Decision-making method of steam generator and training method of intelligent decision-making model

Abstract

The invention provides a decision-making method of a steam generator and a training method of an intelligent decision-making model. The method comprises the steps of obtaining key operation parameters of a nuclear power plant; based on the key operation parameters, predicting thermal hydraulic parameters of a steam generator of the nuclear power plant in a future time period; inputting the key operation parameters and the thermal hydraulic parameters into an intelligent decision-making model, making a decision through a pre-learned decision-making table in the intelligent decision-making model, and determining an optimal adjustment action of the steam generator; wherein the decision table comprises a mapping relation between different decision actions and reward values when the nuclear power plant and the steam generator are in different working states, and the working states are determined according to key operation parameters of the nuclear power plant and thermal hydraulic parameters of the steam generator; the reward value is determined according to the safety degrees corresponding to the liquid level height and the pressure of the steam generator. The embodiment of the invention can improve the decision reliability of the steam generator.

Description

Technical Field

[0001] This application relates to the field of nuclear power technology, and in particular to a decision-making method for steam generators and a training method for intelligent decision-making models. Background Technology

[0002] As a key connecting device in the primary and secondary loops of a pressurized water reactor nuclear power plant, the steam generator plays a crucial role in transferring heat from the reactor core and generating steam to drive the turbine generator unit. The stable control of its water level and pressure directly determines the safety and reliability of the unit. Excessively high water levels can lead to water carryover in the steam, damaging the turbine; excessively low water levels may cause thermal fatigue rupture of the heat transfer tubes. Abnormal pressure can affect the thermodynamic cycle efficiency or induce reactive disturbances. Therefore, the decision-making and control technology of the steam generator has always been a core research direction in the operation and management of nuclear power plants.

[0003] Existing decision-making methods for steam generators primarily rely on digital instrumentation and control systems. These systems collect operating parameters such as temperature, pressure, and flow rate through sensors and combine them with preset control logic for decision-making and control. However, the liquid level and pressure of a steam generator are strongly coupled by multiple parameters in the primary / secondary loops, exhibiting significant nonlinearity and time-varying characteristics. These existing methods, which mostly rely on linear fixed parameters for analysis, are difficult to adapt to the complex operating conditions of nonlinear and time-varying steam generators, resulting in large deviations in the output decision actions and thus low reliability of steam generator decision-making. Summary of the Invention

[0004] This application provides a decision-making method for steam generators and a training method for intelligent decision-making models, which can improve the decision-making reliability of steam generators.

[0005] In a first aspect, embodiments of this application provide a decision-making method for a steam generator, the method comprising: Obtain key operating parameters of nuclear power plants; Based on key operating parameters, predict the thermal-hydraulic parameters of the steam generator in a nuclear power plant over a future period; Key operating parameters and thermal-hydraulic parameters are input into the intelligent decision-making model, and decisions are made through the pre-learned decision table in the intelligent decision-making model to determine the optimal adjustment action of the steam generator; The decision table includes the mapping relationship between different decision actions and reward values ​​for nuclear power plants and steam generators under different operating conditions. The operating conditions are determined based on the key operating parameters of the nuclear power plant and the thermal-hydraulic parameters of the steam generator, and the reward values ​​are determined based on the safety levels corresponding to the liquid level height and pressure of the steam generator.

[0006] Secondly, this application provides a method for training an intelligent decision-making model, the method comprising: Obtain a training sample set, which includes key operating parameters, thermal-hydraulic parameters, and adjustment action labels of the nuclear power plant. The thermal-hydraulic parameters are used to predict the thermal-hydraulic parameters of the nuclear power plant's steam generator in the future time period based on the key operating parameters. The key operating parameters and thermal-hydraulic parameters of the sample are input into the intelligent decision-making model to be trained. The decision is made through the decision table in the intelligent decision-making model to determine the predicted adjustment action of the steam generator. Calculate the loss value based on the predicted adjustment actions and the corresponding adjustment action labels; If the loss value does not meet the preset conditions, the decision table is iteratively updated until the loss value meets the preset conditions, and a trained intelligent decision model is obtained. The decision table includes the mapping relationship between different decision actions and reward values ​​for nuclear power plants and steam generators under different operating conditions. The operating conditions are determined based on the key operating parameters of the nuclear power plant and the thermal-hydraulic parameters of the steam generator, and the reward values ​​are determined based on the safety levels corresponding to the liquid level height and pressure of the steam generator.

[0007] Thirdly, this application provides a decision-making device for a steam generator, the device comprising: The first acquisition module is used to acquire key operating parameters of the nuclear power plant; The prediction module is used to predict the thermal-hydraulic parameters of the steam generator of the nuclear power plant in the future time period based on the key operating parameters. The first decision module is used to input the key operating parameters and the thermal-hydraulic parameters into the intelligent decision model, and make decisions through the pre-learned decision table in the intelligent decision model to determine the optimal adjustment action of the steam generator; The decision table includes a mapping relationship between different decision actions and reward values ​​for the nuclear power plant and the steam generator under different operating conditions. The operating conditions are determined based on the key operating parameters of the nuclear power plant and the thermal-hydraulic parameters of the steam generator. The reward values ​​are determined based on the safety levels corresponding to the liquid level and pressure of the steam generator.

[0008] Fourthly, this application provides a training device for an intelligent decision-making model, the device comprising: The second acquisition module is used to acquire a training sample set, which includes sample key operating parameters, sample thermal-hydraulic parameters and adjustment action labels of the nuclear power plant; the sample thermal-hydraulic parameters are the thermal-hydraulic parameters of the steam generator of the nuclear power plant in the future time period based on the sample key operating parameters. The second decision module is used to input the key operating parameters and thermal-hydraulic parameters of the sample into the intelligent decision model to be trained, and to make a decision through the decision table in the intelligent decision model to determine the predicted adjustment action of the steam generator. The calculation module is used to calculate the loss value based on the predicted adjustment action and the corresponding adjustment action label; The training module is used to iteratively update the decision table when the loss value does not meet the preset conditions, until the loss value meets the preset conditions, so as to obtain a trained intelligent decision model. The decision table includes a mapping relationship between different decision actions and reward values ​​for the nuclear power plant and the steam generator under different operating conditions. The operating conditions are determined based on the key operating parameters of the nuclear power plant and the thermal-hydraulic parameters of the steam generator. The reward values ​​are determined based on the safety levels corresponding to the liquid level and pressure of the steam generator.

[0009] Fifthly, embodiments of this application provide an electronic device, which includes: a processor and a memory storing computer program instructions; When the processor executes computer program instructions, it implements a decision-making method for a steam generator as described in any embodiment of the first aspect, and a training method for an intelligent decision-making model as described in any embodiment of the second aspect.

[0010] In a sixth aspect, embodiments of this application provide a computer storage medium storing computer program instructions. When the computer program instructions are executed by a processor, they implement the decision-making method for a steam generator as described in any embodiment of the first aspect, and the training method for an intelligent decision-making model as described in any embodiment of the second aspect.

[0011] In a seventh aspect, embodiments of this application provide a computer program product in which instructions, when executed by a processor of an electronic device, cause the electronic device to execute a decision-making method for a steam generator as described in any embodiment of the first aspect, and a training method for an intelligent decision-making model as described in any embodiment of the second aspect.

