Abrasion prediction and intelligent pipe blocking method and system for heat transfer pipe of steam generator

Abstract

The invention provides a steam generator heat transfer tube wear prediction and intelligent tube blocking method and system. The method comprises the steps that flow fields, vibration fields and wear databases of a heat transfer tube under different working conditions and positions are constructed through analogue simulation and real overhaul data of a nuclear power plant; constructing a training sample based on the constructed flow fields, vibration fields and wear databases of the heat transfer tube under different working conditions and positions; respectively training the flow field prediction model, the vibration field prediction model and the wear coefficient prediction model; during the unit overhaul period, operation data of a steam generator, position information of each heat transfer pipe and a supporting plate and corresponding abrasion data are obtained; respectively carrying out pretreatment; utilizing the trained flow field prediction model, vibration field prediction model and wear coefficient prediction model to predict the wear depth during the next overhaul; and judging whether the predicted wear depth during the next overhaul reaches or exceeds a set threshold value, and triggering and controlling the automatic pipe plugging operation when the wear depth reaches or exceeds the threshold value.

Description

Steam generator heat transfer tube wear prediction and intelligent tube blockage method and system Technical Field

[0001] This invention relates to a method for predicting heat transfer tube wear, specifically, to a method and system for predicting heat transfer tube wear in steam generators and for intelligent tube plugging. Background Technology

[0002] As a key piece of equipment connecting the primary and secondary loops of a nuclear power plant, the safety of the steam generator is of paramount importance. During unit overhauls, wear assessment of the steam generator heat transfer tubes typically relies on eddy current testing to obtain the comprehensive wear amount of each heat transfer tube at different support plate locations, and to predict the possible wear depth during the next overhaul. The most crucial and challenging task lies in accurately determining the wear coefficient.

[0003] Therefore, in the face of this situation, accurate prediction of the wear degree of the heat transfer tubes of the steam generator is of great importance to the safety and economy of nuclear power plants.

[0004] In existing technologies, the wear depth of the heat transfer tube at the time of the next major overhaul is predicted by assuming that the increase in wear volume during this major overhaul is equal to the increase in wear volume during the next major overhaul, or assuming that the increase in wear depth during this major overhaul is equal to the increase in wear depth during the next major overhaul. Based on these two calculations, the most conservative prediction result is selected, and the heat transfer tubes that reach or exceed the predetermined wear limit during the next major overhaul are plugged.

[0005] In existing technologies, a fretting wear test device for heat transfer tube support plates is built to simulate the fretting wear between the heat transfer tubes and the support plates inside a steam generator. The wear volume is directly measured, the wear coefficient is fitted, and then the results are input into classic wear models such as Archard for engineering calculations.

[0006] The methods listed above are typically replaced by more conservative tube-blocking schemes in actual nuclear power plant operation to ensure safety and stability during operation. However, the above empirical prediction methods may be too conservative for some heat transfer tubes, thereby reducing the heat exchange efficiency of the primary and secondary sides and causing unnecessary economic losses.

[0007] One method involves using equal depth and volume to predict the wear depth that the next major overhaul might reach. Based on these two calculations, the most conservative prediction result is selected. However, in reality, some heat transfer tubes that are deemed necessary to be plugged may actually still have a certain safety margin. Such plugging operations lead to a decrease in the heat exchange efficiency of the primary and secondary loops, causing economic losses to the nuclear power plant.

[0008] Relying on the completeness and accuracy of data from previous major overhauls, and with time-based predictions limited by the overhaul cycle, it is difficult to achieve true real-time online prediction.

[0009] A laboratory-scale fretting wear test apparatus for heat transfer tubes was constructed to simulate the internal operating conditions of a steam generator and directly measure the wear volume. However, the experimental scale and conditions could not fully reproduce the complex operating conditions of an actual steam generator, and the wear coefficient measured experimentally could not accurately reflect the wear situation in an actual steam generator.

[0010] The present invention provides a method and system for predicting wear and intelligent blockage of heat transfer tubes in steam generators. Instead of relying on a single "wear coefficient", it constructs an intelligent model that can predict wear-related loads, fretting behavior, or directly predict the amount of wear by learning historical operating data, vibration signals, and simulation results. Summary of the Invention

[0011] To address the shortcomings of existing technologies, the purpose of this invention is to provide a method and system for predicting wear and intelligently blocking heat transfer tubes in steam generators.

