple candidate exposure times into the corrosion loss prediction model, and output the corrosion loss prediction value D(t) corresponding to each candidate exposure time to form a prediction sequence of corrosion loss as exposure time. S6. Based on the allowable corrosion loss threshold Dcrit and the prediction sequence, the candidate exposure time that first satisfies D(t)≥Dcrit is determined as the predicted service life Life of the weathering steel to be evaluated in the target marine atmospheric environment, where Life=min{t|D(t)≥Dcrit}, t is the candidate exposure time, D(t) is the predicted corrosion loss value corresponding to the candidate exposure time t, and Dcrit is the allowable corrosion loss threshold.

[0009] Furthermore, in step S6, after determining the predicted service life (Life), the following processing is also included: S601. The corrosion loss prediction model was validated using independent outdoor exposure samples, and the error evaluation results between the predicted corrosion loss value and the measured corrosion damage result were obtained. S602. Based on the error evaluation results, calibrate the applicable scope of the corrosion thickness loss prediction model; S603. Verify or correct the prediction trend of the corrosion thickness loss prediction model through the rust layer characterization results.

[0010] Furthermore, in step S601, the independent outdoor exposure samples include weathering steel outdoor exposure samples that did not participate in the training, feature selection, and parameter optimization of the corrosion loss prediction model. The error evaluation results include at least one of the relative error, root mean square error, and mean absolute error between the predicted corrosion loss value and the measured corrosion loss value.

[0011] Furthermore, in step S603, the rust characterization results include one or more of the following: rust morphology, elemental distribution, phase composition, and valence state. The rust characterization results are obtained through one or more of the following methods: SEM / EDS, EPMA, XRD, and XPS.

[0012] Furthermore, in step S1, the weathering steel material composition includes at least one of C, Mn, Cr, Cu, Ni, P, S and Si, and the marine atmospheric environmental parameters include at least one of chloride ion deposition parameters, sulfur dioxide pollution parameters, temperature, relative humidity, pH, wetting time, high humidity duration, rainfall and dry-wet cycle frequency.

[0013] Furthermore, in step S3, the marine atmospheric corrosion mechanism constraint includes retaining at least one of the following parameters related to the marine atmospheric corrosion process: chloride ion deposition parameters, sulfur dioxide pollution parameters, relative humidity, pH, exposure time, Cr, Cu, Ni and S as candidate input features, and the statistical analysis includes at least one of correlation analysis, feature importance analysis and model contribution analysis.

[0014] Furthermore, in step S5, multiple candidate exposure times are set according to a preset time step, and the corrosion loss prediction model outputs the corrosion loss prediction value D(t) corresponding to each candidate exposure time, forming a prediction sequence according to the order of the candidate exposure times.

[0015] Furthermore, in step S6, the allowable corrosion thickness loss threshold Dcrit is determined based on the engineering allowable damage conditions of the weathering steel to be evaluated. When the engineering allowable damage conditions include the initial thickness h0 and the allowable thickness loss ratio η, the allowable corrosion thickness loss threshold Dcrit is determined according to Dcrit=η×h0, and the lifetime threshold is determined by using the allowable corrosion thickness loss threshold Dcrit to determine the predicted sequence.

[0016] Furthermore, in step S1, when the corrosion dataset lacks marine atmospheric environmental parameters synchronized with the exposure year, representative annual parameters, multi-year average parameters, or typical environmental parameters of the same exposure site are used as environmental load descriptors. The environmental load descriptors are used for long-term cumulative corrosion thickness loss prediction, but not for transient corrosion current or daily-scale corrosion rate prediction.

