Ocean engineering structure external corrosion prediction method

Abstract

The invention discloses an ocean engineering structure external corrosion prediction method, which comprises a data acquisition module, a real-time prediction monitoring module, a multi-model integrated learning module and a visualization module, the data acquisition module acquires corrosion data of the ocean engineering structure in real time by using various sensors deployed on the ocean engineering structure, the real-time prediction monitoring module performs initial corrosion prediction by using a prediction model based on key features, and the multi-model integrated learning module performs initial corrosion prediction on the basis of the position of the ocean engineering structure in the ocean. A plurality of training models based on different algorithms are constructed, a visual user interface is provided by a visual module, and an optimized real-time corrosion prediction result is displayed; according to the method, the ocean engineering structure is divided into different areas, different training models are adopted for each area, environmental characteristics and corrosion mechanisms of different areas are fully considered, and the proportion weight of each factor is dynamically adjusted according to the actual conditions of different areas, so that the prediction result is closer to reality.

Description

Technical Field

[0001] The invention relates to the technical field of marine engineering structure corrosion prediction, and in particular to a method for predicting external corrosion of a marine engineering structure. Background Art

[0002] Marine engineering materials are susceptible to corrosion due to a variety of physical, chemical and biological factors in a complex marine environment, which not only causes significant economic losses but also threatens safety. Therefore, in-depth research on the corrosion laws of these materials in the marine environment is crucial for the development and utilization of marine resources, environmental protection and marine equipment manufacturing. Given the diversity and nonlinear characteristics of corrosion factors in the marine environment, traditional mathematical functions are difficult to apply.

[0003] At present, the scientific community mostly uses neural network models, gray correlation analysis, SWM and other advanced algorithms to predict the corrosion rate of marine engineering materials. For example, Song Shizhe et al. established a corrosion prediction model based on artificial neural networks, which comprehensively considered the influencing factors of seawater, carbon steel synthetic elements and their corrosion reactions; Zhu Xiangrong et al. used gray correlation analysis to explore the correlation of seawater corrosion; Deng Zhian et al. used fuzzy algorithms to optimize the neural network model for corrosion rate prediction; Hu Songqing et al. used the BP neural network model to study the internal corrosion reaction of oil pipelines. The application of these research methods has greatly promoted the accuracy and reliability of corrosion prediction of marine engineering materials.

[0004] However, existing methods for predicting corrosion of marine engineering structures often use a single prediction model to cover the entire marine environment, but the marine environment is complex and changeable, and the environmental characteristics and corrosion mechanisms of different regions vary significantly.

[0005] Therefore, there is an urgent need for a method for predicting external corrosion of marine engineering structures to solve the technical problems existing in the above-mentioned prior art. Summary of the invention

[0006] The present invention overcomes the deficiencies of the prior art and provides a method for predicting external corrosion of marine engineering structures.

[0007] To achieve the above object, the technical solution adopted by the present invention is: a method for predicting external corrosion of marine engineering structures, comprising: a data acquisition module, a real-time prediction and monitoring module, a multi-model integrated learning module and a visualization module;

[0008] The data acquisition module uses various sensors deployed on the marine engineering structure to collect the corrosion data of the marine engineering structure in real time, and pre-processes the collected corrosion data to extract key features;

[0009] The real-time prediction and monitoring module uses the prediction model to predict initial corrosion based on the key features to obtain prediction results, and sets an early warning threshold in the real-time prediction and monitoring module to determine the prediction results;

[0010] The multi-model integrated learning module constructs a plurality of training models based on different algorithms based on the location of the marine engineering structure in the ocean, introduces the prediction results into different training models using an integrated learning method, dynamically adjusts the weights of environmental influencing factors in the integrated learning, optimizes the prediction results, and obtains real-time corrosion prediction results;

[0011] The visualization module provides an intuitive user interface to display optimized real-time corrosion prediction results, historical corrosion trend analysis, and early warning information.

[0012] In a preferred embodiment of the present invention, the various sensors include corrosion monitoring sensors and environmental parameter sensors installed at various locations of the marine engineering structure; the preprocessing of the corrosion data includes: data cleaning, missing value processing, outlier detection and correction, and dimensionality reduction processing of the corrosion data using dimensionality reduction technology; the key features include: seawater temperature, salinity, pH value, dissolved oxygen, tidal changes, ocean current changes, structural material type, coating condition and age.