[0012] The present invention provides a method for predicting wear and intelligently blocking heat transfer tubes in a steam generator, comprising: Step S1: constructing a database of flow field, vibration field, and wear of heat transfer tubes under different operating conditions and locations through simulation and real overhaul data of nuclear power plants; Step S2: constructing training samples based on the constructed database of flow field, vibration field, and wear of heat transfer tubes under different operating conditions and locations; training the flow field prediction model, vibration field prediction model, and wear coefficient prediction model using the constructed training samples to obtain the trained flow field prediction model, vibration field prediction model, and wear coefficient prediction model; Step S3: during unit overhaul, acquiring the operating data of the steam generator, the position information of each heat transfer tube and support plate, and the corresponding wear data; and analyzing the acquired operating data and wear data of the steam generator. The wear data is preprocessed separately; Step S4: Based on the preprocessed operating data, the position information of each heat transfer pipe and support plate, and the corresponding preprocessed wear data, the wear depth at the next major overhaul is predicted using the trained flow field prediction model, vibration field prediction model, and wear coefficient prediction model; Step S5: It is determined whether the predicted wear depth at the next major overhaul reaches or exceeds the set threshold. When it reaches or exceeds the set threshold, the automated pipe blocking operation is triggered and controlled; The flow field prediction model is built based on DeepONet and is used to predict flow field parameters; The vibration field prediction model is built based on DeepONet and is used to predict vibration characteristics; The wear coefficient prediction model is built based on least squares inversion and regularized inversion and is used to predict the wear coefficient.

[0013] Preferably, step S1 includes: Step S1.1: Based on the geometric arrangement of the steam generator and the actual operating conditions, establish a numerical model of the flow field of the heat transfer tubes of the steam generator, obtain the local flow field parameters of each heat transfer tube and different support plate positions, and extract the flow field excitation force of each heat transfer tube; establish a dynamic model of the heat transfer tube support plate structure, and solve the vibration characteristics at each support plate position based on the flow field excitation force; classify the simulation results according to the labels "unit", "steam generator number", "heat transfer tube number", "support plate number" and "overhaul number"; Step S1.2: Obtain the heat transfer tube wear data and steam generator operation data of each overhaul, and perform data cleaning processing on the obtained heat transfer tube wear data and steam generator operation data of each overhaul to obtain the cleaned heat transfer tube wear data and steam generator operation data of each overhaul.

[0014] Preferably, step S2 includes: constructing training samples based on the constructed database of flow field, vibration field, and wear of the heat transfer tube under different operating conditions and locations; for the flow field prediction model, during training, a model including operational data is constructed based on the training samples. and heat transfer tube wear location parameters input vector ; Output local flow field prediction parameters of the heat transfer tube at different support plates For the vibration field prediction model, during training, local flow field prediction parameters are constructed based on training samples. and the structural parameters of the heat transfer tube itself The input vector is The vibration characteristics Vp of the heat transfer tube at different support plate positions are output; for the wear coefficient prediction model, during training, a model is constructed based on the training samples, including the heat transfer tube wear location parameter P, the wear depth h of the heat transfer tube detected during this overhaul, and the local flow field prediction parameters. The input is the equivalent full-power operating time; the output is the predicted value k of the wear coefficient.

[0015] Preferably, step S3 includes: during unit overhaul, acquiring the operating data of the steam generator, the position information of each heat transfer tube and support plate, and the corresponding raw eddy current detection data; wherein, the operating data of the steam generator includes: actual operating time and real-time operating power; converting the acquired actual operating time and real-time operating power into an equivalent full-power operating time; cleaning the raw eddy current detection data to meet preset requirements to obtain cleaned raw data; and standardizing the cleaned raw data to obtain processed wear data; wherein, the cleaning process includes: noise filtering and threshold comparison rejection.

[0016] Preferably, step S4 includes: based on the pre-treated operating data of the steam generator and heat transfer tube wear location parameters Construct the input vector The trained flow field prediction model is used to obtain the local flow field prediction parameters for the next major overhaul cycle at the current location. ; Local flow field prediction parameters for the next major overhaul cycle at the current location and the structural parameters of the current heat transfer tube itself. The vibration characteristics of the heat transfer tube in the next cycle are obtained using the trained vibration field prediction model. The wear depth of the heat transfer tube at this location after pretreatment. Local flow field prediction for the next major overhaul cycle Vibration characteristics of the heat transfer tube in the next cycle The equivalent full-power operating time of the next cycle is used as input to the trained wear coefficient prediction model to obtain the wear coefficient of the heat transfer tube at this location in the next wear cycle. Based on the wear coefficient of the heat transfer tube at this location in the next wear cycle. The predicted value of the additional wear depth at this location during the next major overhaul was calculated. The wear depth detected during this overhaul Add them together to get the predicted wear depth for the next major overhaul. .

[0017] Preferably, step S5 includes: determining whether the wear depth predicted for the next major overhaul reaches or exceeds a set threshold; when it reaches or exceeds the set threshold, triggering the positioning robot to move to the corresponding support plate position, and having the mounted pipe-blocking robot perform automated pipe-blocking operation; after the pipe-blocking is completed, acquiring the heat transfer pipe blockage information and transmitting the heat transfer pipe blockage information back to the flow field prediction model and the vibration field prediction model, so that the flow field prediction model and the vibration field prediction model can be automatically iteratively updated.