[0012] In the decision-making method and intelligent decision-making model training method for a steam generator provided in this application embodiment, key operating parameters of the nuclear power plant are acquired to ensure accurate perception of the current operating state of the steam generator. Based on this, the thermal-hydraulic parameters of the steam generator for future periods are predicted based on the key operating parameters. This prediction process allows decision-making to move beyond passive responses to the current operating conditions and instead possess a forward-looking understanding of future operating conditions, reducing decision-making biases caused by the inability to adapt to dynamic changes in operating conditions. Finally, the key operating parameters and the predicted thermal-hydraulic parameters are jointly input into the intelligent decision-making model, and the optimal adjustment action is determined through a pre-learned decision table. The decision table constructs a "decision action-reward value" mapping relationship based on different operating states of the nuclear power plant and the steam generator, and the reward value is determined by the safety of the liquid level and pressure, ensuring that the decision action always prioritizes the safe and stable operation of the steam generator. This process, through multi-parameter fusion to define operating conditions and safety-oriented reward value constraints on the selection of adjustment actions, enables the intelligent decision-making model to adapt to different complex operating conditions, resulting in adjustment actions that better match actual operating requirements, significantly reducing decision-making biases, and thus improving the reliability of steam generator decision-making. Attached Figure Description

[0013] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a flowchart illustrating the decision-making method for a steam generator provided in an embodiment of this application; Figure 2 This is a schematic diagram of the structure of the steam generator provided in the embodiments of this application; Figure 3 This is a schematic diagram of the structure of the digital twin model provided in the embodiments of this application; Figure 4 This is a flowchart illustrating the training method of the intelligent decision-making model provided in the embodiments of this application; Figure 5 This is a schematic diagram of the structure of a decision-making device for a steam generator provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of a training device for an intelligent decision-making model provided in an embodiment of this application; Figure 7 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0015] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0016] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.

[0017] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the term "comprising" or any other variations thereof is intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0018] A pressurized water reactor (PWR) nuclear power plant mainly consists of a reactor, a primary loop system, and a secondary loop system. During operation, the coolant in the primary loop flows through the reactor core, carrying away the large amount of heat released by the core fission reaction, and then flows to the steam generator. The steam generator is a key piece of equipment in the PWR nuclear power steam supply system, connecting the primary and secondary loop systems. It plays a crucial role in heat and mass transfer, transferring heat from the reactor core to the secondary loop working fluid through heat transfer tubes to generate steam.

[0019] As a crucial link between the primary and secondary loops, the steam generator in a nuclear power plant transfers heat generated by the reactor to the secondary working fluid through thousands of heat transfer tubes, thereby producing saturated steam to drive the turbine generator unit. This process involves extremely complex multiphase flow, heat transfer, and thermal hydraulic phenomena. The complexity, strong nonlinearity, and high coupling of its dynamic characteristics make the control of the steam generator, especially the control of water level and pressure, a critical factor determining the safe, reliable, and economical operation of the entire nuclear power plant unit.

[0020] The working fluid flow and heat transfer process on the secondary side of the steam generator can be clearly divided into three distinct stages: in the descending channel, the underheated water from the feedwater system undergoes single-phase heat transfer; upon entering the tube bundle region, the working fluid begins subcooled boiling heat transfer; and finally, in the ascending channel, it develops into intense saturated two-phase boiling heat transfer. Correspondingly, the flow characteristics also exhibit significant differences: the descending channel exhibits relatively stable single-phase flow, while the steam-water two-phase mixture in the ascending channel is prone to inherent density wave oscillations. These oscillations manifest as periodic "contraction-expansion" of the working fluid mass flow rate and cavitation fraction, severely restricting the stability of the natural circulation and posing inherent challenges to the design of the control system.

[0021] Of all the dynamic characteristics, the most deceptive and difficult to control is the "false water level" phenomenon inherent in the system's non-minimum phase behavior. The physical essence of this phenomenon stems from the effect of sudden pressure changes on the density of the steam-water mixture. Specifically, when the load increases, causing a sudden surge in steam flow, the internal pressure of the steam generator drops instantaneously, leading to a decrease in the saturation temperature. This change causes the previously saturated water to instantly become superheated, triggering a rapid intensification of the boiling process and generating a large number of additional steam bubbles. These bubbles significantly increase the total volume of the steam-water mixture as they rise, causing its apparent density to drop rapidly. Water level gauges operating on the differential pressure principle sense the density of the mixture, not the actual water level, and therefore output a signal indicating a "false rise" in water level. However, from the perspective of mass conservation, the outflow of steam is now instantaneously greater than the inflow of feedwater, and the actual total water storage in the steam generator is decreasing. Conversely, when the low-temperature feedwater flow suddenly increases, it temporarily inhibits boiling in the rising channel, causing bubbles to collapse, the mixture to contract, and its density to increase, resulting in a "false drop" in the water level indication. These counterintuitive dynamic responses have amplitudes that are proportional to the size of the disturbance, while the lag time is closely related to initial operating parameters such as feedwater subcooling.

[0022] Precise control of the steam generator water level is absolutely crucial for its safe operation. Excessive water level severely impacts the separation efficiency of the steam-water separator, leading to increased humidity and deterioration in the quality of the outlet steam. Saturated steam carrying water droplets can erode turbine blades at high speed, severely affecting their lifespan and even causing serious damage such as rotor imbalance. More critically, in extreme accident conditions such as steam pipe rupture, the enormous energy contained in excessive water can rush into the rupture site, causing excessive cooling of the reactor core, introducing positive reactivity, and potentially triggering complex reactive accidents. Conversely, insufficient water level means inadequate coverage of the heat transfer tube bundles. Some heat transfer tube tops may be periodically exposed to the steam environment, causing the tube walls to experience alternating wet and dry cycles, generating enormous alternating thermal stress, ultimately leading to thermal fatigue cracks or even rupture of the heat transfer tubes, causing primary loop boundary failure due to radioactive leakage. Excessive water level can also drain the feedwater pipeline, inducing highly destructive water hammer. Therefore, maintaining the steam generator water level strictly within a pre-set narrow window during the startup, commissioning, and entire operational life of a nuclear power plant is a crucial task that cannot be compromised.

[0023] Meanwhile, the saturated steam pressure on the secondary side of the steam generator is a core indicator for evaluating the performance of the secondary loop thermodynamic cycle. During steady-state full-power operation, higher saturated steam pressure translates to higher Rankine cycle thermal efficiency, directly improving the power plant's economics. However, during transients and commissioning, the saturated steam pressure is affected by a combination of parameters, including reactor power, primary loop inlet temperature and pressure, and steam flow rate. Excessive pressure may trigger the secondary side safety valves, leading to energy and working fluid losses; excessively low pressure results in a lower saturated temperature, enhancing heat exchange between the primary and secondary sides, potentially causing overcooling of the reactor coolant system and introducing unnecessary reactive disturbances. Furthermore, pressure changes can also affect the water level by altering the cavitation fraction, creating another complex control coupling loop. Traditional pressure control primarily involves adjusting the valve openings on the steam pipeline. Achieving coordinated and smooth control of water level and pressure amidst strong coupling is another major challenge facing existing control systems.

[0024] Faced with these complex challenges, the steam generator level and pressure control schemes currently used in most operating nuclear power plants are based on a distributed control system (DCS) architecture rooted in classical control theory. This scheme forms the cornerstone of automatic control, and its technical content can be broken down into three levels: sensing and measurement, control algorithms, and actuators.