[0013] In a preferred embodiment of the present invention, the prediction model integrates the data contained in the key features into a unified data set, divides the data into a training set and a test set according to the characteristics of the data, uses a linear regression model to train the selected training set, and uses a test set to verify the linear regression model to obtain the prediction result. The formula expression of the linear regression model is:

[0014]

[0015] In the formula, y is the prediction result, that is, the predicted corrosion rate; β0 is the intercept term; For each feature x i The linear effect on the corrosion rate, where β i is the corresponding regression coefficient; is the interaction between feature data, that is, two different features x i and x j The effect of the presence of both on the corrosion rate, β ij is the regression coefficient of the interaction term; is the quadratic influence of each feature, that is, feature x i The square of the effect on the corrosion rate, β ii is the regression coefficient of the quadratic term; ε is the error term.

[0016] In a preferred embodiment of the present invention, based on the position of the marine engineering structure in the ocean, the marine engineering structure is divided into an ocean atmosphere zone, a wave splash zone, an ocean tidal range zone, a seawater full immersion zone and a seabed mud zone; the training models based on different algorithms are: support vector machine, random forest, gradient boosting decision tree, neural network and Bayesian network respectively; the training model used in the marine atmosphere zone is the support vector machine, the training model used in the wave splash zone is the random forest, the training model used in the ocean tidal range zone is the gradient boosting decision tree, the training model used in the seawater full immersion zone is the neural network, and the training model used in the seabed mud zone is the Bayesian network.

[0017] In a preferred embodiment of the present invention, the environmental influencing factors include: climate factors, seawater factors, marine biological factors and seabed soil factors. In the marine atmosphere zone, the climate factors account for the highest proportion, and the seabed soil factors account for the lowest proportion; in the wave splash zone and the ocean tidal zone, the seawater factors account for the highest proportion, followed by the climate factors, and the seabed soil factors account for the lowest proportion; in the seawater full immersion zone, the seawater factors account for the highest proportion, followed by the marine biological factors, and the seabed soil factors account for the lowest proportion; in the seabed soil zone, the seabed soil factors account for the highest proportion, and the climate factors account for the lowest proportion.

[0018] In a preferred embodiment of the present invention, the ensemble learning method optimizes the prediction result based on the change in the proportion of the prediction result and the environmental impact factor to obtain the real-time corrosion prediction result, and the formula expression of the real-time corrosion prediction result is:

[0019]

[0020] In the formula, y actual is the real-time corrosion prediction result, that is, the actual prediction value after considering all complex factors; y is the prediction result, that is, the predicted corrosion rate; ω i is the weight ratio of the i-th environmental impact factor; f i (E i ) is a function used to describe the i-th environmental impact factor E i How does it affect the predicted value? Further explanation is that this function is nonlinear or depends on other variables; m is the number of environmental factors; g is an adjustment function that depends on the variable h j , where variable h j are other types of environmental factors or conditions; n is the number of variables that the adjustment function g depends on; c k is the coefficient of the Kth additional adjustment term; d k is the value of the Kth additional adjustment item.

[0021] In a preferred embodiment of the present invention, the environmental impact factor E i The vector expression is: E=[E1,E2,E3......,E i ];

[0022] The environmental impact factor ω i The vector expression of weight proportion is: ω=[ω1,ω2,ω3......,ω i ].

[0023] In a preferred embodiment of the present invention, a computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0024] In a preferred embodiment of the present invention, a computer-readable storage medium stores a computer program, and the computer program implements the steps of the above method when executed by a processor.