[0018] According to the present invention, a heat transfer tube wear prediction and intelligent tube blockage system for a steam generator includes: Module M1: constructing a database of flow field, vibration field, and wear of the heat transfer tube under different operating conditions and locations through simulation and real overhaul data of a nuclear power plant; Module M2: constructing training samples based on the constructed database of flow field, vibration field, and wear of the heat transfer tube under different operating conditions and locations; using the constructed training samples to train the flow field prediction model, vibration field prediction model, and wear coefficient prediction model respectively, to obtain the trained flow field prediction model, vibration field prediction model, and wear coefficient prediction model; Module M3: during unit overhaul, acquiring the operating data of the steam generator, the position information of each heat transfer tube and support plate, and the corresponding wear data; and analyzing the acquired operating data and wear data of the steam generator. The wear data is preprocessed separately; Module M4: Based on the preprocessed operating data, the position information of each heat transfer pipe and support plate, and the corresponding preprocessed wear data, it uses the trained flow field prediction model, vibration field prediction model, and wear coefficient prediction model to predict the wear depth during the next major overhaul; Module M5: Determines whether the predicted wear depth during the next major overhaul reaches or exceeds a set threshold. When it reaches or exceeds the set threshold, it triggers and controls automated pipe blocking operation; The flow field prediction model is built based on DeepONet and is used to predict flow field parameters; The vibration field prediction model is built based on DeepONet and is used to predict vibration characteristics; The wear coefficient prediction model is built based on least squares inversion and regularized inversion and is used to predict the wear coefficient.

[0019] Preferably, module M1 includes: Module M1.1: Based on the geometric arrangement of the steam generator and the actual operating conditions, establish a numerical model of the flow field of the heat transfer tubes of the steam generator, obtain the local flow field parameters of each heat transfer tube and different support plate positions, and extract the smooth excitation force of each heat transfer tube; establish a dynamic model of the heat transfer tube support plate structure, and solve the vibration characteristics at each support plate position based on the smooth excitation force; classify the simulation results according to the labels "unit", "steam generator number", "heat transfer tube number", "support plate number" and "overhaul number"; Module M1.2: acquire the heat transfer tube wear data and steam generator operation data of each overhaul, and perform data cleaning processing on the acquired heat transfer tube wear data and steam generator operation data of each overhaul to obtain the cleaned heat transfer tube wear data and steam generator operation data of each overhaul.

[0020] Preferably, module M2 includes: constructing training samples based on the constructed database of flow field, vibration field, and wear of the heat transfer tube under different operating conditions and locations; for the flow field prediction model, during training, constructing a model based on the training samples, including operational data... and heat transfer tube wear location parameters input vector ; Output local flow field prediction parameters of the heat transfer tube at different support plates For the vibration field prediction model, during training, local flow field prediction parameters are constructed based on training samples. and the structural parameters of the heat transfer tube itself The input vector is The vibration characteristics Vp of the heat transfer tube at different support plate positions are output; for the wear coefficient prediction model, during training, a model is constructed based on the training samples, including the heat transfer tube wear location parameter P, the wear depth h of the heat transfer tube detected during this overhaul, and the local flow field prediction parameters. The input is the equivalent full-power operating time; the output is the predicted value k of the wear coefficient.

[0021] Preferably, module M3 includes: acquiring operating data of the steam generator, position information of each heat transfer tube and support plate, and corresponding raw eddy current detection data during unit overhaul; wherein, the operating data of the steam generator includes: actual operating time and real-time operating power; equivalently converting the acquired actual operating time and real-time operating power to obtain equivalent full-power operating time; cleaning the raw eddy current detection data to meet preset requirements to obtain cleaned raw data; and standardizing the cleaned raw data to obtain processed wear data; wherein, the cleaning process includes: noise filtering and threshold comparison rejection; module M4 includes: based on the pre-processed operating data of the steam generator and heat transfer tube wear location parameters Construct the input vector The trained flow field prediction model is used to obtain the local flow field prediction parameters for the next major overhaul cycle at the current location. ; Local flow field prediction parameters for the next major overhaul cycle at the current location and the structural parameters of the current heat transfer tube itself. The vibration characteristics of the heat transfer tube in the next cycle are obtained using the trained vibration field prediction model. The wear depth of the heat transfer tube at this location after pretreatment. Local flow field prediction for the next major overhaul cycle Vibration characteristics of the heat transfer tube in the next cycle The equivalent full-power operating time of the next cycle is used as input to the trained wear coefficient prediction model to obtain the wear coefficient of the heat transfer tube at this location in the next wear cycle. Based on the wear coefficient of the heat transfer tube at this location in the next wear cycle. The predicted value of the additional wear depth at this location during the next major overhaul was calculated. The wear depth detected during this overhaul Add them together to get the predicted wear depth for the next major overhaul. The module M5 includes: determining whether the wear depth predicted for the next major overhaul reaches or exceeds a set threshold; when it reaches or exceeds the set threshold, triggering the positioning robot to move to the corresponding support plate position, where the onboard pipe-blocking robot performs automated pipe-blocking operation; after the pipe-blocking is completed, acquiring the heat transfer pipe blockage information and transmitting the heat transfer pipe blockage information back to the flow field prediction model and the vibration field prediction model, so that the flow field prediction model and the vibration field prediction model can be automatically iteratively updated.