[0025] At the sensing and measurement level, the system relies on a carefully arranged set of field instruments for data acquisition. Key measurement parameters include: the steam generator water level measured by a differential pressure transmitter (its output value is directly affected by the density of the steam-water mixture, directly reflecting a "false water level"); the main steam flow rate and feedwater flow rate measured by an orifice plate and a differential pressure transmitter; and the steam generator top pressure measured by a pressure transmitter. These analog signals, after sampling, are sent to the input module of the DCS controller, becoming the "eyes" of the entire control system for sensing the external environment.

[0026] At the control algorithm level, Proportional-Integral-Derivative (PID) control, with its simple structure and ease of implementation, has become the dominant control law. To overcome the inherent limitations of a single PID controller in dealing with "false water levels," a more advanced "three-impulse" control strategy is widely adopted in engineering. This strategy forms the core logic of water level control: it uses the water level deviation as the main feedback signal (impulse 1), responsible for long-term, steady-state error-free regulation; it introduces steam flow as a feedforward signal (impulse 2), aiming to detect load changes in real time, enabling the controller to anticipate and act in advance to partially offset the impact of impending false water levels; simultaneously, it uses feedwater flow as a feedback compensation signal (impulse 3), forming a rapid feedwater flow inner loop to promptly overcome internal disturbances such as feedwater pressure fluctuations and stabilize the feedwater flow. The weighted sum of these three signals jointly determines the controller's output. Pressure control typically employs an independent PID control loop, maintaining the pressure setpoint by adjusting the opening of the regulating valve on the main steam pipeline.

[0027] At the actuator level, the controller's output commands (typically 4-20mA analog signals) drive the final regulating equipment. For water level control, the core actuator is the feedwater regulating valve (or, for once-through boilers or certain specific operating conditions, the speed regulation of the feedwater pump). For pressure control, it mainly operates the opening degree of the main steam isolation valve or bypass discharge valve. These actuators are mostly pneumatic or electric, and their response speed is limited by mechanical inertia, dead zone, stroke, and other characteristics. From receiving the signal to fully executing the command, it usually takes several seconds or even longer.

[0028] Furthermore, it must be recognized that in existing technical solutions, a highly qualified operator team is itself an indispensable and crucial component of the control system, especially acting as the highest-level decision-making and intervention layer outside of DCS automatic control. During power plant startup, commissioning, load adjustments, or handling of abnormal operating conditions, operators need to continuously monitor the complex trend curves on the DCS human-machine interface. Relying on their deep understanding of system mechanisms, proficiency in operating procedures, and long-term accumulated experience, they can select automatic control modes, fine-tune setpoints, or, when necessary, switch to manual control for direct intervention to compensate for the shortcomings of the automatic control system in terms of intelligence and adaptability.

[0029] The dynamic performance of the entire control loop is constrained by the combined effects of the DCS scan cycle (typically 100-500 milliseconds), sensor response time, signal transmission delay, and the mechanical inertia of the actuator. Its overall closed-loop response time is usually in the second range, which is difficult to match the millisecond-level changes in the physical processes inside the steam generator.

[0030] Although the existing DCS combined with three-impulse control scheme has been tested in long-term engineering practice, its inherent technical limitations have exposed a series of fundamental defects that are difficult to overcome when facing the highly nonlinear, strongly coupled, and rapidly changing dynamic processes of steam generators. This has become a bottleneck in improving the safety, economy, and automation level of nuclear power plants.

[0031] First, there is an inherent deficiency in response speed and a timing mismatch. The existing system's second-level response time is orders of magnitude mismatched with the millisecond-level evolution of physical processes within the steam generator (such as the instantaneous formation and collapse of bubbles, and the propagation of density waves). The control system is always "half a beat late"; by the time it completes calculations based on sampled data from hundreds of milliseconds ago and ultimately drives the valves, the internal operating conditions of the equipment have already changed. This delay causes the control action to always lag behind the disturbance, not only significantly reducing the regulation effect but also frequently causing overshoot and oscillations, and even exacerbating system instability. In the initial transient phase of a rapid load change, the system is completely unable to suppress "false water levels" in real time.

[0032] Second, the control strategy suffers from model dependence and a lack of intelligence. Classical PID and three-impulse control are essentially model-driven controls based on linear and time-invariant assumptions. Their control effectiveness heavily relies on precise mathematical models and a pre-tuned set of fixed parameters. However, the dynamic characteristics of steam generators exhibit strong nonlinearity and time-varying behavior depending on factors such as power level, feedwater subcooling, and equipment aging. A fixed set of PID parameters cannot maintain optimal performance under all operating conditions. More importantly, the existing system completely lacks "cognitive" and "learning" capabilities. It cannot understand, like a human expert, that "false water levels" are physical deceptions caused by changes in mixture density, nor can it intelligently identify and decouple them, let alone perform forward-looking predictive compensation based on physical models. It can only mechanically react based on the current error and its historical and changing trends. Therefore, under severe disturbances, it is easily misled by false signals, making erroneous adjustments such as "reducing feedwater when the water level rises," thereby artificially exacerbating system fluctuations and even inducing accidents.

[0033] Third, the system has a weak ability to handle complex coupling and multi-objective optimization. There is a strong coupling relationship between water level and pressure. Traditional solutions involve designing independent single-loop controllers and minimizing mutual interference through decoupling. However, this decoupling is often imperfect; under large disturbances, the interaction between the two loops becomes very significant, potentially leading to unpredictable system instability. Existing systems lack the comprehensive decision-making capability to coordinate and optimize multiple key parameters such as water level and pressure, making it difficult to achieve optimal balance among multiple objectives. For example, how to coordinate the adjustment of feedwater valves and steam valves to simultaneously and smoothly control water level and pressure during rapid load increases.

[0034] Fourth, the limitations of automation and human capabilities. The high degree of automation in existing solutions is relative; under complex transient conditions, manual intervention by the operator is still ultimately necessary. This not only introduces the risk of human error, but more importantly, on a millisecond-scale timescale, the physiological limits of human reaction mean that humans are fundamentally incapable of effectively participating in the control process. When a "false water level" phenomenon occurs and ends within seconds, the operator may only just notice the anomaly in the trend curve and, before making a judgment, have already missed the optimal intervention opportunity. Therefore, relying entirely on human operators for control faces an insurmountable physiological performance ceiling when dealing with rapid transients.

[0035] Currently, nuclear power plant steam generators generally employ DCS control systems based on PID and three-impulse principles for water level control, similar to "cruise control" or early driver assistance technologies in automobiles. These systems are slow to respond, rely on precise linear models, and have low levels of intelligence, making them unable to handle inherent nonlinear and rapidly changing dynamic processes such as "false water levels." When faced with severe disturbances such as sudden load changes, these systems behave like a slow-reacting driver, exhibiting the following technical problems: First, the system's slow response speed prevents timely detection and response to dynamic processes such as "false water levels" that change within the steam generator at millisecond levels, resulting in significant control lag. Second, the control algorithm is designed based on linear and steady-state assumptions, lacking the intelligent understanding and prediction capabilities for nonlinear and strongly coupled operating conditions. It is easily interfered with by misleading signals such as "false water levels," and may even produce erroneous adjustments, exacerbating system fluctuations. Third, the existing systems have limited intelligence, heavily relying on manual intervention under complex transient conditions. However, manual operation has physiological limitations and safety ceilings at the millisecond timescale, making timely and accurate decision-making difficult.