[0025] The present invention solves the defects existing in the background technology and has the following beneficial effects:

[0026] (1) The present invention divides marine engineering structures into different areas, and adopts different training models for each area, taking full account of the environmental characteristics and corrosion mechanisms of different areas. Environmental influencing factors include climate factors, seawater factors, marine biological factors and seabed soil factors. The weight of each factor is dynamically adjusted according to the actual situation of different areas, so that the prediction results are closer to reality. In addition, the corrosion data is collected in real time through the data acquisition module, and preprocessing and key feature extraction are performed to ensure the quality and relevance of the input data. The real-time prediction and monitoring module uses the prediction model to make preliminary predictions, and the multi-model integrated learning module further combines multiple training models based on different algorithms to dynamically adjust the weight of environmental influencing factors, thereby optimizing the prediction results and improving the accuracy and reliability of the prediction.

[0027] (2) The present invention sets a warning threshold in the real-time prediction monitoring module to judge the prediction result. Once the prediction result exceeds the threshold, a warning can be triggered to promptly remind relevant personnel to take measures to prevent further deterioration of corrosion. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art are briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work.

[0029] Figure 1 is a data transmission flow chart of a preferred embodiment of the present invention;

[0030] Figure 2 It is a calculation flow chart of the prediction method of the preferred embodiment of the present invention. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0033] Structural corrosion in the marine environment is a complex and changeable process, which is affected by a variety of environmental factors. According to the different corrosion characteristics, the marine environment is subdivided into five specific zones: marine atmosphere zone, wave splash zone, ocean tidal zone, seawater immersion zone and seabed mud zone. Each zone exhibits different corrosion characteristics due to its unique environmental conditions. At present, there is obvious inaccuracy in using a single calculation prediction model or method to predict the corrosion condition of the overall structure, because this "one-size-fits-all" approach ignores the differences and dynamic changes in environmental conditions in each zone.

[0034] What needs to be further explained here is the characteristics of corrosion in the marine atmosphere: the atmosphere is the environmental medium that metal materials are most exposed to. According to statistics, 80% of metal components are used in the atmospheric environment. Most metal materials are in the marine atmospheric environment; and the marine atmosphere is rich in various salt particles, which are highly corrosive to metals. The atmospheric corrosion of metal materials is carried out by electrochemical mechanisms. When metals are exposed to the atmospheric environment, they absorb moisture from the atmospheric environment and form a thin layer of electrolyte film on the metal surface. The existence of this electrolyte film makes it possible for electrochemical reactions to occur on the metal surface. The amount of water absorbed by the metal surface is related to the relative humidity of the atmospheric environment. When the relative humidity is 75%, the thickness of the adsorbed water molecule layer is about 5 layers. Generally speaking, when the thickness of the water film is greater than 5 layers, the electrochemical corrosion process can be carried out. Atmospheric pollutants and atmospheric suspended particles will settle in the thin liquid layer on the metal surface and affect and participate in the atmospheric corrosion process of the material.

[0035] Corrosion characteristics of the splash zone: In the marine environment, the seawater droplets can be sprayed onto the surface, but the area that cannot be submerged by the seawater at high tide is usually called the splash zone. In the splash zone, the material is in a dry and wet alternating environment for a long time, with sufficient oxygen supply, and due to the splash of seawater, its droplets can directly hit the metal surface, causing it to corrode severely, generally 5 to 10 times that of the full seawater immersion area, which is a typical strong corrosion area in the marine environment. The Key Laboratory of Marine Environmental Corrosion and Protection of the Chinese Academy of Sciences has carried out continuous research on the corrosion mechanism and anti-corrosion technology of the marine splash zone, and successfully developed mineral coating protection technology to solve the corrosion problem in the splash zone. At the same time, the anti-corrosion and repair technology of reinforced concrete structures has been developed, which has been widely used in the protection of marine engineering structures.

[0036] Corrosion characteristics of the ocean tidal range area: The ocean tidal range area is between the average high tide line and the low tide line. This area is protected as the cathode area of ​​the oxygen concentration difference cell and has a low corrosion rate. However, there is marine biofouling and aerobic bacteria adhesion corrosion in this area. Corrosive organisms can accelerate the corrosion of metal substrates through various means such as oxygen concentration difference cells, induce local corrosion damage and other problems. It is reported that corrosion losses caused by microbial corrosion account for about 20% of the total corrosion losses.