[0017] An evaluation system for predicting marine atmospheric corrosion damage and assessing service life of weathering steel as described above includes a data input module, a data preprocessing module, a corrosion prediction module, a rate calculation module, a service life threshold determination module, an interpretability module, and a result output module. The data input module includes an alloy composition input terminal, an environmental parameter input terminal, and a lifetime parameter input terminal. The data input module is connected to the data preprocessing module, and the data preprocessing module is connected to the corrosion prediction module, forming a front-end data processing link from alloy composition, environmental parameters, and lifetime parameters into the corrosion prediction module. The corrosion prediction module is connected to the rate calculation module, the lifetime threshold determination module, and the interpretability module, respectively, forming a corrosion rate calculation branch, a lifetime threshold determination branch, and a model interpretation branch based on the corrosion thickness loss prediction results. The rate calculation module, the service life threshold determination module, and the interpretability module are connected to the result output module to form a summary output link of corrosion rate, predicted service life, and model interpretation results. The results output module includes output terminals for predicted corrosion thickness loss, periodic average corrosion rate, predicted service life, and risk level.

[0018] This invention incorporates the composition of weathering steel materials, marine atmospheric environmental parameters, and candidate exposure times into the corrosion loss prediction process. Furthermore, it correlates the predicted corrosion loss results with allowable corrosion loss thresholds, thus forming a continuous evaluation link from corrosion loss prediction to predicted service life output. Outdoor exposure verification and rust layer characterization results correct and support the prediction trend, combining the model prediction results with the marine atmospheric corrosion mechanism of weathering steel and the determination of its engineering service life. This avoids the problem of simply combining material composition, environmental parameters, and conventional prediction models, which still only results in single-point predictions of corrosion rate or corrosion loss.

[0019] By adopting the above technical solution, the present invention can also bring the following beneficial effects: 1. This invention discloses a method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel. It integrates the composition of weathering steel material, marine atmospheric environmental parameters, and exposure time into the corrosion damage prediction process. This allows the corrosion loss prediction model to simultaneously consider the combined effects of the material's own corrosion-resistant elements, marine atmospheric environmental load, and service time on the evolution of corrosion damage. Compared with methods that rely solely on natural exposure tests, empirical models, or single corrosion rate evaluations, this method uses corrosion loss values ​​as the evaluation object to predict the long-term cumulative corrosion damage of weathering steel in the target marine atmospheric environment. This improves the correspondence between corrosion evaluation results and actual service damage status, and has the advantages of strong corrosion damage prediction and clearer evaluation objects.

[0020] 2. This invention provides a method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel. After obtaining the predicted corrosion loss value, it can determine the predicted service life by combining the allowable corrosion loss threshold, so that the corrosion damage prediction result can be converted into the engineering service life assessment result. This avoids the problem that ordinary corrosion prediction models only output corrosion rate or corrosion loss value and are difficult to use for service life determination. By forming a prediction sequence of corrosion loss changing with exposure time through multiple candidate exposure times, the service life assessment process can be carried out around the threshold determination, no longer stopping at the prediction of single-point corrosion results. It has the advantages of clear service life determination basis and strong engineering application conversion.

[0021] 3. This invention proposes a method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel. By constraining the marine atmospheric corrosion mechanism and using statistical analysis, the model input characteristics are determined, ensuring a correspondence between these characteristics and corrosion-influencing factors such as chloride ion deposition, sulfur dioxide pollution, relative humidity, pH, exposure time, and corrosion-resistant elements. This reduces the black-box nature of purely data-driven modeling. For cases where environmental records in historical outdoor exposure data are not fully synchronized, representative annual parameters, multi-year average parameters, or typical environmental parameters from the exposure sites are used as environmental load descriptors, improving the usability of multi-source corrosion data. Combined with independent outdoor exposure samples, error evaluation results, and rust layer characterization results, the applicability and prediction trend of the corrosion thickness prediction model are verified or corrected. This method has the advantages of strong data adaptability and high reliability of prediction results. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This invention provides a flowchart illustrating a method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel. Figure 2 This is a schematic diagram illustrating the changes in corrosion thickness loss with exposure time and the determination of the lifetime threshold in the embodiments. Figure 3 This is a schematic diagram illustrating the relationship between the combination of corrosion-resistant elements and the predicted corrosion thickness loss in the examples. Figure 4 This is a schematic diagram illustrating the closed-loop verification between the corrosion thickness loss prediction model, outdoor exposure verification, and rust layer characterization verification in the embodiment. Figure 5 This invention provides a schematic diagram of the module structure of a system for predicting marine atmospheric corrosion damage and assessing service life of weathering steel. Detailed Implementation

[0024] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0025] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0026] It should be noted that various aspects of embodiments within the scope of the appended claims are described below. It will be apparent that the aspects described herein can be embodied in a wide variety of forms, and any particular structure and / or function described herein is merely illustrative. Based on this invention, those skilled in the art will understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects set forth herein can be used to implement the device and / or practice the method. Additionally, this device and / or method can be implemented using other structures and / or functionalities besides one or more of the aspects set forth herein.