[0036] To address the problems existing in related technologies, embodiments of this application provide a decision-making method for steam generators and a training method for intelligent decision-making models.

[0037] The decision-making method for the steam generator provided in the embodiments of this application will be introduced first below. For example... Figure 1 As shown, the method specifically includes the following steps: S101, obtain key operating parameters of the nuclear power plant.

[0038] Optionally, in this embodiment, key operating parameters refer to a set of core parameters that can accurately reflect the overall operating status of the nuclear power plant and the core operating conditions of the steam generator, providing basic data support for subsequent thermal-hydraulic parameter prediction and intelligent decision-making. These parameters need to cover the key operating dimensions of the primary and secondary loops of the steam generator, as well as the status information of related equipment. Specifically, they may include: the operating pressure, inlet and outlet temperatures, and average coolant temperature of the primary side of the nuclear power plant steam generator (reflecting the basic state of heat transfer in the primary loop); the current water level, steam pressure, steam flow rate, feedwater flow rate, and feedwater temperature of the secondary side of the steam generator (directly related to the core targets of secondary side water level and pressure control, reflecting the flow and heat transfer characteristics of the working fluid in the secondary loop); in addition, they may also include equipment status parameters related to the control execution of the steam generator, such as the current opening degree of the feedwater regulating valve, the current opening degree of the steam discharge valve, and the current speed of the feedwater pump (reflecting the real-time operating status of the actuator, providing a basis for judging the rationality of adjustment actions).

[0039] These key operating parameters need to be real-time and accurate, and can be acquired synchronously through sensor arrays. They are the basic data prerequisites for the effective implementation of the entire decision-making method.

[0040] S102, Based on the key operating parameters, predict the thermal-hydraulic parameters of the steam generator of the nuclear power plant in the future time period.

[0041] Optionally, in this embodiment, the thermal-hydraulic parameters refer to the core parameter set calculated by a prediction model based on the key operating parameters of the nuclear power plant obtained in S101. These parameters reflect the internal thermal processes (heat transfer, working fluid state changes) and hydraulic characteristics (working fluid flow, pressure distribution) of the steam generator over a future period, and are a quantitative representation of the future operating conditions of the steam generator. Specifically, these parameters may include: the predicted value of the secondary side water level of the steam generator over a future period (reflecting the dynamic change trend of the liquid level, providing a basis for responding to the risk of excessively high / low water levels in advance), the predicted value of the secondary side steam pressure (reflecting the evolution of the thermodynamic cycle state, supporting the advanced prediction of pressure anomalies), the cavitation fraction of the steam-water mixture (reflecting the degree of phase change of the secondary side working fluid, associated with the identification of nonlinear operating conditions such as "false water levels"), the predicted value of the working fluid mass flow rate of each key flow channel on the secondary side (such as the rising channel and the falling channel) (characterizing the working fluid flow characteristics, reflecting cycle stability), and the predicted value of the heat transfer power between the primary and secondary sides (reflecting heat transfer efficiency, associated with the economic efficiency of the thermodynamic cycle), etc. Providing complete operating condition information, including "current operating conditions + future trends," for subsequent intelligent decision-making models is a key basis for realizing the transformation of steam generator decision-making from "passive response" to "proactive control," and directly affects the accuracy and foresight of optimal adjustment actions.

[0042] Optionally, in one feasible implementation of this application, the key operating parameters obtained in S101 are first preprocessed, including outlier removal, missing value completion, and standardization. At the same time, the temporal features (such as the rate of change of water level and the fluctuation amplitude of steam flow rate in the past 100ms) and correlation features (such as the difference between feedwater flow rate and steam flow rate and the heat transfer temperature difference between primary and secondary side temperatures) of the key parameters are extracted to construct a high-dimensional feature vector, which comprehensively describes the operating status and parameter coupling relationship of the steam generator under the current operating conditions.

[0043] Secondly, data-driven prediction models (such as Long Short-Term Memory networks and gradient boosting trees) trained on historical operating data of nuclear power plants are used as the core prediction tools. These models can effectively capture the nonlinear and time-varying characteristics of steam generators by learning the time-series mapping patterns between key operating parameters and thermal-hydraulic parameters in historical data. For example, the Long Short-Term Memory network model can memorize long-term operating condition evolution trends through a gating mechanism and accurately learn the hysteresis response patterns of water level and pressure under load fluctuations.

[0044] Finally, the preprocessed high-dimensional feature vector is input into the trained prediction model. Based on the parameter evolution patterns learned from history, the model outputs predicted values ​​of thermal and hydraulic parameters for future time periods (such as 500ms or 1s). At the same time, to ensure the reliability of the prediction, an integrated prediction strategy can be introduced. By fusing the prediction results of multiple different models, a weighted average method is used to reduce the prediction bias of a single model. Finally, the predicted thermal and hydraulic parameters with the required accuracy for decision-making are obtained, providing advanced operating condition basis for subsequent intelligent decision-making in S103.

[0045] S103, The key operating parameters and the thermal-hydraulic parameters are input into the intelligent decision-making model, and the decision is made through the pre-learned decision table in the intelligent decision-making model to determine the optimal adjustment action of the steam generator; The decision table includes a mapping relationship between different decision actions and reward values ​​for the nuclear power plant and the steam generator under different operating conditions. The operating conditions are determined based on the key operating parameters of the nuclear power plant and the thermal-hydraulic parameters of the steam generator. The reward values ​​are determined based on the safety levels corresponding to the liquid level and pressure of the steam generator.

[0046] Optionally, in this embodiment, the decision table is a pre-stored "condition-action-reward" mapping relationship library in the intelligent decision-making model, which serves as the basic data support for the intelligent decision-making model to achieve rapid decision-making. The decision action includes all executable adjustment instruction options.

[0047] The adjustment action is the specific operation output by the intelligent decision-making model used to regulate the operating state of the steam generator. It should be noted that all adjustment actions must meet the physical limits of the actuator (such as valve opening degree 0~100%).

[0048] Operating status is a comprehensive quantitative indicator used to describe the current and future operating conditions of nuclear power plants and steam generators. It is the core index dimension of the decision table and is constructed by combining key operating parameters obtained from S101 with thermal-hydraulic parameters predicted from S102. Different combinations of parameters form different operating states, ensuring that the model can accurately distinguish between all operating scenarios and provide a precise basis for action matching.

[0049] The reward value is a core indicator in the decision table used to quantitatively evaluate the merits and demerits of different adjustment actions under corresponding working conditions. Its calculation logic is guided by the safe and stable operation of the steam generator and is directly related to the control effect of water level and pressure.

[0050] Safety is the core basis for calculating reward value. It is an indicator used to quantitatively assess whether the current and future water level and pressure status of the steam generator meet the requirements for safe operation.

[0051] Optionally, in one feasible implementation of this application, firstly, the key operating parameters obtained in S101 and the thermal-hydraulic parameters predicted in S102 are fused together, and the dimensional differences are eliminated through standardization transformation. Then, a high-dimensional state vector is constructed by combining the coupling relationship between the parameters to accurately define the current and future working states, ensuring a complete characterization of the nonlinear and time-varying working conditions of the steam generator.