[0037] Corrosion characteristics of the seawater full immersion area: In the seawater full immersion area, since corrosion is closely related to oxygen content, the corrosion rate decreases with increasing depth. However, this area covers a wide range. With the increase of depth, the seawater pressure, pH value, salinity, marine life and oxygen content will change significantly, the uniform corrosion rate will decrease, and local corrosion such as galvanic corrosion, crevice corrosion and pitting corrosion will intensify. In addition, the corrosion in this area is characterized by the interaction of multiple factors. The laws and mechanisms of hydrogen-induced cracking, stress corrosion and corrosion fatigue will change significantly, and the tendency of materials to undergo environmentally sensitive fractures in the marine environment will be more serious.

[0038] Corrosion characteristics of submarine mud areas: In submarine mud areas, the corrosion rate is usually low due to the low oxygen concentration. However, in the presence of corrosive microorganisms, the metabolic activity of microorganisms will accelerate the corrosion of metals and change the corrosion mechanism. The Institute of Oceanology, Chinese Academy of Sciences has accumulated a large number of achievements in the field of microbial corrosion research, studied the mechanism of marine microbial corrosion, developed sensor technology for rapid detection of corrosive microorganisms, and a variety of new material technologies for marine microbial corrosion protection.

[0039] like Figure 1 As shown, a method for predicting external corrosion of marine engineering structures includes: a data acquisition module, a real-time prediction and monitoring module, a multi-model integrated learning module and a visualization module.

[0040] like Figure 2 As shown, the data acquisition module uses various sensors deployed on marine engineering structures to collect corrosion data of marine engineering structures in real time, and pre-processes the collected corrosion data to extract key features;

[0041] It is further explained that various types of sensors include corrosion monitoring sensors and environmental parameter sensors installed in various locations of marine engineering structures; the preprocessing of corrosion data includes: data cleaning, missing value processing, outlier detection and correction, and dimensionality reduction processing of corrosion data using dimensionality reduction technology; key features include: seawater temperature, salinity, pH value, dissolved oxygen, tidal changes, ocean current changes, structural material type, coating condition and age.

[0042] The real-time prediction and monitoring module uses the prediction model to predict initial corrosion based on key features, obtains prediction results, and sets warning thresholds in the real-time prediction and monitoring module to judge the prediction results;

[0043] The prediction model integrates the data contained in the key features into a unified data set, divides it into a training set and a test set according to the characteristics of the data, uses a linear regression model to train the selected training set, and uses the test set to verify the linear regression model to obtain the prediction result. The formula expression of the linear regression model is:

[0044]

[0045] In the formula, y is the prediction result, that is, the predicted corrosion rate; β0 is the intercept term; For each feature x i The linear effect on the corrosion rate, where β i is the corresponding regression coefficient; is the interaction between feature data, that is, two different features x i and x j The effect of the presence of both on the corrosion rate, βij is the regression coefficient of the interaction term; is the quadratic influence of each feature, that is, feature x i The square of the effect on the corrosion rate, β ii is the regression coefficient of the quadratic term; ε is the error term.

[0046] The multi-model ensemble learning module builds multiple training models based on different algorithms based on the location of marine engineering structures in the ocean. It uses ensemble learning methods to introduce prediction results into different training models, dynamically adjusts the weights of environmental influencing factors in ensemble learning, optimizes prediction results, and obtains real-time corrosion prediction results.

[0047] It is further explained that, based on the position of the marine engineering structure in the ocean, the marine engineering structure is divided into the marine atmosphere zone, the wave splash zone, the ocean tidal range zone, the seawater full immersion zone and the seabed mud zone; the training models based on different algorithms are: support vector machine, random forest, gradient boosting decision tree, neural network and Bayesian network; the training model used in the marine atmosphere zone is support vector machine, the training model used in the wave splash zone is random forest, the training model used in the ocean tidal range zone is gradient boosting decision tree, the training model used in the seawater full immersion zone is neural network, and the training model used in the seabed mud zone is Bayesian network.

[0048] Environmental influencing factors include: climate factors, seawater factors, marine biological factors and seabed soil factors. In the marine atmosphere zone, climate factors account for the highest proportion, and seabed soil factors account for the lowest proportion; in the wave splash zone and the ocean tidal range zone, seawater factors account for the highest proportion, followed by climate factors, and seabed soil factors account for the lowest proportion; in the seawater full immersion area, seawater factors account for the highest proportion, followed by marine biological factors, and seabed soil factors account for the lowest proportion; in the seabed soil zone, seabed soil factors account for the highest proportion, and climate factors account for the lowest proportion.