[0027] It should also be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. The drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0028] Furthermore, specific details are provided in the following description to facilitate a thorough understanding of the examples. However, those skilled in the art will understand that the described aspects can be practiced without these specific details. Example

[0029] This invention provides a method for predicting marine atmospheric corrosion damage and assessing the service life of weathering steel. The method includes constructing a corrosion dataset, converting corrosion damage results, determining model input features, training a corrosion loss prediction model, outputting a corrosion loss prediction sequence, determining the predicted service life, conducting outdoor exposure verification, and conducting rust layer characterization verification. It does not merely aim to predict corrosion rate or single-point corrosion loss as the final goal, but further compares the corrosion loss prediction results with the allowable corrosion loss threshold to obtain the predicted service life of weathering steel in the target marine atmospheric environment.

[0030] S1. Construct the corrosion dataset; The corrosion dataset includes the composition of weathering steel, marine atmospheric environmental parameters, exposure time, and corrosion damage results. The composition of weathering steel can include C, Mn, Cr, Cu, Ni, P, S, and Si. Marine atmospheric environmental parameters can include chloride ion deposition parameters, sulfur dioxide pollution parameters, temperature, relative humidity, pH, wetting time, duration of high humidity, rainfall, and frequency of wet-dry cycles. Exposure time is recorded in days. The corrosion damage results are corrosion thickness loss values, which can also be replaced with corrosion weight loss values ​​as needed.

[0031] In this embodiment, corrosion data can be sourced from the National Corrosion Data Center, enterprise outdoor exposure test data, and publicly available literature data. For data from different sources, the material composition field, environmental parameter field, exposure time field, and corrosion damage result field are first unified to ensure they can be included in the same corrosion dataset. For some historical outdoor exposure data, if marine atmospheric environmental parameters synchronized with the exposure year are lacking, representative annual parameters, multi-year average parameters, or typical environmental parameters from the same exposure site are used as environmental load descriptors. The corresponding environmental load descriptors are used for long-term cumulative corrosion thickness loss prediction, not for transient corrosion current or daily-scale corrosion rate prediction, thereby avoiding the misinterpretation of site-representative parameters as transient corrosion process parameters.

[0032] S2. Convert the corrosion damage results into corrosion thickness loss values; When corrosion damage results are recorded as corrosion loss values, they can be directly used as the model's target variable. When corrosion damage results are in the form of corrosion weight loss, the corrosion weight loss is converted into an equivalent corrosion loss value based on the sample's exposed area and material density, and this converted corrosion loss value is used as the model's target variable. Through this process, corrosion damage results from different sources and in different recording formats can be unified into the corrosion loss evaluation dimension, avoiding the problem that corrosion weight loss data and corrosion loss data cannot be used together in modeling.

[0033] S3. Determine the model input characteristics based on marine atmospheric corrosion mechanism constraints and statistical analysis; The constraints on marine atmospheric corrosion mechanisms include retaining chloride ion deposition parameters, sulfur dioxide pollution parameters, relative humidity, pH, exposure time, Cr, Cu, Ni, and S as candidate input features related to the marine atmospheric corrosion process of weathering steel. Statistical analysis includes correlation analysis, feature importance analysis, and model contribution analysis. In this way, the model input features are not simply automatically selected by the algorithm, but are constrained in conjunction with the corrosion mechanism of weathering steel in the marine atmospheric environment, so that the model input features maintain a correspondence with the actual corrosion influencing factors.