[0052] Secondly, the constructed state vector is input into the intelligent decision-making model. The intelligent decision-making model uses a pre-trained feature matching algorithm (such as K-nearest neighbor search or deep learning feature extraction) to locate the most similar working state in the pre-learned decision table and calls the "decision action-reward value" mapping relationship stored in that state. The reward value in the decision table is dynamically calculated based on the safety level of liquid level and pressure. For example, when the predicted water level is close to the safety threshold, the reward value of the action to suppress water level fluctuation is increased. When the pressure is in the high-efficiency range, the reward value of the action to maintain pressure stability is increased, ensuring that the reward value is directly related to the safety control target.

[0053] Finally, the intelligent decision-making model sorts all matched decision actions in descending order of reward value, selects the candidate action with the highest reward value for safety verification, and verifies whether the action exceeds the physical limits of the actuator and whether it may cause abnormal coupling parameters (such as whether adjusting the water flow will cause pressure overpressure). If the verification passes, it is determined as the optimal adjustment action; if the verification fails, the next highest reward value action is selected for re-verification until an adjustment action that meets the safety requirements is output, thus achieving dual protection of safety and accuracy.

[0054] In a decision-making method for a steam generator provided in this application embodiment, key operating parameters of the nuclear power plant are acquired to ensure accurate perception of the current operating status of the steam generator. Based on this, the thermal-hydraulic parameters of the steam generator for future periods are predicted using these key operating parameters. This prediction process allows decision-making to move beyond passive responses to current operating conditions and instead provides a forward-looking understanding of future operating conditions, reducing decision-making biases caused by the inability to adapt to dynamic changes in operating conditions. Finally, the key operating parameters and the predicted thermal-hydraulic parameters are jointly input into an intelligent decision-making model, and the optimal adjustment action is determined through a pre-learned decision table. The decision table constructs a "decision action-reward value" mapping relationship based on different operating states of the nuclear power plant and the steam generator, with the reward value determined by the safety of the liquid level and pressure, ensuring that the decision action always prioritizes the safe and stable operation of the steam generator. This process, through multi-parameter fusion to define operating conditions and safety-oriented reward value constraints on the selection of adjustment actions, enables the intelligent decision-making model to adapt to different complex operating conditions, resulting in adjustment actions that better match actual operating requirements, significantly reducing decision-making biases, and thus improving the reliability of steam generator decision-making.

[0055] In one embodiment, predicting the thermal-hydraulic parameters of the nuclear power plant's steam generator over a future time period based on the key operating parameters includes: Construct a digital twin model of the steam generator; The key operating parameters are input into the digital twin model, and the digital twin model performs joint calculations based on the key operating parameters to obtain the thermal-hydraulic parameters of the steam generator in the future time period.

[0056] Optionally, in the embodiments of this application, the digital twin model is a digital mirror image of the physical entity of the steam generator. It is a tool for dynamically simulating the thermal-hydraulic characteristics and predicting the future state of the steam generator by constructing a virtual model that is highly consistent with the actual steam generator structure and operating mechanism.

[0057] The digital twin model is based on the actual structural parameters of the steam generator and combines physical laws such as thermal balance, mass conservation, energy conservation, and momentum conservation to establish control equations covering the primary and secondary side loops, accurately simulating the physical processes such as flow, heat transfer, and phase change of the working fluid in the steam generator.

[0058] Optionally, in one specific implementation of this application, firstly, a digital twin model of the steam generator is constructed. This model uses the physical entity of the steam generator as a prototype and, based on its structural characteristics and operating mechanism, abstracts the complex system into a multi-dimensional virtual mapping. During modeling, the structural parameters and physical laws in the design drawings need to be integrated to form a mathematical model framework that reflects the actual operating characteristics, covering the coupling relationship between the primary and secondary sides, ensuring a high degree of consistency between the virtual model and the physical entity in key characteristics.

[0059] Next, the key operating parameters obtained by S100 are input into the digital twin model. These parameters serve as the boundary conditions and initial conditions for the calculation of the digital twin model, providing a real-time operating condition benchmark for the simulation.

[0060] Finally, joint calculations are performed using a digital twin model: based on the input key operating parameters, the digital twin model invokes built-in mathematical equations and employs numerical calculation methods (such as the finite difference method and iterative solution algorithms) to simultaneously simulate the thermal-hydraulic processes on the primary and secondary sides, simulating the flow of the working fluid, heat transfer, and state changes between the control volumes. The digital twin model can extrapolate parameter evolution over future time periods (e.g., 1-5 seconds), outputting predicted secondary water levels, steam pressure trends, and flow distributions in each channel, among other thermal-hydraulic parameters.

[0061] In these alternative embodiments, a digital twin model is used to accurately predict the future operating state of the steam generator, providing advanced data with physical mechanism support for subsequent decision-making and ensuring the real-time performance and accuracy of the simulation results.

[0062] In one embodiment, the steam generator includes a primary side circuit and a secondary side circuit; The construction of the digital twin model of the steam generator includes: Based on the heat balance relationship of the primary side loop, the primary side control equation is constructed; the primary side control equation is used to solve for the corresponding thermal-hydraulic parameters of the primary side loop. Based on the primary-side control equations, a first physical model corresponding to the primary-side loop is constructed; the first physical model includes a network consisting of multiple primary-side control bodies and primary-side flow channels connecting the primary-side control bodies. Based on the mass conservation, energy conservation, and momentum conservation relationships of the secondary side loop, the secondary side control equations are constructed, which are used to solve for the corresponding thermo-hydraulic parameters of the secondary side loop. Based on the secondary-side control equations, a second physical model corresponding to the secondary-side loop is constructed; the second physical model includes a network consisting of multiple secondary-side control bodies and secondary-side flow channels connecting the secondary-side control bodies. The digital twin model is constructed based on the first physical model and the second physical model.

[0063] Optionally, in this embodiment, the heat balance relationship is the fundamental physical law for constructing the primary-side control equations, meaning that the heat input and output of the primary-side loop of the steam generator remain balanced per unit time (in the dynamic process, the rate of change of heat equals the difference between the input and output). For the primary-side loop, the input heat mainly comes from the heat brought in by the reactor coolant, while the output heat is the heat transferred to the secondary working fluid through the heat transfer tubes. At the same time, the changes in heat storage within the loop itself (such as heat storage / release on the metal wall) must be considered.

[0064] The secondary-side control equations are a set of equations based on the mass, energy, and momentum conservation relationships of the secondary-side loop, used to solve for the thermo-hydraulic parameters of the secondary side. The mass conservation relationship describes the mass change of the control volume per unit time; the energy conservation relationship describes the energy change within the control volume; and the momentum conservation relationship describes the momentum change of the working fluid within the control volume. By solving this set of equations in a coupled manner, the complex processes of flow, heat transfer, and phase change of the working fluid on the secondary side can be dynamically reflected.