[0022] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention deeply integrates the heat transfer tube positioning robot, eddy current detection equipment and AI prediction model to realize full automation of the entire process from heat transfer tube positioning, wear detection, data uploading, intelligent analysis to automatic generation of tube blocking instructions and execution of tube blocking operations; the whole process does not require manual judgment or manual intervention, shortens the overhaul cycle and reduces the number of times staff enter the radiation area, and significantly improves the safety and execution efficiency of nuclear power plant maintenance operations.

[0023] 2. Key physical quantities inside the steam generator cannot be directly obtained through sensors, and the amount of actual wear data is limited. Traditional methods cannot obtain the main controlling factors of wear. This invention constructs a high-quality training database through flow field simulation and structural vibration simulation, and introduces AI models such as DeepONet to achieve "virtual sensing" of the flow field and vibration field. It can predict unmeasurable key loads without physical sensors, breaking through the limitations of traditional experimental and detection methods. It transforms unmeasurable physical quantities into input predictable variables, thereby significantly improving the accuracy and generalization ability of the wear prediction model. 3. Traditional judgment methods only extrapolate from the current wear value based on experience, which often leads to conservative or misjudgment of the pipe blocking strategy. This invention uses an AI model to jointly learn the current wear state, the predicted local flow field, the predicted vibration field, and the operating parameters to directly give the wear increment for the next cycle, realizing the prediction of future wear. It can identify high-risk heat transfer tubes in advance, improving safety margins; avoid excessive pipe blocking caused by conservative strategies, improving heat exchange efficiency and economy; transform traditional offline static detection into dynamic monitoring, improving the foresight of decision-making; and establish a mathematical correlation between wear degree, pipe blocking decision, and operating data. This has enabled the scientific and standardized decision-making process for heat transfer tube maintenance.

[0024] 4. After completing wear prediction, this invention automatically generates a tube-blocking strategy, and the tube-blocking operation is executed by a robot or automated device, avoiding inconsistencies in maintenance standards and potential misoperations caused by differences in human judgment. The tube-blocking execution robot of this invention can automatically locate the target heat transfer tube position and directly execute the tube-blocking action according to the tube-blocking decision, and send the tube-blocking result back to the database for model iteration. Through data-driven objective decision-making, this invention can effectively reduce the economic losses caused by overly conservative tube-blocking, while ensuring that heat transfer tubes with potential failure risks are dealt with in a timely manner, achieving dual optimization of safety and economy. 5. The wear prediction and automatic tube-blocking method for steam generator heat transfer tubes based on AI and data-driven principles proposed in this invention realizes a complete intelligent closed loop from heat transfer tube detection to wear prediction and finally automatic tube-blocking execution. Attached Figure Description

[0025] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 is a flowchart of a method for predicting wear of heat transfer tubes in a steam generator and for intelligent tube blocking.

[0026] Figure 2 shows the simulation results of the heat transfer tubes of the steam generator.

[0027] Figure 3 is a flowchart for determining pipe blockage. Detailed Implementation

[0028] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0029] Example 1 presents a method and system for predicting and intelligently blocking heat transfer tube wear in steam generators according to the present invention. Addressing the wear problem of heat transfer tubes in actual steam generators of nuclear power plants, this invention combines AI and data-driven approaches to achieve precise prediction of the wear location and depth of the heat transfer tubes. Based on this, it further completes wear trend assessment and automatic generation of blocking strategies. After acquiring on-site inspection data, this invention utilizes a simulation-enhanced flow field, vibration, and wear chain inference model to predict potential wear risks in the next major overhaul cycle, thus providing a quantitative basis for blocking decisions. Simultaneously, by combining robots to achieve automatic positioning and blocking execution, the entire steam generator maintenance process is transformed from traditional manual experience-based judgment into an intelligent closed loop of "perception-prediction-execution," significantly improving the safety, accuracy, and efficiency of steam generator heat transfer tube maintenance. The overall process is shown in Figure 1.

[0030] The method for predicting heat transfer tube wear and intelligently preventing tube blockage in steam generators includes: The first stage: database establishment and model training, which is an offline preparation stage. This stage is fundamental for online heat transfer tube wear prediction. A database of flow fields, vibration fields, and wear patterns of the heat transfer tubes under different operating conditions and locations is built using simulation and real nuclear power plant overhaul data. Intelligent methods such as DeepONet are used to learn the mapping relationships between these fields for online prediction in the second stage. After each overhaul, the tube blockage information is integrated to update the database model for the next prediction.

