Nuclear power plant seawater pipeline inner wall coating aging monitoring device

Abstract

The invention belongs to the technical field of nuclear power plant seawater pipeline inner wall coating aging monitoring and evaluation, and particularly relates to a nuclear power plant seawater pipeline inner wall coating aging monitoring device. The watertight cover is provided with a probe flange, the array electrode, the reference electrode and the auxiliary anode are located in an insulating part, the insulating part is installed in a pipeline, and an installation flange is arranged on the outer side of the insulating part and connected with the probe flange. The intelligent detector is respectively connected with the array electrode, the reference electrode and the auxiliary anode through an array electrode cable, a reference electrode cable and an auxiliary anode cable; and the array electrode cable, the reference electrode cable and the auxiliary anode cable penetrate through the watertight cover and the insulating part. The mapping relation between the damage rate of the seawater pipeline inner wall coating and the cathode protection current is established, so that the failure condition of the coating is judged and evaluated by monitoring the magnitude of the cathode protection current of the nuclear power plant seawater pipeline inner wall, and the corrosion state of the seawater pipeline inner wall is mastered in real time. And online monitoring and early warning of the failure process of the inner wall coating of the seawater pipeline are realized.

Description

Technical Field

[0001] The invention belongs to the technical field of aging monitoring and evaluation of inner wall coatings of seawater pipelines in nuclear power plants, and particularly relates to an aging monitoring device for inner wall coatings of seawater pipelines in nuclear power plants. Background Art

[0002] The inner wall of seawater cooling source pipelines is protected against corrosion using a combination of coating and cathodic protection to mitigate seawater corrosion. Cathodic protection typically uses sacrificial anodes to prevent over-protection and cathodic delamination of the coating. Under normal circumstances, the inner wall coating failure process of seawater pipelines involves electrolyte penetration of the coating pores, the coating pores filling with electrolyte (the coating capacitance remains essentially unchanged), the electrolyte reaching the base metal surface (metal corrosion begins), and the coating failing due to visible bubbling (the corrosion reaction pathway is opened).

[0003] Assuming the coating on the inner wall of a seawater pipeline is uniform and defect-free during the coating process, under the protection of a combined sacrificial anode, the coating's resistance decreases and its capacitance increases as seawater gradually penetrates its internal pores. Once seawater fills the coating's pores and reaches the inner metal surface, the coating's capacitance and resistance characteristics stabilize. Over a long period of time, a cathodic reaction occurs at the pipe-metal interface, the reaction rate of which is dependent on the corrosive environment within the pipe. Once the coating's water absorption reaches saturation, the coating itself remains largely unchanged, except for a possible significant effect on the diffusion of electrochemical reactants. From this point on, until blistering occurs, the coating is considered intact. In actual nuclear power plant seawater pipelines, blistering typically begins in small, localized areas, often due to localized coating defects or the inability of the sacrificial anode to protect certain areas during alternating wet-dry conditions. Less frequently, bare metal may also be exposed. Therefore, in most cases, attention should be paid to coating damage and failure after the coating's pores are filled with electrolyte and the solution reaches the inner metal. This stage is referred to as the mid-immersion phase of the coating.