[0065] A control volume is a basic unit in a physical model used to simplify complex systems and calculate local parameters. It refers to a spatial region with fixed boundaries and volume, and can be regarded as a "node" constituting a loop network. For a primary-side control volume, its boundary is usually a specific region of a heat transfer pipe or header, and its volume is determined by the actual structural dimensions (such as pipe diameter and length). It is filled with primary-side coolant and used to calculate parameters such as average temperature, pressure, and internal energy within this region. A secondary-side control volume corresponds to the gas-water mixture region on the secondary side (such as segments of rising or falling channels). Its boundaries are also fixed, and the local thermal-hydraulic characteristics are reflected by calculating the mass and energy changes of the working fluid within the control volume.

[0066] Flow channels are the channel structures connecting adjacent control bodies, used to describe the flow path of the working fluid between control bodies, forming a "connection network" in the physical model. Primary-side flow channels correspond to the flow channels of the coolant in the primary-side loop (such as the connection between heat transfer tube sections), and their parameters include length, cross-sectional area, and friction coefficient, used to calculate the pressure loss, flow distribution, and heat transfer of the working fluid during flow. Secondary-side flow channels correspond to the flow path of the secondary-side working fluid (water, steam, or a steam-water mixture) (such as the connection between the rising channel and the steam drum), and the influence of phase change on flow characteristics (such as velocity changes caused by an increase in the steam fraction) needs to be considered. By establishing the mass, momentum, and energy transfer relationships between adjacent control bodies, flow channels couple the local calculations of each control body, ensuring that the model can reflect the working fluid flow and parameter transfer process of the entire loop.

[0067] Optionally, in one specific implementation of this application, firstly, for the primary side loop, based on the heat balance relationship and loop structural parameters (control volume, heat transfer area), the primary side control equation is derived. Then, the primary side loop is discretized into multiple primary side control bodies according to the flow path and structural characteristics (e.g., each heat transfer tube corresponds to one control body). These control bodies are connected into a network through the primary side flow channel (simulating the coolant flow channel between control bodies) to form the first physical model, thereby realizing the local and global calculation of the primary side thermal-hydraulic parameters.

[0068] Next, for the secondary side loop, based on the conservation of mass, energy, and momentum, a secondary side control equation including mass, internal energy, and flow rate equations is constructed. Similarly, the secondary side is divided into secondary side control volumes corresponding to rising and falling channels, and connected by secondary side flow channels to form a second physical model, adapted to the steam-water two-phase flow characteristics of the secondary side. Finally, through a heat transfer coupling module (linking the heat transfer power of the primary side control volume and the corresponding secondary side control volume), the first and second physical models are integrated, enabling synchronous calculation and data interaction between the two models, ultimately forming a complete digital twin model of the steam generator.

[0069] In these alternative embodiments, by modeling and integrating the sub-loops, equations are constructed based on physical laws to ensure accurate calculation of primary and secondary thermal-hydraulic parameters. The actual structure is also restored using the control body and flow channel network to adapt to complex operating conditions. The final integrated digital twin model can accurately map the operating state of the steam generator, providing reliable model support for subsequent prediction of thermal-hydraulic parameters, ensuring accurate prediction results, and providing an effective basis for decision-making.

[0070] In one embodiment, the step of inputting the key operating parameters into the digital twin model, and using the digital twin model to perform joint calculations based on the key operating parameters to obtain the thermal-hydraulic parameters of the steam generator for a future time period, includes: By substituting the key operating parameters as boundary and initial conditions into the primary and secondary control equations, the thermal-hydraulic parameters of the steam generator in the future time period can be obtained.

[0071] Optionally, in this embodiment, boundary conditions refer to the constraints in the digital twin model used to limit the parameter values ​​of the primary and secondary loops at specific locations or interfaces. For the primary loop, boundary conditions may include the pressure, temperature, and flow rate of the primary inlet coolant, and the pressure constraint at the primary outlet; for the secondary loop, boundary conditions cover the flow rate and temperature of the feedwater inlet, and the pressure of the steam outlet. These boundary conditions simulate the interaction between the steam generator and other systems in the nuclear power plant (such as the reactor coolant system, feedwater system, and turbine system), ensuring that the working fluid flow and heat transfer processes calculated by the model conform to the external constraints of the actual operating scenario.

[0072] Initial conditions refer to the parameter values ​​of each control element in the primary and secondary loops at the initial moment when the digital twin model begins calculating the thermal-hydraulic parameters for future time periods. These are the "starting point" data for the digital twin model's extrapolation. Specifically, they include: the initial coolant temperature, pressure, and internal energy of each control element on the primary side; the initial working fluid mass, temperature, and pressure of each control element on the secondary side; and the initial water level on the secondary side. Initial conditions ensure that the model starts from the current actual operating state of the steam generator, rather than from an idealized initial state, laying the foundation for accurately extrapolating parameter trends over future time periods.

[0073] In these alternative embodiments, substituting key operating parameters as boundary and initial conditions into the equations allows the digital twin model to fit the actual operating scenario of the steam generator, avoiding deviations caused by idealized assumptions; joint calculation ensures that primary and secondary parameters are solved in a coordinated manner, accurately outputting future thermal-hydraulic parameters and providing reliable advanced data support for subsequent decision-making.

[0074] In one embodiment, constructing the primary-side control equations based on the thermal balance relationship of the primary-side loop includes: Calculate the first product based on the mass of the i-th primary side control body and the first derivative; the first derivative is the derivative of the internal energy and time of the i-th primary side control body. The second product is calculated based on the difference between the fluid enthalpy values ​​entering and leaving the i-th primary side control body and the flow rate of the primary side loop; Calculate the first difference between the second product and the heat power transferred to the secondary side circuit; Calculate the first difference and the first sum of the heat capacities of the tube sheet and lower head metal of the steam generator; By establishing an equation with the first product and the first sum, the primary control equation is obtained.

[0075] Optionally, in one specific implementation of this application, the primary-side governing equation is (single-phase incompressible fluid assumption): (1) in, Let the mass of the i-th primary control body be denoted as . Let i be the internal energy of the i-th primary control volume. It is the first derivative; The first product; and These are the enthalpy values ​​of the fluid entering and leaving the control volume, respectively. The difference between the fluid enthalpy entering and leaving the i-th primary side control volume; For the flow rate of the primary side loop, This is the second product; The heat power transferred to the secondary circuit; It is the worst; The heat capacity of the tube sheet and lower end cap metal of the steam generator; For the first and the last.

[0076] In these alternative embodiments, by refining the energy balance (mass and internal energy derivative, enthalpy difference and flow rate, heat transfer power, etc.) of the primary control volume and incorporating the influence of metal heat capacity, the constructed primary control equations can accurately reflect the heat balance relationship, reflecting both the dynamic changes in working fluid flow and heat transfer and taking into account the structural heat storage characteristics, thereby improving the accuracy of primary thermal-hydraulic parameter calculations.

[0077] In one embodiment, the secondary-side control equations include the mass equation, the internal energy equation, and the flow rate equation; The process of constructing the secondary-side control equations based on the mass conservation, energy conservation, and momentum conservation relationships of the secondary-side loop includes: The mass equation is determined based on the flow rates entering and leaving the secondary side flow channels of the secondary side control body, and the flow rates exchanged between the steam generator body and the external boundary. The internal energy equation is determined based on the flow rates entering and leaving the secondary side flow channels of the secondary side control body, the enthalpy values ​​of the fluids entering and leaving the secondary side control body, the flow rates exchanged between the steam generator body and the external boundary, the enthalpy values ​​of the fluids exchanged between the steam generator body and the external boundary, the internal energy of the heat transfer tube metal of the steam generator, and the generalized heat; the generalized heat is used to characterize the heat exchange of the non-core heat transfer process within the secondary side control body. The flow equation is determined based on the equivalent length and cross-sectional area of ​​the secondary side channel, the pressure of the two secondary side control bodies connected to the secondary side channel, and the frictional pressure drop, local pressure drop, heavy pressure drop, and acceleration pressure drop within the secondary side channel.