[0031] Specifically, the first stage includes: Step 1: Establishing a simulation database; As shown in Figure 2, based on the geometric layout of the steam generator and the actual operating conditions, a numerical model of the flow field of the heat transfer tube bundle of the steam generator is established to obtain the local flow field parameters of each heat transfer tube and different support plate positions, and the smooth excitation force of each heat transfer tube is extracted; A dynamic model of the heat transfer tube support plate structure is established, and the vibration characteristics at each support plate position are solved based on the smooth excitation force, i.e., the vibration displacement-time curve at each position; The simulation results are classified according to the tags "Unit", "Steam Generator Number", "Heat Transfer Tube Number", "Support Plate Number", and "Overhaul Number".

[0032] Step 2: Collect historical overhaul inspection and operation data; organize the wear data of heat transfer tubes and the operation data of steam generator from previous overhauls, including outliers and values ​​with significantly large errors in the cleaning history data.

[0033] Step 3: Construct training samples and labels; for each support plate position of each heat transfer tube during each overhaul, extract the operational data for that position from the simulation database. and heat transfer tube wear location parameters Construct flow field feature vectors Extract the corresponding vibration characteristics from the simulation database. The wear increment for each major overhaul was calculated based on the wear levels of the two overhauls. The different positions of the heat transfer tubes and the bend radius R of the heat transfer tubes are used as input parameters.

[0034] Step 4: AI Model Training; Three prediction sub-models need to be trained: a flow field prediction model, a vibration field prediction model, and a wear coefficient prediction model. These are trained independently using supervised learning, based on a simulation database and historical overhaul inspection data. The DeepONet deep operator network is used to simulate the local flow and vibration fields.

[0035] For the flow field prediction model, the construction includes operational data. and heat transfer tube wear location parameters input vector ; Output the predicted vector of the local flow field of the heat transfer tube at different support plates The training samples are derived from real flow field results in a previously established simulation library.

[0036] For vibration field prediction models, during the initial model training, the input of the vibration field comes from the simulated values ​​of the flow field from the previous training. And the structural parameters of the heat transfer tube itself, such as the radius of the bend section. etc., construct the input vector as Vibration characteristics of the output heat transfer tube at different support plate positions The training samples are derived from real vibration field results in a previously established simulation library.

[0037] For the wear coefficient prediction model, predict the wear coefficient for the next cycle. This is the core of the entire AI model. During training, the input includes parameters such as the wear location of the heat transfer tubes. During this major overhaul, the depth of wear on the heat transfer tubes was inspected. Flow field characteristics The output is a predicted value of the wear coefficient. The training samples are derived from the vibration characteristics of different heat transfer tube locations in the previously established simulation library, as well as wear data generated during previous overhauls. Training is conducted using data from multiple historical overhaul intervals, with a subset of overhaul cycles used as the validation set. This embodiment supplements the scarcity of real data using a simulation database and predicts the flow and vibration fields using the DeepONet deep operator network, solving the problems of slow offline simulation speed and inability to meet the timeliness requirements of nuclear power plant overhauls; it also overcomes the limitation of unmeasurable loads, achieving intelligent prediction of the entire flow-vibration-wear chain based on simulation enhancement. In this embodiment, the wear coefficient is inverted using the eddy current detection results from multiple overhaul cycles; the defect depth of the same heat transfer tube at the same support plate location during each overhaul is considered as the observed value of wear accumulation over time, thereby inferring the wear coefficient of the wear model.

[0038] The second stage: online prediction and pipe plugging process during nuclear power unit overhaul. This stage is the online prediction stage. This stage mainly realizes the automatic positioning of robots and the perception of heat transfer tube wear data, combined with AI-driven models to predict the wear degree of heat transfer tubes in the next overhaul cycle, as well as intelligent pipe plugging decision-making and execution. This is the core application stage of the invention.

[0039] Step 5: Positioning Robot Inspection and Data Acquisition; During the unit overhaul, a heat transfer tube positioning robot is sent into the steam generator. The robot positions the heat transfer tubes, and the eddy current detection robot inserts the ECT probe into the designated heat transfer tube through a mechanical drive mechanism. It scans the positions of multiple support plates one by one, automatically detecting the original ECT data of wear on each heat transfer tube. Through the positioning robot and the eddy current detection system, the position number of each heat transfer tube and support plate, as well as the wear data at each corresponding number, can be obtained, realizing automated perception of heat transfer tube wear information.

[0040] Step Six: Data Preprocessing and Feature Construction; The ECT data obtained in Step Five is subjected to noise filtering and threshold judgment, and then integrated to output a standardized heat transfer tube wear depth. Based on the numbering of different wear data in step five, the corresponding flow field and vibration field feature data are matched from the AI ​​training model in step four, and the historical overhaul data records with corresponding labels are matched from the database.