[0004] Currently, the most mature coating monitoring method involves monitoring certain impedance parameters of coating specimens, such as characteristic frequency and coating capacitance. Failure of coatings on the inner wall of nuclear power plant seawater pipelines originates locally or in micro-regions, resulting in uncertainty. Small coating probes cannot represent large-scale coating failures. During the mid-immersion period, the coating damage rate is related to AC impedance measurements such as coating capacitance and characteristic frequency, as well as the cathodic protection current. At a given cathodic protection potential, the greater the coating damage rate, the greater the cathodic protection current density. However, the cathodic protection current density of the pipeline inner wall is also affected by environmental factors such as seawater flow rate, dissolved oxygen concentration, and temperature. Therefore, a mapping relationship between coating damage rate and cathodic protection current density was established through seawater pipeline dynamic bench testing and AI algorithms. Coating damage was then monitored based on the cathodic protection current density. By constructing a digital twin of coating damage and combining it with the spatiotemporal evolution of the cathodic protection current, an intelligent evaluation system with environmental adaptability was established. According to the current coating aging grading standard (GB / T1766-2008), coating bubble density is classified into six levels, ranging from 0 to 5, with 0 indicating no visible bubbles and 5 indicating dense bubbles. This rating system is inherently subjective and general, and even if a precise equation for coating damage rate existed, it would not be directly applicable. Therefore, it is necessary to develop a device and evaluation method for monitoring the aging of the inner wall coating of nuclear power plant seawater pipelines. A grading system for the pipeline coating damage rate, represented by the cathodic protection current density of the inner wall of nuclear power plant seawater pipelines, should be established within a reasonable range to guide aging monitoring and early warning of the inner wall coating of nuclear power plant seawater pipelines. Summary of the Invention

[0005] The purpose of the present invention is to provide a device for monitoring the aging of the inner wall coating of a nuclear power plant seawater pipeline. Based on machine learning modeling technology, a mapping relationship between the damage rate of the inner wall coating of the seawater pipeline and the cathodic protection current is established. By monitoring the magnitude of the cathodic protection current of the inner wall of the nuclear power plant seawater pipeline, the failure of the coating can be judged and evaluated, the corrosion state of the inner wall of the seawater pipeline can be grasped in real time, and online monitoring and early warning of the failure process of the inner wall coating of the seawater pipeline can be realized.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is:

[0007] A device for monitoring aging of the inner wall coating of a seawater pipeline in a nuclear power plant comprises a watertight cover provided with a probe flange, an array electrode, a reference electrode and an auxiliary anode located in an insulating member installed in the pipeline, a mounting flange provided on the outside of the insulating member, the mounting flange being connected to the probe flange, an intelligent detector being connected to the array electrode, the reference electrode and the auxiliary anode via array electrode cables, reference electrode cables and auxiliary anode cables respectively, the array electrode cables, reference electrode cables and auxiliary anode cables passing through the watertight cover and the insulating member.

[0008] Monitoring methods include:

[0009] a) Multi-source data fusion acquisition: A monitoring device is installed on a nuclear power plant seawater pipeline or a simulated seawater pipeline. Array electrodes are used to simulate specimens with different coating damage rates. The array electrodes are electrically connected to the pipeline. Auxiliary anodes are used to apply cathodic protection to the pipeline. Reference electrodes monitor the pipeline protection potential. An intelligent monitor is used to monitor the protection current density passing through the array electrodes after the pipeline reaches the normal cathodic protection potential.

[0010] b) Intelligent data preprocessing: Establishing a mapping relationship between the coating damage rate and cathodic protection current of seawater pipelines, and forming a database of seawater pipeline coating damage rate, cathodic protection potential, and current density;

[0011] c) Machine Learning Model Construction:

[0012] 1) Training phase: A training set was constructed using dynamic water test data, hyperparameters were tuned using Bayesian optimization, and a hybrid prediction model combining a random forest regression model and a deep neural network was established;

[0013] 2) Validation phase: Model performance was evaluated through k-fold cross-validation and feature interpretability analysis was performed using SHAP values;

[0014] 3) Deployment phase: The optimized model is integrated into an intelligent monitoring instrument, establishing a real-time data stream processing channel. The intelligent monitoring instrument can determine the coating damage rate by monitoring the cathodic protection current density of the pipeline, thereby evaluating the aging status of the inner wall coating of the seawater pipeline;

[0015] 4) Online monitoring and prediction: The deployed machine learning model analyzes the cathodic protection current characteristics in real time and outputs a dynamic trend chart of the coating damage rate. When the predicted coating damage rate exceeds the threshold, a graded warning is triggered. At the same time, the remaining life of the coating is estimated based on the time series prediction model.