[0034] The model input features include C, Mn, Cr, Cu, Ni, P, S, Si, Time, Cl-, SO2, Temp, RH, and pH, where Time represents exposure time, Temp represents temperature, and RH represents relative humidity. In practical use, the model input features are added or removed based on data integrity, target marine atmospheric environment characteristics, and weathering steel material type.

[0035] S4. Train a machine learning regression model based on the model input features and the model target variable to obtain a corrosion thickness prediction model; The corrosion loss prediction model is used to establish the mapping relationship between weathering steel material composition, marine atmospheric environmental parameters, exposure time and corrosion loss value. This invention does not take a specific type of machine learning model as the core innovation point. The machine learning regression model in this example adopts the GBDT model. Data cleaning, missing value imputation, outlier handling, target variable transformation and input variable standardization can all be used as auxiliary processing methods for model training.

[0036] In the completed experiment of this example, the number of samples after cleaning was 708, the training set had 566 samples, and the test set had 142 samples. Then, the GBDT model was used to establish a corrosion loss prediction model, with the corrosion loss value as the model's objective variable. The test set results are shown in the table below:

[0037] As shown in the table above, the corrosion loss prediction model can effectively reflect the cumulative corrosion damage trend of weathering steel in atmospheric or marine environments.

[0038] S5. Input the material composition of the weathering steel to be evaluated, the target marine atmospheric environment parameters, and multiple candidate exposure times into the corrosion loss prediction model, and output the corrosion loss prediction value D(t) corresponding to each candidate exposure time to form a prediction sequence of corrosion loss as exposure time. Multiple candidate exposure times can be set according to a preset time step, with months as the time step. By continuously inputting multiple candidate exposure times, corrosion thickness prediction results under different service times can be obtained.

[0039] like Figure 2 As shown, the predicted corrosion thickness D(t) can form a corrosion thickness variation curve with candidate exposure time. Figure 2 The horizontal axis represents the exposure time, the vertical axis represents the predicted corrosion loss value, and the horizontal threshold line represents the allowable corrosion loss threshold Dcrit. The intersection of the corrosion loss prediction curve and the allowable corrosion loss threshold Dcrit corresponds to the predicted service life, further illustrating the determination method for converting corrosion damage prediction results into service life assessment results.

[0040] S6. Based on the allowable corrosion loss threshold Dcrit and the prediction sequence, the candidate exposure time that first satisfies D(t)≥Dcrit is determined as the predicted service life Life of the weathering steel to be evaluated in the target marine atmospheric environment. Where Life = min{t|D(t)≥Dcrit}, t is the candidate exposure time, D(t) is the predicted corrosion thickness loss value corresponding to the candidate exposure time t, and Dcrit is the allowable corrosion thickness loss threshold.

[0041] The allowable corrosion loss threshold Dcrit is determined based on the engineering allowable damage conditions of the weathering steel to be evaluated. When the engineering allowable damage conditions include the initial thickness h0 and the allowable thickness loss ratio η, the allowable corrosion loss threshold Dcrit is determined according to Dcrit=η×h0. The allowable corrosion loss threshold Dcrit is then used to determine the life threshold of the prediction sequence. In this embodiment, the initial thickness of the steel plate is selected as 10 mm, and the allowable thickness loss ratio is 10%. Therefore, the allowable corrosion loss threshold Dcrit is 1 mm. Subsequently, the corrosion loss prediction value corresponding to each candidate exposure time is obtained step by step according to the preset time step. When the corrosion loss prediction value reaches or exceeds the allowable corrosion loss threshold Dcrit for the first time, the candidate exposure time is taken as the predicted service life Life.

[0042] The main process involves transforming the corrosion loss prediction results into predictable service life that can be used in engineering through steps S5 and S6, which can provide direct basis for the selection of weathering steel materials, evaluation of marine atmospheric environment adaptability, and maintenance decisions.