[0078] In one embodiment, the generalized heat includes at least one of the following: The heat transferred through the sleeve wall by the descending and ascending channels of the steam generator, the heat transferred by the thermal components of the steam generator, the heat transferred by thermal diffusion of the steam generator, and the heat transferred by the condensation of steam upon encountering the wall of the steam generator.

[0079] Optionally, in the embodiments of this application, the mass equation is an equation constructed based on the mass conservation relationship of the secondary control volume, used to describe the mass change law of the control volume per unit time.

[0080] The internal energy equation is an equation constructed based on the energy conservation relationship of the secondary control volume, used to describe the change law of total energy within the control volume.

[0081] The flow equation is based on the momentum conservation relationship of the secondary side channel and is used to describe the relationship between the flow rate and pressure difference of the working fluid in the channel.

[0082] The flow rate exchanged between the steam generator body and the external boundary refers to the amount of working fluid exchanged between the secondary circuit and other external systems of the nuclear power plant (such as the feedwater system and the turbine system). It is an important boundary parameter in the mass equation and the internal energy equation. For example, the feedwater flow rate input to the secondary side through the feedwater pipeline (external input flow rate), the steam flow rate output from the secondary side to the turbine (external output flow rate), and the steam discharge flow rate when the safety valve is activated, etc.

[0083] Generalized heat is a heat term in the internal energy equation used to supplement the description of non-core heat transfer processes within the secondary control volume. It covers secondary heat exchanges that cannot be covered by the heat transferred from the primary side or the energy of the working fluid flow. Examples include heat loss between the control volume and the steam generator shell, weak heat exchange between different control volumes through radiation or natural convection, and compensation for energy conservation deviations caused by measurement errors. Although these heats usually account for a very small percentage, including generalized heat can improve the calculation accuracy of the internal energy equation and ensure that the energy conservation relationship is strictly satisfied under complex operating conditions.

[0084] The internal energy of the heat transfer tube metal refers to the thermal energy stored on the metal wall of the heat transfer tube corresponding to the secondary control volume. Its magnitude is related to the metal mass, specific heat capacity, and temperature. In the internal energy equation, changes in the internal energy of the heat transfer tube metal (such as heat absorption when the temperature rises and heat release when the temperature falls) will affect the energy balance of the secondary control volume. When the metal temperature is higher than the secondary working fluid temperature, the metal releases internal energy (a positive contribution); conversely, it absorbs internal energy (a negative contribution).

[0085] The equivalent length of the secondary flow channel is a simplified value of the actual flow channel geometry, used to uniformly quantify the flow channel resistance in the flow equation. The actual secondary flow channel may have complex structures such as bends and diameter changes (e.g., the connection section between the riser channel and the header). The equivalent length is obtained by converting the resistance of local structures (such as elbows and valves) into the length of a straight pipe with equal resistance, and adding it to the length of the straight pipe section of the flow channel.

[0086] The cross-sectional area of ​​the secondary flow channel refers to the cross-sectional area of ​​the flow channel perpendicular to the direction of the working fluid flow.

[0087] The pressure of the two secondary side control bodies connected to the secondary side flow channel refers to the pressure values ​​at both ends of the secondary side flow channel, belonging to two adjacent secondary side control bodies respectively.

[0088] Frictional pressure drop is the pressure loss caused by the friction between the fluid viscosity and the channel wall when the working fluid flows in the secondary side channel. Its magnitude is related to the channel length, roughness, working fluid velocity and density.

[0089] Local pressure drop is the pressure loss caused by flow field disturbances (such as eddies, changes in flow velocity direction) when the working fluid flows through local structures (such as elbows, valves, and abrupt changes in cross-section) in the secondary side channel. Its magnitude is related to the shape coefficient of the local structure (such as the curvature ratio of the elbow) and the kinetic energy of the working fluid (usually proportional to the square of the flow velocity).

[0090] Gravity pressure drop (also known as gravity pressure drop) is the pressure difference caused by uneven density distribution of the working fluid in the secondary flow channel (such as the density difference between steam and water in a steam-water two-phase flow) or by the inclined / vertical arrangement of the flow channel. Its magnitude is related to the height of the flow channel, the average density of the working fluid, and the gravitational acceleration (such as in a vertically rising flow channel, the pressure formed by high-density water at the bottom is higher than that of low-density steam at the top).

[0091] Accelerated pressure drop is the pressure loss caused by changes in momentum of the working fluid in the secondary flow channel due to changes in flow velocity (such as changes in cross-section leading to increases or decreases in flow velocity) or phase changes (such as the increase in steam content due to water evaporation, resulting in a sharp increase in flow velocity). Its magnitude is related to the density of the working fluid and the amount of change in flow velocity. In the flow of steam-water mixtures in the secondary side (such as in the rising channel near saturation), the accelerated pressure drop caused by phase changes is particularly significant.

[0092] In these alternative embodiments, when constructing the secondary-side control equations, the mass equation accurately correlates the internal and external flow rates, the internal energy equation incorporates multiple energy terms and generalized heat, and the flow equation refines the flow channel parameters and pressure drop, comprehensively covering the conservation of mass, energy, and momentum. This not only fits the complex characteristics of the secondary-side steam-water two-phase flow but also reduces the omission of non-core factors, improves the calculation accuracy of the equations for thermal-hydraulic parameters, and provides reliable mathematical support for the digital twin model.

[0093] In one embodiment, determining the mass equation based on the flow rates entering and leaving the secondary side flow channels of the secondary side control body, and the flow rates exchanged between the steam generator body and the external boundary, includes: Calculate the second sum of the flow rates of all secondary side channels entering the i-th secondary side control body; Calculate the third sum of the flow rates of all secondary side channels leaving the i-th secondary side control body; Calculate the second difference between the second sum and the third sum; Calculate the third difference between the second difference and the flow rate exchanged between the steam generator body and the external boundary; By establishing an equation between the second derivative and the third difference, the mass equation is obtained; the second derivative is the mass-time derivative of the i-th quadratic control volume.

[0094] Alternatively, in one specific implementation of this application, the mass equation is as follows: (2) in, The second sum of the flow rates entering the secondary side channels of the i-th secondary side control body; The third sum of the flow rates leaving the secondary side channels of the i-th secondary side control body; It is the second worst; The total flow rate exchanged between the steam generator body and the external boundary; It is the third worst; Let the mass of the i-th secondary side control body be . This is the second derivative.

[0095] In these alternative embodiments, by subdividing the flow rates entering and leaving the control body and the flow rates exchanged with the outside, the rate of change in mass is accurately calculated, and the constructed mass equation strictly follows the law of mass conservation. It encompasses both the working fluid flow within the internal channels and the flow interactions with the external system, comprehensively reflecting the mass dynamics of the secondary control body and providing a reliable foundation for the secondary side model.