[0016] The mapping relationship between the coating damage rate and cathodic protection current of the seawater pipeline is established through orthogonal experiments or other statistical test methods and numerical simulations.

[0017] Machine learning specifically includes:

[0018] a) Establish a cathodic protection dynamic model based on machine learning: by collecting multi-dimensional data of array electrodes in real time, constructing a spatiotemporal feature matrix, and using a random forest algorithm to rank feature importance and screen key modeling parameters;

[0019] b) Develop a deep neural network prediction model: Construct a hybrid neural network architecture with an LSTM layer. The input layer receives the real-time monitored cathodic protection current density, ambient temperature, and seawater flow rate parameters. The nonlinear relationship is fitted through three hidden layers. The output layer provides the coating damage rate prediction value and confidence interval.

[0020] c) Establish an adaptive optimization mechanism: Using an online learning algorithm, when the deviation between the monitoring data and the predicted value exceeds the preset threshold, the model parameter update is automatically triggered, and the model's dynamic tracking capability of the pipeline aging process is maintained through incremental learning.

[0021] The multi-dimensional data of array electrodes include current density distribution, potential gradient, and flow field parameters.

[0022] The array electrode is a standard cylinder made of Q235 steel, which is consistent with the material of the pipeline.

[0023] The reference electrode was a silver / silver chloride electrode.

[0024] The auxiliary electrode is a metal oxide coated cylindrical electrode with a surface area significantly larger than that of the array electrode.

[0025] The array electrodes are 5×5 arrays.

[0026] The array electrodes are 10×10 arrays.

[0027] The beneficial effects achieved by the present invention are:

[0028] This invention enables online, real-time monitoring of the inner coating of nuclear power plant seawater pipelines, eliminating the need to shut down the pipelines for drainage and manual visual inspection. The monitoring scope covers the coating damage rate over a large area of the seawater pipeline, avoiding the limitations of existing monitoring methods that only monitor or detect the degree of coating aging at a specific point. The monitoring data is highly reliable and representative. This is in-situ monitoring, eliminating the need for external coating test specimens. Combined with cathodic protection data, it enables real-time monitoring, resulting in high monitoring efficiency. The more data mapping the coating damage rate and cathodic protection current, the more accurate the monitoring, minimizing the impact of external environmental factors. By establishing a database of coating failure processes throughout the coating's lifecycle, the remaining coating life can be predicted. The array electrode simulates coating damage rates with high accuracy, capable of simulating coating damage with gradients of 1% or even smaller. A new machine learning-based intelligent evaluation system automatically identifies nonlinear mapping relationships in complex environments, improving prediction accuracy by over 30% compared to traditional statistical models. Utilizing deep feature extraction technology, it effectively processes multi-physics field coupled data, achieving 95% accuracy in identifying special operating conditions such as localized corrosion and abnormal damage. Therefore, the nuclear power plant seawater pipeline inner wall coating aging monitoring device and evaluation method of the present invention have the advantages of real-time, high detection efficiency and high reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 This is a schematic front view cross-sectional diagram of a device for monitoring aging of the inner wall coating of a seawater pipeline in a nuclear power plant;

[0030] Figure 2A three-dimensional diagram of a device for monitoring the aging of the inner wall coating of a seawater pipeline in a nuclear power plant;

[0031] In the figure: 1 is the intelligent detector, 2 is the array electrode cable, 3 is the reference electrode cable, 4 is the auxiliary anode cable, 5 is the watertight cover, 6 is the probe flange, 7 is the array electrode, 8 is the reference electrode, 9 is the auxiliary anode, 10 is the insulating part, 11 is the mounting flange, and 12 is the pipeline. DETAILED DESCRIPTION