[0043] S7. The corrosion loss prediction model was validated using independent outdoor exposure samples, and the applicable scope of the corrosion loss prediction model was calibrated based on the error evaluation results. Independent outdoor exposure samples include weathering steel outdoor exposure samples that were not involved in the training, feature selection, and parameter optimization of the corrosion loss prediction model. The error evaluation results include the relative error, root mean square error, and mean absolute error between the predicted corrosion loss value and the measured corrosion loss value. If the error evaluation results are within the preset error range, the corrosion loss prediction model is considered applicable to the current weathering steel material type and the target marine atmospheric environment range. If the error evaluation results exceed the preset error range, the applicability of the corrosion loss prediction model is calibrated according to the error distribution, material composition range, environmental parameter range, and exposure time range to avoid the model outputting unreliable conclusions under data conditions that exceed the applicable range.

[0044] In this embodiment, thirty-one independent outdoor verification samples were used to evaluate the generalization ability of the model. These independent outdoor verification samples did not participate in model training, feature selection, or parameter optimization. Then, representative Cr-Cu weathering steel WS was selected for outdoor natural exposure verification at 3 M, 6 M, 12 M, and 24 M. The corrosion thickness loss value predicted by the model was converted into corrosion weight loss and compared with the measured corrosion weight loss. The results showed that the relative error between the predicted corrosion weight loss and the measured corrosion weight loss under different exposure periods was less than 7%, proving that the corrosion thickness loss prediction model can still maintain a low prediction error on outdoor exposure samples that did not participate in training. It can better reflect the actual corrosion damage change trend of representative Cr-Cu weathering steel under different exposure periods, thus providing a reliable data basis for the subsequent threshold determination of service life prediction.

[0045] S8. Verify or correct the prediction trend of the corrosion thickness prediction model through rust layer characterization results; Rust layer characterization results include rust layer morphology, element distribution, phase composition and valence state. Rust layer characterization results are obtained through one or more of SEM / EDS, EPMA, XRD and XPS. Through rust layer characterization results, it can be determined whether the corrosion-resistant element combination trend identified by the model is consistent with the actual rust layer structure and corrosion product evolution law.

[0046] In this embodiment, the rust layer of WS steel after exposure to sunlight was characterized by SEM / EDS, EPMA, XRD and XPS. The characterization results showed that Cr was enriched in the inner rust layer and existed in the form of Cr2O3, CrOOH and FeCr2O4. Cu participated in the transformation of corrosion products in the form of CuFeO2 and CuO. The proportion of protective rust layer increased with exposure time.

[0047] The experimental results above demonstrate that the trend identified by the model, where the combination of high Cr and high Cu is beneficial to reducing corrosion thickness loss, is consistent with the prediction trend of the corrosion thickness loss prediction model. This shows that the prediction trend of the corrosion thickness loss prediction model can be supported by outdoor exposure data and rust corrosion evidence.

[0048] like Figure 4 As shown, this invention forms a closed-loop verification system between model prediction, outdoor exposure, error evaluation, and rust characterization. The model prediction results are not directly used as the final, uncorrectable conclusion, but can be verified or corrected through independent outdoor exposure samples and rust characterization results, thereby improving the reliability of the corrosion loss prediction model in the assessment of marine atmospheric corrosion life of weathering steel.

[0049] like Figure 3 As shown, this invention can also utilize interpretable analysis methods to analyze the relationship between corrosion-resistant element combinations and corrosion loss prediction results. Figure 4Taking the relationship between the Cr-Cu combination range and corrosion loss or life assessment results as an example, this study demonstrates the influence of different combinations of corrosion-resistant elements on the corrosion loss prediction trend. In this embodiment, samples were grouped according to Cr and Cu content, and the corrosion loss prediction results under different Cr-Cu combinations were compared. The results show that low Cu and low Cr combinations correspond to higher corrosion loss, while high Cu and high Cr combinations correspond to lower corrosion loss. These trends were further verified by rust layer characterization results, thus helping users understand the output of the corrosion loss prediction model. This ensures that the model prediction results not only have numerical output but also establish a correspondence with the influence of weathering steel alloy elements on rust layer protection.