[0049] The ensemble learning method optimizes the prediction results based on the changes in the proportion of the prediction results and environmental influencing factors to obtain the real-time corrosion prediction results. The formula for the real-time corrosion prediction results is:

[0050]

[0051] In the formula, y actual is the real-time corrosion prediction result, that is, the actual prediction value after considering all complex factors; y is the prediction result, that is, the predicted corrosion rate; ω i is the weight ratio of the i-th environmental impact factor; f i (E i ) is a function used to describe the i-th environmental impact factor E iHow does it affect the predicted value? Further explanation is that this function is nonlinear or depends on other variables; m is the number of environmental factors; g is an adjustment function that depends on the variable h j , where variable h j are other types of environmental factors or conditions; n is the number of variables that the adjustment function g depends on; c k is the coefficient of the Kth additional adjustment term; d k is the value of the Kth additional adjustment item.

[0052] Environmental impact factor E i The vector expression is: E=[E1,E2,E3......,E i ]; Environmental impact factor ω i The vector expression of weight proportion is: ω=[ω1,ω2,ω3......,ω i ].

[0053] The visualization module provides an intuitive user interface to display optimized real-time corrosion prediction results, historical corrosion trend analysis, and early warning information.

[0054] In a preferred embodiment, in order to improve the efficiency and accuracy of the dynamic adjustment mechanism, automation and intelligent technologies can be introduced. For example, a large amount of historical data can be trained and learned using machine learning algorithms to discover the complex relationship between environmental factors and corrosion rates; the Internet of Things technology can be used to achieve real-time data transmission and processing; and artificial intelligence technology can be used to achieve intelligent analysis and decision-making of prediction results.

[0055] Take seawater temperature as an example. When seawater temperature rises, on the one hand, it will accelerate the corrosion reaction of metals (because the temperature increase will increase the reaction rate); on the other hand, the temperature increase may also reduce the amount of dissolved oxygen in seawater (because the solubility of oxygen decreases with the temperature increase), thereby slowing down the corrosion process. The dynamic adjustment mechanism needs to comprehensively consider these two opposite effects and automatically adjust the proportion of the seawater temperature factor in the overall environmental factors according to the actual situation. If the experimental data and model analysis show that the current temperature rise has a major impact on the corrosion rate, then the weight of the seawater temperature factor in the prediction model should be increased accordingly.

[0056] Embodiment 1

[0057] In order to predict the corrosion rate of marine engineering materials in the marine environment, this embodiment specially selected three representative materials, 5052 aluminum alloy, Ti-6Al-4V and white copper B30, as experimental objects. All data are from the China Corrosion and Protection Network.

[0058] In view of the influencing factors of marine engineering material corrosion, this embodiment uses a multi-model ensemble learning algorithm based on the collected environmental data to conduct an in-depth analysis and calculate the influence rate of each relevant influencing factor. As shown in Table 1, the influence rate table of influencing factors clearly reveals that the six factors of water temperature, dissolved oxygen content, salinity, conductivity, pH value and redox potential in the marine environment are closely related to the corrosion rate of materials during the corrosion process of marine engineering materials.

[0059] Table 1

[0060]

[0061] The relevant parameter data of 5052 aluminum alloy, Ti-6Al-4V and white copper B30 are input as training data into the training model of the multi-model integrated learning module for training. The surrounding seawater temperature, salinity, pH value, dissolved oxygen, tidal changes, ocean current changes of 5052 aluminum alloy, Ti-6Al-4V and white copper B30, as well as the structural material type, coating condition and age of the materials are used as input data, and the threshold and weight are input into the model integrated learning module, and the real-time corrosion prediction results are calculated by the integrated learning method.

[0062] In this embodiment, the learning efficiency of the ensemble learning method is 0.5, the error capacity is 0.00001, and the number of repeated training is 10000. After multiple experiments, the number of iterations is set to 100 to ensure that the optimal real-time corrosion prediction result is obtained through the ensemble learning method.