[0032] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0033] A nuclear power plant seawater cooling source pipeline coating monitoring device includes an intelligent monitoring instrument, a monitoring probe, an auxiliary anode cable, a reference electrode cable, an array electrode cable, a mounting flange, a machine learning modeling system, and intelligent monitoring software. The monitoring probe includes an array electrode, an auxiliary anode, and a reference electrode. The accompanying machine learning modeling system includes a data preprocessing module, a feature engineering module, a model training module, and an online prediction module. A corresponding method for using the nuclear power plant seawater cooling source pipeline coating monitoring device has also been developed. This invention is suitable for measuring the coating damage rate and cathodic protection current mapping data of the inner wall of nuclear power plant seawater pipelines. It can also monitor the failure of the inner wall coating of seawater pipelines based on a database established through dynamic water testing and numerical simulation.

[0034] Machine learning methods include:

[0035] a) Establish a cathodic protection dynamic model based on machine learning: By collecting multi-dimensional data of array electrodes in real time (including current density distribution, potential gradient, and flow field parameters), a spatiotemporal feature matrix is constructed. The random forest algorithm is used to rank the feature importance and screen key modeling parameters.

[0036] b) Develop a deep neural network prediction model: Construct a hybrid neural network architecture containing an LSTM layer. The input layer receives real-time monitored parameters such as cathodic protection current density, ambient temperature, and seawater flow rate. Nonlinear relationships are fitted through three hidden layers. The output layer provides a predicted value and confidence interval for the coating damage rate.

[0037] c) Establish an adaptive optimization mechanism: Using an online learning algorithm, when the deviation between the monitoring data and the predicted value exceeds the preset threshold, the model parameter update is automatically triggered, and the model's dynamic tracking capability of the pipeline aging process is maintained through incremental learning.

[0038] Specific monitoring methods include:

[0039] a) Multi-source data fusion acquisition: A coating monitoring probe is installed on a nuclear power plant seawater pipeline or a simulated seawater pipeline. The probe's array electrodes are used to simulate specimens with different coating damage rates. The array electrodes are electrically connected to the pipeline. Auxiliary anodes are used to apply cathodic protection to the pipeline. Reference electrodes monitor the pipeline protection potential. An intelligent monitor is used to monitor the protection current density passing through the array electrodes after the pipeline reaches the normal cathodic protection potential.

[0040] b) Intelligent data preprocessing: Establish a mapping relationship between the coating damage rate and cathodic protection current of seawater pipelines through orthogonal experiments or other statistical test methods and numerical simulations, and form a database of seawater pipeline coating damage rate, cathodic protection potential, and current density;

[0041] c) Machine Learning Model Construction:

[0042] 1) Training phase: The dynamic water test data was used to construct a training set, Bayesian optimization was used to tune hyperparameters, and a hybrid prediction model of random forest regression model and deep neural network was established.

[0043] 2) Validation phase: Model performance was evaluated through k-fold cross-validation, and feature interpretability analysis was performed using SHAP values.

[0044] 3) Deployment phase: The optimized model is integrated into an intelligent monitoring instrument to establish a real-time data stream processing channel. The intelligent monitoring instrument can determine the coating damage rate by monitoring the cathodic protection current density of the pipeline, thereby evaluating the aging status of the coating on the inner wall of the seawater pipeline.

[0045] 4) Online monitoring and prediction: The deployed machine learning model analyzes the cathodic protection current characteristics in real time and outputs a dynamic trend chart of the coating damage rate. When the predicted coating damage rate exceeds the threshold, a graded warning is triggered. At the same time, the remaining life of the coating is estimated based on the time series prediction model.

[0046] A coating monitoring device for seawater cooling source pipelines in nuclear power plants includes an intelligent detector 1, an array electrode cable 2, a reference electrode cable 3, an auxiliary anode cable 4, a watertight cover 5, a probe flange 6, an array electrode 7, a reference electrode 8, an auxiliary anode 9, and an insulating member 10. The coating monitoring probe is mounted on the inner wall of the pipeline via a flange connection, slightly protruding from or flush with the inner wall. The coating monitoring probe is connected to the intelligent detector via the array electrode cable, reference electrode cable, and auxiliary anode cable.