[0050] like Figure 5 As shown, this invention discloses a system for predicting marine atmospheric corrosion damage and assessing service life of weathering steel, including a data input module, a data preprocessing module, a corrosion prediction module, a rate calculation module, a service life threshold determination module, an interpretability module, and a result output module. The data input module includes an alloy composition input terminal, an environmental parameter input terminal, and a service life parameter input terminal. The data input module is connected to the data preprocessing module, and the data preprocessing module is connected to the corrosion prediction module, forming a front-end data processing link from alloy composition, environmental parameters, and service life parameters into the corrosion prediction module.

[0051] The corrosion prediction module is connected to the rate calculation module, the lifetime threshold determination module, and the interpretability module, forming a corrosion rate calculation branch, a lifetime threshold determination branch, and a model interpretation branch based on the corrosion loss prediction results. The rate calculation module, the lifetime threshold determination module, and the interpretability module are connected to the result output module, forming a summary output link of corrosion rate, predicted service life, and model interpretation results. The result output module includes a corrosion loss prediction value output terminal, a periodic average corrosion rate output terminal, a predicted service life output terminal, and a risk level output terminal, thereby converting material composition, environmental parameters, candidate exposure time, and lifetime parameters into corrosion loss prediction values, periodic average corrosion rates, predicted service life, and risk levels, thus realizing the software application of the method of this invention.

[0052] In practical applications, the corresponding method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel can also be implemented by electronic devices. The electronic devices include a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, it can implement one or more steps in the above-mentioned method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel.

[0053] In practical applications, this invention also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements one or more steps in the above-mentioned method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel. The computer-readable storage medium can be a hard disk, solid-state drive, memory card, read-only memory, random access memory, or other media capable of storing computer programs.

[0054] In summary, this invention combines corrosion thickness loss prediction, lifespan threshold determination, and verification and correction processes, enabling the corrosion damage prediction results of weathering steel in marine atmospheric environments to be further transformed into predicted service life that can be used for engineering judgment. Furthermore, the prediction conclusions can be supported by outdoor exposure data and rust layer characterization results, thus forming a complete technical solution applicable to the marine atmospheric corrosion life assessment of weathering steel. It has the advantages of strong engineering applicability of prediction results, clear basis for lifespan determination, and high reliability of model evaluation.

[0055] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel, characterized in that, Includes the following steps: S1. Construct a corrosion dataset, which includes weathering steel material composition, marine atmospheric environment parameters, exposure time, and corrosion damage results; S2. Convert the corrosion damage result into corrosion loss value; when the corrosion damage result is corrosion weight loss, convert the corrosion weight loss into an equivalent corrosion loss value according to the exposed area of ​​the sample and the material density, and use the converted corrosion loss value as the target variable of the model. S3. Based on the constraints of marine atmospheric corrosion mechanism and statistical analysis, determine the model input characteristics, so that the model input characteristics include material composition characteristics, environmental load characteristics and exposure time characteristics related to the marine atmospheric corrosion process of weathering steel; S4. Train a machine learning regression model based on the model input features and the model target variable to obtain a corrosion loss prediction model. The corrosion loss prediction model is used to establish the mapping relationship between weathering steel material composition, marine atmospheric environment parameters, exposure time and corrosion loss value. S5. Input the material composition of the weathering steel to be evaluated, the target marine atmospheric environment parameters, and multiple candidate exposure times into the corrosion loss prediction model, and output the corrosion loss prediction value D(t) corresponding to each candidate exposure time to form a prediction sequence of corrosion loss as exposure time. S6. Based on the allowable corrosion loss threshold Dcrit and the prediction sequence, the candidate exposure time that first satisfies D(t)≥Dcrit is determined as the predicted service life Life of the weathering steel to be evaluated in the target marine atmospheric environment, where Life=min{t|D(t)≥Dcrit}, t is the candidate exposure time, D(t) is the predicted corrosion loss value corresponding to the candidate exposure time t, and Dcrit is the allowable corrosion loss threshold.