[0063] After repeated training of the above training model, the real-time corrosion prediction results of 5052 aluminum alloy, Ti-6Al-4V and nickel silver B30 are shown in Table 2.

[0064] Table 2

[0065] Material Brand True value Predicted value Error rate time consuming 5052 Aluminum Alloy 0.005 0.00486 2.8％ 3.6 / ms Ti-6Al-4V 0.0196 0.0186 5.1％ 6.1 / ms White copper B30 0.0135 0.0131 3.0％ 4.2 / ms

[0066] This study explored the construction of multiple training models based on different algorithms based on the location of marine engineering structures in the ocean, the use of ensemble learning methods to introduce prediction results into different training models, dynamically adjust the weight of environmental influencing factors in ensemble learning, optimize prediction results, and obtain real-time corrosion prediction results. By using this fusion method, the data of three marine engineering materials, 5052 aluminum alloy, Ti-6Al-4V and white copper B30, in seawater environment were trained to build a prediction model.

[0067] The model showed the advantages of reducing prediction error, reducing average time consumption and significantly improving prediction accuracy when predicting the seawater corrosion rate of marine engineering materials. In short, the use of the integrated learning method model not only improves the actual accuracy of the prediction of the corrosion rate of marine engineering materials, but also helps to reduce related hazards, and shows good scalability, applicability and promotion potential.

[0068] When the present invention is used, corrosion monitoring sensors and environmental parameter sensors deployed on marine engineering structures are used to collect corrosion-related data in real time, and the collected data is preprocessed, including data cleaning, missing value processing, outlier detection and correction, and dimensionality reduction processing, to extract key features.

[0069] Among them, the key characteristics include sea water temperature, salinity, pH value, dissolved oxygen, tidal changes, ocean current changes, structural material type, coating condition and age, etc. These characteristics will serve as important inputs for subsequent prediction models.

[0070] Based on the key features extracted by the data acquisition module, the prediction model is used to make preliminary corrosion predictions, and the prediction results are obtained. The warning threshold is set in the real-time prediction monitoring module to judge the prediction results. If the prediction result exceeds the threshold, an early warning is triggered to remind relevant personnel to take timely measures.

[0071] Marine engineering structures are divided into different regions according to their locations in the ocean. For each region, training models based on different algorithms are constructed. The preliminary prediction results are introduced into different training models using an integrated learning method. The prediction results are optimized based on the weights of environmental influencing factors (such as climate factors, seawater factors, marine biological factors, and seabed soil factors).

[0072] By dynamically adjusting the weights of environmental influencing factors and considering the interactions between characteristic data and the effects of quadratic terms, more accurate real-time corrosion prediction results can be obtained. An intuitive user interface is provided to display optimized real-time corrosion prediction results, allowing relevant personnel to clearly understand the corrosion situation and prediction trends. Historical corrosion trend analysis is displayed to help relevant personnel understand the changing patterns and trends of corrosion rates.

[0073] Display early warning information. When the prediction result exceeds the warning threshold, timely remind relevant personnel to take measures to prevent further deterioration of corrosion.

[0074] The above is based on the ideal embodiment of the present invention. Through the above description, relevant personnel can make various changes and modifications without departing from the technical concept of the present invention. The technical scope of the present invention is not limited to the content in the specification, and the technical scope must be determined according to the scope of the claims.

Claims

1. A method for predicting external corrosion of marine engineering structures, comprising: Data acquisition module, real-time prediction and monitoring module, multi-model integrated learning module and visualization module; characterized in that, The data acquisition module uses various sensors deployed on the marine engineering structure to collect the corrosion data of the marine engineering structure in real time, and pre-processes the collected corrosion data to extract key features; The real-time prediction and monitoring module uses the prediction model to predict initial corrosion based on the key features to obtain prediction results, and sets an early warning threshold in the real-time prediction and monitoring module to determine the prediction results; The multi-model integrated learning module constructs a plurality of training models based on different algorithms based on the location of the marine engineering structure in the ocean, introduces the prediction results into different training models using an integrated learning method, dynamically adjusts the weights of environmental influencing factors in the integrated learning, optimizes the prediction results, and obtains real-time corrosion prediction results; The visualization module provides an intuitive user interface to display optimized real-time corrosion prediction results, historical corrosion trend analysis, and early warning information.