[0047] The watertight cover 5 is provided with a probe flange 6, the array electrode 7, the reference electrode 8 and the auxiliary anode 9 are located in the insulating part 10, the insulating part 10 is installed in the pipe 12, and a mounting flange 11 is provided on the outside of the insulating part 10, and the mounting flange 11 is connected to the probe flange 6. The intelligent detector 1 is connected to the array electrode 7, the reference electrode cable 3 and the auxiliary anode cable 4 respectively. The array electrode cable 2, the reference electrode cable 3 and the auxiliary anode cable 4 pass through the watertight cover 5 and the insulating part 10.

[0048] Example:

[0049] Step 1: Intelligent data acquisition system construction

[0050] The seawater cooling source pipeline coating monitoring device developed by the present invention was used at a nuclear power plant's seawater cooling source pipeline dynamic water test site to monitor pipeline coatings. The pipeline under test was made of Q235 steel and had a length of L = 10m. The pipeline coating monitoring probe developed by the present invention was used. The probe consisted of a working electrode, a reference electrode, and an auxiliary electrode. The working electrode was a standard Q235 steel cylinder, consistent with the pipeline material; the reference electrode was a silver / silver chloride (Ag / AgCl) electrode; and the auxiliary electrode was a metal oxide (MMO)-coated cylindrical electrode with a significantly larger surface area than the working electrode. Each electrode was encapsulated in a corrosion-resistant housing with insulating sealant and secured to the pipeline's inner wall via a flange assembly, with its end faces flush with or slightly convex to the inner wall. During installation, a temporary protective structure was placed on the surfaces of the auxiliary and reference electrodes. The working electrode surface and the pipeline inner wall were coated using the same pretreatment and coating processes. During the monitoring process, the working electrode of the probe is insulated from the monitored pipeline. The monitoring probe independently provides a reference potential signal and is insulated and disconnected from the monitored pipeline. The probe is externally connected to an independent intelligent cathodic protection potential monitoring and polarization device. Its cathodic protection potential is set to be consistent with the cathodic protection potential of the pipeline. Eight groups of monitoring probes are arranged at equal intervals inside the Q235 steel pipeline being measured (L = 10m).

[0051] Step 2: Construction of spatiotemporal feature database

[0052] Construct a multidimensional dataset, including cathodic protection characteristic parameter data, environmental parameter time series, and coating status labels. Cathodic protection characteristic parameters include cathodic protection potential, cathodic protection current, reference electrode monitoring potential, output potential, and output current parameters; environmental parameter time series include minute-level continuous monitoring data of seawater flow rate, temperature, salinity, conductivity, and pH; coating status labels include coating damage rate data.

[0053] Step 3: Machine Learning Model Deployment

[0054] A spatiotemporal convolutional feature extractor was used to extract the cathodic protection current density distribution data for the Q235 pipeline. The input layer included cathodic protection characteristic parameter data and environmental parameter time series; the output layer included the coating damage rate. Data pre-training was performed on the aforementioned laboratory dynamic water test bench, and the coating damage rate was mapped using a Sigmoid activation function to form a machine learning model.

[0055] Step 4: Online monitoring and quantitative diagnosis

[0056] By collecting the polarization current density data of the working electrode in real time, combining it with a coating state analysis model based on an AI algorithm, and comparing the current density change characteristics with the preset threshold, quantitative diagnosis of the coating damage rate can be achieved.