[0041] Step 7: AI-driven heat transfer tube wear prediction; In step 7, the AI ​​model is automatically invoked to predict the flow field, vibration field, and other information at different support plate positions of each heat transfer tube in the next overhaul cycle.

[0042] First, obtain the wear depth of the heat transfer tubes at different support plate locations obtained in step six; then, call the "flow field prediction model" from step four and apply the running data. and heat transfer tube wear location parameters Constructing input vectors The local flow field prediction parameters for the next major overhaul cycle at this location are obtained through model prediction. Using the "vibration field prediction model", the local flow field is predicted. And the structural parameters of the heat transfer tube itself, such as the radius of the bend section. The vibration characteristics of the heat transfer tube in the next cycle are obtained. Then, using the "wear coefficient prediction model," the wear depth of the heat transfer tube at this location is calculated. Local flow field prediction for the next major overhaul cycle Vibration characteristics of the heat transfer tube in the next cycle As input, the wear coefficient of the heat transfer tube at this location in the next wear cycle is obtained. Finally, the predicted value of the increased wear depth at this location during the next major overhaul was obtained through calculation. The wear depth detected during this overhaul Add them together to get the predicted wear depth for the next major overhaul. This embodiment uses an AI model to intelligently predict the wear amount during the next major overhaul, refining the wear prediction results, rather than relying solely on experience with equal depth and volume based on current wear data; Step 8: Automatic pipe blockage decision; As shown in Figure 3, based on the predicted wear depth for the next major overhaul obtained in Step 7. Determine the predicted values ​​for each heat transfer tube. If the set 40% threshold is reached or exceeded, a tube blocking command will be issued for heat transfer tubes that reach or exceed the 40% threshold.

[0043] Step Nine: Automatic Pipe Blocking Execution; For the heat transfer tubes that require pipe blocking operation as determined in Step Eight, the designated positioning robot moves to the corresponding support plate position, and the onboard pipe blocking robot performs the automated pipe blocking operation.

[0044] Step 10: Record pipe blockage information and update the model; After the pipe blockage operation is completed, record the information of each blocked heat transfer tube, record the pipe blockage information and send it back to Step 4, and update the flow field and vibration field prediction models.

[0045] This embodiment automatically processes the data after eddy current detection and combines the predicted flow field and vibration field information with the results of previous overhauls. It calls the AI-trained model to intelligently predict the wear amount of the next overhaul cycle, skipping manual judgment to directly determine whether the predicted wear amount requires pipe plugging operation, and outputs pipe plugging command for automatic execution, reducing the difference in manual judgment and taking into account both safety and economy.

[0046] The present invention also provides a system for predicting and intelligently blocking heat transfer tube wear in a steam generator. The system can be implemented by executing the process steps of the method for predicting and intelligently blocking heat transfer tube wear in a steam generator. That is, those skilled in the art can understand the method for predicting and intelligently blocking heat transfer tube wear in a steam generator as a preferred embodiment of the system for predicting and intelligently blocking heat transfer tube wear in a steam generator.

[0047] Those skilled in the art will understand that, in addition to implementing the system, apparatus, and their modules provided by this invention in purely computer-readable program code, the same program can be implemented in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, the system, apparatus, and their modules provided by this invention can be considered a hardware component, and the modules included therein for implementing various programs can also be considered structures within the hardware component; alternatively, modules for implementing various functions can be considered both software programs implementing the method and structures within the hardware component.

[0048] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. A method for predicting wear and intelligently blocking heat transfer tubes in a steam generator, characterized in that, The process includes steps S1 and S2. Constructing a database of flow field, vibration field, and wear data for heat transfer tubes under different operating conditions and locations using simulation data and real nuclear power plant overhaul data. Step S2. Building training samples based on the constructed database of flow field, vibration field, and wear data for heat transfer tubes under different operating conditions and locations. Using these training samples, the flow field prediction model, vibration field prediction model, and wear coefficient prediction model are trained to obtain the trained models. Step S3. During unit overhaul, acquiring the operating data of the steam generator, the location information of each heat transfer tube and support plate, and the corresponding wear data. Preprocessing the acquired steam generator operating data and wear data. Step S4. Based on the preprocessed operating data, the position information of each heat transfer tube and support plate, and the corresponding preprocessed wear data, the wear depth at the next major overhaul is predicted using the trained flow field prediction model, vibration field prediction model, and wear coefficient prediction model. Step S5: Determine whether the predicted wear depth at the next major overhaul reaches or exceeds a set threshold. If it reaches or exceeds the set threshold, an automated tube plugging operation is triggered and controlled. The flow field prediction model is built based on DeepONet and is used to predict flow field parameters. The vibration field prediction model is built based on DeepONet and is used to predict vibration characteristics. The wear coefficient prediction model is built based on least squares inversion and regularized inversion and is used to predict the wear coefficient.