2. A method for predicting external corrosion of marine engineering structures according to claim 1, characterized in that: The various sensors include corrosion monitoring sensors and environmental parameter sensors installed at various locations of the marine engineering structures; the preprocessing of the corrosion data includes: data cleaning, missing value processing, outlier detection and correction, and dimensionality reduction processing of the corrosion data using dimensionality reduction technology; the key features include: seawater temperature, salinity, pH value, dissolved oxygen, tidal changes, ocean current changes, structural material type, coating condition and age.

3. The method for predicting external corrosion of marine engineering structures according to claim 1, characterized in that: The prediction model integrates the data contained in the key features into a unified data set, divides the data into a training set and a test set according to the characteristics of the data, uses a linear regression model to train the selected training set, and uses a test set to verify the linear regression model to obtain the prediction result. The formula expression of the linear regression model is: In the formula, y is the prediction result, that is, the predicted corrosion rate; β0 is the intercept term; For each feature x i The linear effect on the corrosion rate, where β i is the corresponding regression coefficient; is the interaction between feature data, that is, two different features x i and x j The effect of the presence of both on the corrosion rate, β ij is the regression coefficient of the interaction term; is the quadratic influence of each feature, that is, feature x i The square of the effect on the corrosion rate, β ii is the regression coefficient of the quadratic term; ε is the error term.

4. The method for predicting external corrosion of marine engineering structures according to claim 1, characterized in that: Based on the position of the marine engineering structure in the ocean, the marine engineering structure is divided into a marine atmosphere zone, a wave splash zone, a marine tidal range zone, a seawater full immersion zone and a seabed mud zone; the training models based on different algorithms are: support vector machine, random forest, gradient boosting decision tree, neural network and Bayesian network respectively; the training model used in the marine atmosphere zone is the support vector machine, the training model used in the wave splash zone is the random forest, the training model used in the marine tidal range zone is the gradient boosting decision tree, the training model used in the seawater full immersion zone is the neural network, and the training model used in the seabed mud zone is the Bayesian network.

5. A method for predicting external corrosion of marine engineering structures according to claims 1 and 4, characterized in that: The environmental influencing factors include: climate factors, seawater factors, marine biological factors and seabed soil factors. In the marine atmosphere zone, the climate factors account for the highest proportion, and the seabed soil factors account for the lowest proportion; in the wave splash zone and the ocean tidal zone, the seawater factors account for the highest proportion, followed by the climate factors, and the seabed soil factors account for the lowest proportion; in the seawater full immersion zone, the seawater factors account for the highest proportion, followed by the marine biological factors, and the seabed soil factors account for the lowest proportion; in the seabed soil zone, the seabed soil factors account for the highest proportion, and the climate factors account for the lowest proportion.

6. A method for predicting external corrosion of marine engineering structures according to claims 1, 3 and 5, characterized in that: The ensemble learning method optimizes the prediction result based on the prediction result and the change in the proportion of the environmental influencing factors to obtain the real-time corrosion prediction result. The formula expression of the real-time corrosion prediction result is: In the formula, y actual is the real-time corrosion prediction result, that is, the actual prediction value after considering all complex factors; y is the prediction result, that is, the predicted corrosion rate; ω i is the weight ratio of the i-th environmental impact factor; f i (E i ) is a function used to describe the i-th environmental impact factor E i How it affects the predicted value, further indicating that the function is nonlinear or depends on other variables; m is the number of environmental factors; g is an adjustment function that depends on the variable h j , where variable h j are other types of environmental factors or conditions; n is the number of variables that the adjustment function g depends on; c k is the coefficient of the Kth additional adjustment term; d k is the value of the Kth additional adjustment item.

7. A method for predicting external corrosion of marine engineering structures according to claim 6, characterized in that: The environmental impact factor E i The vector expression is: E=[E1,E2,E3......,E i ]; the environmental impact factor ω i The vector expression of weight proportion is: ω=[ω1,ω2,ω3......,ω i ].

8. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images