[0057] The present invention discloses a coating monitoring device for seawater cooling source pipelines in nuclear power plants and a method for use, which belongs to the technical field of nuclear power plant cooling source pipeline monitoring. In view of the problem that existing coating monitoring methods are limited to local detection, rely on subjective ratings, and cannot evaluate large-scale coating damage online in real time, the present invention proposes a monitoring scheme based on the relationship between array electrode simulation and cathodic protection current mapping. The device includes components such as an intelligent monitor, a monitoring probe (including array electrodes, reference electrodes, and auxiliary anodes), and a mounting flange. It simulates different coating damage rates through array electrode gradients, and combines orthogonal experiments with AI algorithms to establish a mapping database of coating damage rate and cathodic protection current density to achieve real-time online monitoring. The monitoring method of the present invention can obtain the coating status of a large area of the pipeline in situ without the need to stop work for inspection. It has strong resistance to environmental interference, high reliability of monitoring data, and supports the prediction of the remaining life of the coating. This technology solves the problem of early warning of coating failure of seawater cooling source pipelines in nuclear power plants. It has the advantages of high efficiency, precision, and strong real-time performance, and is suitable for the long-term health management of the anti-corrosion layer of the cooling source pipelines of nuclear power plants.

[0058] A device and evaluation method for monitoring the aging of the coating on the inner wall of a nuclear power plant seawater pipeline. For the monitored nuclear power plant seawater pipeline, a coating aging monitoring probe is set on the inner wall of the seawater pipeline. The array electrode, auxiliary anode and reference electrode provided on the coating aging monitoring probe are used. The material and coating process of the array electrode are consistent with those of the monitored pipeline. Under a specific cathodic protection potential, the cathodic protection current density on the array electrode is monitored. Combined with a coating state analysis model based on an AI artificial intelligence algorithm, the cathode current density variation characteristics of the array electrode are compared with a preset threshold value to achieve quantitative diagnosis of the coating damage rate, thereby realizing the detection and evaluation of the aging state of the coating on the inner wall of the seawater pipeline.

[0059] The coating aging monitoring probe includes an auxiliary anode, a reference electrode, and an array electrode. The array electrode is a 5×5 array, which can also be made into a 10×10 array. The auxiliary anode is an MMO cylindrical anode or a platinum-niobium cylindrical anode, and the reference electrode is a high-purity zinc or Ag / AgCl reference electrode. The coating aging monitoring probe is insulated from the monitored pipeline. The cathodic protection potential of the array electrode in the coating aging monitoring probe is kept consistent with the cathodic protection potential of the pipeline for a long time.

[0060] Through seawater pipeline bench dynamic water tests or other statistical test methods and numerical simulations, a mapping relationship between the coating damage rate of the inner wall of the seawater pipeline and the cathodic protection current is established, and a database of seawater pipeline coating damage rate, cathodic protection potential, and current density is formed. An AI algorithm is used to establish a monitoring model. Then, the intelligent monitor can determine the coating damage rate by monitoring the cathodic protection current density of the pipeline, thereby realizing the monitoring and evaluation of the aging status of the inner wall coating of the seawater pipeline.

[0061] The above embodiments are intended only to illustrate the technical concepts and features of the present invention. Their purpose is to enable those skilled in the art to understand the contents of the present invention and implement them accordingly. They are not intended to limit the scope of protection of the present invention. Any equivalent changes or modifications made in accordance with the spirit of the present invention are intended to be covered by the scope of protection of the present invention.

Claims

1. A device for monitoring the aging of the inner wall coating of a seawater pipeline in a nuclear power plant, characterized by: The watertight cover is provided with a probe flange, the array electrode, reference electrode and auxiliary anode are located in the insulating part, the insulating part is installed in the pipeline, a mounting flange is provided on the outside of the insulating part, the mounting flange is connected to the probe flange, the intelligent detector is connected to the array electrode, reference electrode and auxiliary anode through the array electrode cable, reference electrode cable and auxiliary anode cable respectively, and the array electrode cable, reference electrode cable and auxiliary anode cable pass through the watertight cover and the insulating part.