2. The method for predicting wear and intelligent blockage of heat transfer tubes in a steam generator according to claim 1, characterized in that, Step S1 includes: Step S1.1: Based on the geometric layout of the steam generator and the actual operating conditions, establish a numerical model of the flow field of the heat transfer tubes of the steam generator, obtain the local flow field parameters of each heat transfer tube and different support plate positions, and extract the smooth excitation force of each heat transfer tube; establish a dynamic model of the heat transfer tube support plate structure, and solve the vibration characteristics at each support plate position based on the smooth excitation force; classify the simulation results according to the labels "unit", "steam generator number", "heat transfer tube number", "support plate number" and "overhaul number"; Step S1.2: Obtain the heat transfer tube wear data and steam generator operation data of each overhaul, and perform data cleaning processing on the obtained heat transfer tube wear data and steam generator operation data of each overhaul to obtain the cleaned heat transfer tube wear data and steam generator operation data of each overhaul.

3. The method for predicting wear and intelligent blockage of heat transfer tubes in a steam generator according to claim 1, characterized in that, Step S2 includes: constructing training samples based on the constructed database of flow field, vibration field, and wear of the heat transfer tube under different operating conditions and locations; for the flow field prediction model, during training, constructing a database including operational data based on the training samples. and heat transfer tube wear location parameters input vector ; Output local flow field prediction parameters of the heat transfer tube at different support plates For the vibration field prediction model, during training, local flow field prediction parameters are constructed based on training samples. and the structural parameters of the heat transfer tube itself The input vector is The vibration characteristics Vp of the heat transfer tube at different support plate positions are output; for the wear coefficient prediction model, during training, a model is constructed based on the training samples, including the heat transfer tube wear location parameter P, the wear depth h of the heat transfer tube detected during this overhaul, and the local flow field prediction parameters. The input is the equivalent full-power operating time; the output is the predicted value k of the wear coefficient.

4. The method for predicting wear and intelligent blockage of heat transfer tubes in a steam generator according to claim 1, characterized in that, Step S3 includes: during unit overhaul, acquiring the operating data of the steam generator, the position information of each heat transfer tube and support plate, and the corresponding raw eddy current detection data; wherein, the operating data of the steam generator includes: actual operating time and real-time operating power; the acquired actual operating time and real-time operating power are equivalently converted to obtain the equivalent full-power operating time; the raw eddy current detection data is cleaned to meet preset requirements to obtain cleaned raw data; the cleaned raw data is standardized to obtain processed wear data; wherein, the cleaning process includes: noise filtering and threshold comparison rejection.

5. The method for predicting wear and intelligent blockage of heat transfer tubes in a steam generator according to claim 1, characterized in that, Step S4 includes: based on the pre-processed operating data of the steam generator and heat transfer tube wear location parameters Construct the input vector The trained flow field prediction model is used to obtain the local flow field prediction parameters for the next major overhaul cycle at the current location. ; Local flow field prediction parameters for the next major overhaul cycle at the current location and the structural parameters of the current heat transfer tube itself. The vibration characteristics of the heat transfer tube in the next cycle are obtained using the trained vibration field prediction model. The wear depth of the heat transfer tube at this location after pretreatment. Local flow field prediction for the next major overhaul cycle Vibration characteristics of the heat transfer tube in the next cycle The equivalent full-power operating time of the next cycle is used as input to the trained wear coefficient prediction model to obtain the wear coefficient of the heat transfer tube at this location in the next wear cycle. Based on the wear coefficient of the heat transfer tube at this location in the next wear cycle. The predicted value of the additional wear depth at this location during the next major overhaul was calculated. The wear depth detected during this overhaul Add them together to get the predicted wear depth for the next major overhaul. 。 6. The method for predicting wear and intelligent blockage of heat transfer tubes in a steam generator according to claim 1, characterized in that, Step S5 includes: determining whether the wear depth predicted for the next major overhaul has reached or exceeded a set threshold; when it has reached or exceeded the set threshold, triggering the positioning robot to move to the corresponding support plate position, and having the onboard pipe-blocking robot perform automated pipe-blocking operation; after the pipe-blocking is completed, obtaining the heat transfer pipe blockage information and transmitting the heat transfer pipe blockage information back to the flow field prediction model and the vibration field prediction model, so that the flow field prediction model and the vibration field prediction model can be automatically iteratively updated.