2. The method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel according to claim 1, characterized in that: In step S6, after determining the predicted service life (Life), the following processing is also included: S601. The corrosion loss prediction model was validated using independent outdoor exposure samples, and the error evaluation results between the predicted corrosion loss value and the measured corrosion damage result were obtained. S602. Based on the error evaluation results, calibrate the applicable scope of the corrosion thickness loss prediction model; S603. Verify or correct the prediction trend of the corrosion thickness loss prediction model through the rust layer characterization results.

3. The method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel according to claim 2, characterized in that: In step S601, the independent outdoor exposure samples include weathering steel outdoor exposure samples that did not participate in the training, feature screening and parameter optimization of the corrosion loss prediction model. The error evaluation results include at least one of the relative error, root mean square error and mean absolute error between the predicted corrosion loss value and the measured corrosion loss value.

4. The method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel according to claim 3, characterized in that, In step S603, the rust layer characterization results include one or more of the following: rust layer morphology, elemental distribution, phase composition, and valence state. The rust layer characterization results are obtained through one or more of the following methods: SEM / EDS, EPMA, XRD, and XPS.

5. The method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel according to claim 4, characterized in that: In step S1, the weathering steel material composition includes at least one of C, Mn, Cr, Cu, Ni, P, S, and Si, and the marine atmospheric environmental parameters include at least one of chloride ion deposition parameters, sulfur dioxide pollution parameters, temperature, relative humidity, pH, wetting time, high humidity duration, rainfall, and wet-dry cycle frequency.

6. The method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel according to claim 5, characterized in that: In step S3, the marine atmospheric corrosion mechanism constraint includes retaining at least one of the following parameters related to the marine atmospheric corrosion process: chloride ion deposition parameters, sulfur dioxide pollution parameters, relative humidity, pH, exposure time, Cr, Cu, Ni and S as candidate input features. The statistical analysis includes at least one of correlation analysis, feature importance analysis and model contribution analysis.

7. The method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel according to claim 6, characterized in that: In step S5, multiple candidate exposure times are set according to a preset time step. The corrosion loss prediction model outputs the corrosion loss prediction value D(t) corresponding to each candidate exposure time and forms a prediction sequence according to the order of the candidate exposure times.

8. The method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel according to claim 1, characterized in that: In step S6, the allowable corrosion thickness loss threshold Dcrit is determined based on the engineering allowable damage conditions of the weathering steel to be evaluated. When the engineering allowable damage conditions include the initial thickness h0 and the allowable thickness loss ratio η, the allowable corrosion thickness loss threshold Dcrit is determined according to Dcrit=η×h0, and the lifetime threshold is determined by using the allowable corrosion thickness loss threshold Dcrit.

9. The method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel according to claim 5, characterized in that: In step S1, when the corrosion dataset lacks marine atmospheric environmental parameters synchronized with the exposure year, representative annual parameters, multi-year average parameters, or typical environmental parameters of the same exposure site are used as environmental load descriptors. These environmental load descriptors are used for long-term cumulative corrosion thickness loss prediction, but not for transient corrosion current or daily-scale corrosion rate prediction.

10. An evaluation system employing the method for predicting marine atmospheric corrosion damage and assessing service life of weathering steel according to claim 1, characterized in that: It includes a data input module, a data preprocessing module, a corrosion prediction module, a rate calculation module, a lifetime threshold determination module, an interpretability module, and a result output module; The data input module includes an alloy composition input terminal, an environmental parameter input terminal, and a lifetime parameter input terminal. The data input module is connected to the data preprocessing module, and the data preprocessing module is connected to the corrosion prediction module, forming a front-end data processing link from alloy composition, environmental parameters, and lifetime parameters into the corrosion prediction module. The corrosion prediction module is connected to the rate calculation module, the lifetime threshold determination module, and the interpretability module, respectively, forming a corrosion rate calculation branch, a lifetime threshold determination branch, and a model interpretation branch based on the corrosion thickness loss prediction results. The rate calculation module, the lifespan threshold determination module, and the interpretability module are respectively connected to the result output module to form a summary output link of corrosion rate, predicted service life, and model interpretation results; The result output module includes a corrosion thickness prediction output terminal, a periodic average corrosion rate output terminal, a predicted service life output terminal, and a risk level output terminal.

Images