2. The device for monitoring the aging of the inner wall coating of a seawater pipeline in a nuclear power plant according to claim 1, characterized in that: Monitoring methods include: a) Multi-source data fusion acquisition: A monitoring device is installed on a nuclear power plant seawater pipeline or a simulated seawater pipeline. Array electrodes are used to simulate specimens with different coating damage rates. The array electrodes are electrically connected to the pipeline. Auxiliary anodes are used to apply cathodic protection to the pipeline. Reference electrodes monitor the pipeline protection potential. An intelligent monitor is used to monitor the protection current density passing through the array electrodes after the pipeline reaches the normal cathodic protection potential. b) Intelligent data preprocessing: Establishing a mapping relationship between the coating damage rate and cathodic protection current of seawater pipelines, and forming a database of seawater pipeline coating damage rate, cathodic protection potential, and current density; c) Machine Learning Model Construction: 1) Training phase: A training set was constructed using dynamic water test data, hyperparameters were tuned using Bayesian optimization, and a hybrid prediction model combining a random forest regression model and a deep neural network was established; 2) Validation phase: Model performance was evaluated through k-fold cross-validation and feature interpretability analysis was performed using SHAP values; 3) Deployment phase: The optimized model is integrated into an intelligent monitoring instrument, establishing a real-time data stream processing channel. The intelligent monitoring instrument can determine the coating damage rate by monitoring the cathodic protection current density of the pipeline, thereby evaluating the aging status of the inner wall coating of the seawater pipeline; 4) Online monitoring and prediction: The deployed machine learning model analyzes the cathodic protection current characteristics in real time and outputs a dynamic trend chart of the coating damage rate. When the predicted coating damage rate exceeds the threshold, a graded warning is triggered. At the same time, the remaining life of the coating is estimated based on the time series prediction model.

3. The device for monitoring the aging of the inner wall coating of a seawater pipeline in a nuclear power plant according to claim 2, characterized in that: The mapping relationship between the coating damage rate and cathodic protection current of the seawater pipeline is established through orthogonal experiments or other statistical test methods and numerical simulations.

4. The device for monitoring aging of inner wall coating of seawater pipeline in nuclear power plant according to claim 2, characterized in that: Machine learning specifically includes: a) Establish a cathodic protection dynamic model based on machine learning: by collecting multi-dimensional data of array electrodes in real time, constructing a spatiotemporal feature matrix, and using a random forest algorithm to rank feature importance and screen key modeling parameters; b) Develop a deep neural network prediction model: Construct a hybrid neural network architecture with an LSTM layer. The input layer receives the real-time monitored cathodic protection current density, ambient temperature, and seawater flow rate parameters. The nonlinear relationship is fitted through three hidden layers. The output layer provides the coating damage rate prediction value and confidence interval. c) Establish an adaptive optimization mechanism: Using an online learning algorithm, when the deviation between the monitoring data and the predicted value exceeds the preset threshold, the model parameter update is automatically triggered, and the model's dynamic tracking capability of the pipeline aging process is maintained through incremental learning.

5. The device for monitoring aging of inner wall coating of seawater pipeline in nuclear power plant according to claim 4, characterized in that: The multi-dimensional data of array electrodes include current density distribution, potential gradient, and flow field parameters.

6. The device for monitoring aging of inner wall coating of seawater pipeline in nuclear power plant according to claim 1, characterized in that: The array electrode is a standard cylinder made of Q235 steel, which is consistent with the material of the pipeline.

7. The device for monitoring the aging of the inner wall coating of a seawater pipeline in a nuclear power plant according to claim 1, characterized in that: The reference electrode was a silver / silver chloride electrode.

8. The device for monitoring the aging of the inner wall coating of a seawater pipeline in a nuclear power plant according to claim 1, characterized in that: The auxiliary electrode is a metal oxide coated cylindrical electrode with a surface area significantly larger than that of the array electrode.

9. The device for monitoring the aging of the inner wall coating of a seawater pipeline in a nuclear power plant according to claim 1, characterized in that: The array electrodes are 5×5 arrays.

10. The device for monitoring the aging of the inner wall coating of a seawater pipeline in a nuclear power plant according to claim 1, characterized in that: The array electrodes are 10×10 arrays.

Images