7. A system for predicting wear and intelligent blockage of heat transfer tubes in a steam generator, characterized in that, The system includes: Module M1: Constructing a database of flow field, vibration field, and wear data for heat transfer tubes under different operating conditions and locations using simulation and real nuclear power plant overhaul data; Module M2: Building training samples based on the constructed database of flow field, vibration field, and wear data for heat transfer tubes under different operating conditions and locations; using the constructed training samples to train the flow field prediction model, vibration field prediction model, and wear coefficient prediction model respectively, obtaining the trained flow field prediction model, vibration field prediction model, and wear coefficient prediction model; Module M3: Acquiring operating data of the steam generator, location information of each heat transfer tube and support plate, and corresponding wear data during unit overhaul; preprocessing the acquired operating data and wear data of the steam generator respectively; Module M4... Based on preprocessed operating data, the position information of each heat transfer pipe and support plate, and the corresponding preprocessed wear data, the wear depth at the next major overhaul is predicted using trained flow field prediction models, vibration field prediction models, and wear coefficient prediction models. Module M5 determines whether the predicted wear depth at the next major overhaul reaches or exceeds a set threshold. If it does, it triggers and controls automated pipe plugging operations. The flow field prediction model is based on DeepONet and is used to predict flow field parameters. The vibration field prediction model is based on DeepONet and is used to predict vibration characteristics. The wear coefficient prediction model is based on least squares inversion and regularized inversion and is used to predict the wear coefficient.

8. The steam generator heat transfer tube wear prediction and intelligent tube blockage system according to claim 7, characterized in that, Module M1 includes: Module M1.1: Based on the geometric layout of the steam generator and actual operating conditions, establish a numerical model of the flow field of the heat transfer tubes of the steam generator, obtain the local flow field parameters of each heat transfer tube and different support plate positions, and extract the smooth excitation force of each heat transfer tube; establish a dynamic model of the heat transfer tube support plate structure, and solve the vibration characteristics at each support plate position based on the smooth excitation force; classify the simulation results according to the labels "unit", "steam generator number", "heat transfer tube number", "support plate number" and "overhaul number"; Module M1.2: acquire the heat transfer tube wear data and steam generator operating data of each overhaul, and perform data cleaning processing on the acquired heat transfer tube wear data and steam generator operating data of each overhaul to obtain the cleaned heat transfer tube wear data and steam generator operating data of each overhaul.

9. The steam generator heat transfer tube wear prediction and intelligent tube blockage system according to claim 7, characterized in that, Module M2 includes: constructing training samples based on the constructed database of flow field, vibration field, and wear of the heat transfer tube under different operating conditions and locations; for the flow field prediction model, during training, constructing a database based on the training samples, including operational data. and heat transfer tube wear location parameters input vector ; Output local flow field prediction parameters of the heat transfer tube at different support plates For the vibration field prediction model, during training, local flow field prediction parameters are constructed based on training samples. and the structural parameters of the heat transfer tube itself The input vector is The vibration characteristics Vp of the heat transfer tube at different support plate positions are output; for the wear coefficient prediction model, during training, a model is constructed based on the training samples, including the heat transfer tube wear location parameter P, the wear depth h of the heat transfer tube detected during this overhaul, and the local flow field prediction parameters. The input is the equivalent full-power operating time; the output is the predicted value k of the wear coefficient.

10. The steam generator heat transfer tube wear prediction and intelligent tube blockage system according to claim 7, characterized in that, Module M3 includes: acquiring operating data of the steam generator, position information of each heat transfer tube and support plate, and corresponding raw eddy current detection data during unit overhaul; wherein, the operating data of the steam generator includes: actual operating time and real-time operating power; equivalently converting the acquired actual operating time and real-time operating power to obtain equivalent full-power operating time; cleaning the raw eddy current detection data to meet preset requirements to obtain cleaned raw data; and standardizing the cleaned raw data to obtain processed wear data; wherein, the cleaning process includes: noise filtering and threshold comparison rejection; Module M4 includes: based on the pre-processed operating data of the steam generator and heat transfer tube wear location parameters Construct the input vector The trained flow field prediction model is used to obtain the local flow field prediction parameters for the next major overhaul cycle at the current location. ; Local flow field prediction parameters for the next major overhaul cycle at the current location and the structural parameters of the current heat transfer tube itself. The vibration characteristics of the heat transfer tube in the next cycle are obtained using the trained vibration field prediction model. The wear depth of the heat transfer tube at this location after pretreatment. Local flow field prediction for the next major overhaul cycle Vibration characteristics of the heat transfer tube in the next cycle The equivalent full-power operating time of the next cycle is used as input to the trained wear coefficient prediction model to obtain the wear coefficient of the heat transfer tube at this location in the next wear cycle. Based on the wear coefficient of the heat transfer tube at this location in the next wear cycle. The predicted value of the additional wear depth at this location during the next major overhaul was calculated. The wear depth detected during this overhaul Add them together to get the predicted wear depth for the next major overhaul. The module M5 includes: determining whether the wear depth predicted for the next major overhaul reaches or exceeds a set threshold; when it reaches or exceeds the set threshold, triggering the positioning robot to move to the corresponding support plate position, where the onboard pipe-blocking robot performs automated pipe-blocking operation; after the pipe-blocking is completed, acquiring the heat transfer pipe blockage information and transmitting the heat transfer pipe blockage information back to the flow field prediction model and the vibration field prediction model, so that the flow field prediction model and the vibration field prediction model can be automatically iteratively updated.

Images