[0096] In one embodiment, determining the internal energy equation based on the flow rates entering and leaving the secondary side flow channels of the secondary side control body, the enthalpy values ​​of the fluids entering and leaving the secondary side control body, the flow rates exchanged between the steam generator body and the external boundary, the enthalpy values ​​of the fluids exchanged between the steam generator body and the external boundary, the internal energy of the heat transfer tube metal of the steam generator, and generalized heat includes: Calculate the fourth sum of the products between the flow rates of all secondary flow channels entering the i-th secondary control body and the enthalpy values ​​of all fluids entering the i-th secondary control body's secondary flow channels; Calculate the fifth sum of the products between the flow rates of all secondary side channels leaving the i-th secondary side control body and the fluid enthalpy values ​​of all secondary side channels leaving the i-th secondary side control body; Calculate the fourth difference between the fourth sum and the fifth sum; The sixth sum of the product of the flow rate exchanged between the steam generator body and the external boundary and the enthalpy of the fluid exchanged between the steam generator body and the external boundary; Calculate the fifth difference between the fourth difference and the sixth sum; Calculate the fifth difference, the internal energy of the heat transfer tube metal of the steam generator, and the seventh sum between the generalized heat; By establishing an equation between the seventh and third derivatives, the internal energy equation is obtained; the third derivative is the derivative of the internal energy of the i-th quadratic control volume with time.

[0097] Alternatively, in one specific implementation of this application, the internal energy equation is as follows: (3) Among them, subscript in and out Indicates entering and leaving the control unit; Enthalpy of the fluid; subscript ex This indicates that the feedwater and steam connections are located away from the steam generator body; The internal energy of the secondary control body; It is the product of the flow rate entering the secondary side channel of the secondary side control body and the enthalpy of the fluid entering the corresponding secondary side channel of the secondary side control body; For the fourth and; It is the product of the flow rate leaving the secondary side control body and the enthalpy of the fluid leaving the corresponding secondary side control body. For the fifth and; It is the fourth worst; The product of the flow rate exchanged between the steam generator body and the external boundary and the enthalpy of the fluid exchanged between the steam generator body and the external boundary; For the sixth and; It is the fifth worst; This refers to the internal energy of the heat transfer tube metal in the steam generator; For generalized heat; For the seventh and; Let i be the internal energy of the i-th secondary side control body; This is the third derivative.

[0098] In these alternative embodiments, by refining the energy exchange between internal and external flow rates and enthalpy values, and superimposing the internal energy of the heat transfer tube metal and generalized heat, the constructed internal energy equation strictly adheres to the law of energy conservation. It comprehensively covers energy terms such as working fluid flow, boundary heat transfer, and structural heat storage, accurately reflecting the dynamic changes in internal energy within the secondary control system, and improving the calculation accuracy of parameters such as temperature and pressure.

[0099] In one embodiment, determining the flow equation based on the equivalent length and cross-sectional area of ​​the secondary side channel, the pressures of the two secondary side control bodies connected to the secondary side channel, and the frictional pressure drop, local pressure drop, gravity pressure drop, and acceleration pressure drop within the secondary side channel includes: The sum of the frictional pressure drop, local pressure drop, heavy pressure drop, and acceleration pressure drop in the secondary side channel is determined as the drag loss; The seventh difference between the sixth difference and the resistance loss is determined as the net pressure difference of the secondary side flow channel; the sixth difference is the pressure difference between the two secondary side control bodies connected to the secondary side flow channel. The ratio between the equivalent length and cross-sectional area of ​​the secondary side flow channel is determined as the proportional coefficient of the secondary side flow channel. The flow equation is obtained by establishing an equation using the fourth derivative and the third product; the fourth derivative is the derivative of the flow rate and time in the secondary side channel; the third product is the product between the proportionality coefficient and the net pressure difference.

[0100] Alternatively, in one specific implementation of this application, the flow equation is as follows: (4) Wherein, the subscripts i and j represent the two secondary side control bodies connected to the flow channel k; For frictional pressure drop; For localized pressure drop; For heavy position voltage drop; To accelerate the pressure drop; For resistance loss; The pressure of the secondary control body i; The pressure of the secondary control body j; It is the sixth worst; The seventh difference (i.e., the net pressure difference in the secondary side channel); Let k be the equivalent length of the secondary side channel; Let k be the cross-sectional area of ​​the secondary side channel; The proportionality coefficient of the secondary side channel k; The flow rate of the secondary side channel k; It is the fourth derivative.

[0101] In these alternative embodiments, the flow equation constructed by integrating channel resistance losses, net pressure difference, and structural proportionality coefficients strictly adheres to momentum conservation. By correlating various pressure drops with pressure differences and combining the influence of channel structure on flow, the dynamic changes in flow rate are accurately reflected, improving the accuracy of secondary working fluid flow calculations.

[0102] In one embodiment, constructing the second physical model corresponding to the secondary-side loop based on the secondary-side control equation includes: The secondary side loop is divided into multiple physical branches, including: a first branch for characterizing the steam space between the top of the steam-water separator and the steam outlet pipe; a second branch for characterizing the water supply chamber between the top of the separator and the top of the tube bundle sleeve; a third branch for characterizing the rising channel between the top of the separator and the top of the inner sleeve; a fourth branch for characterizing the falling channel; and a fifth branch for characterizing the rising channel. Based on the multiple physical branches and the secondary-side control equations, a second physical model is constructed; wherein, the secondary-side control body is used to describe the changes in pressure, temperature, specific enthalpy, and mass of the working fluid within each of the physical branches; the secondary-side flow channel is used to describe the changes in the mass flow rate of the working fluid connecting each of the secondary-side control bodies and the head loss of the circulation loop.

[0103] Optionally, in this embodiment, the first branch is a physical branch in the secondary side loop used to characterize the steam space between the top of the steam-water separator and the steam outlet inlet pipe. It mainly covers the channel through which qualified steam after steam-water separation flows to the steam outlet. The working medium in this branch is mainly steam, and its core function is to simulate the flow and state changes (such as pressure loss and temperature fine-tuning) of steam after leaving the separator and before entering the external steam system, to ensure the accurate calculation of steam outlet parameters (such as pressure and dryness), and to reflect the process of steam being transported to external equipment such as the steam turbine.

[0104] The second branch characterizes the physical branch of the feedwater chamber between the top of the steam-water separator and the top of the tube bundle sleeve, corresponding to the area where secondary side feedwater is temporarily stored and distributed after entering. The working fluid in this branch is mainly unsaturated or saturated water, primarily simulating the mixing, temporary storage, and initial distribution process of feedwater entering from the external system in the feedwater chamber, towards the descending or ascending channel. Its thermo-hydraulic parameters (such as feedwater temperature and pressure) directly affect the heating and phase change efficiency of the subsequent working fluid.

[0105] The third branch is a physical branch representing the rising channel from the top of the steam-water separator to the top of the inner sleeve, corresponding to the final section of the steam-water mixture flowing into the separator. The working fluid in this branch is a two-phase flow of steam and water with a high steam content. It mainly simulates the flow state of the working fluid as it approaches the separator and its heat exchange with the surrounding structure. It is a crucial transition section before the steam-water mixture enters the separation stage and has a direct impact on the separation efficiency.

[0106] The fourth branch characterizes the physical branch of the descending channel in the secondary loop, which is the channel through which the working fluid flows from the feedwater chamber or the lower part of the steam drum to the bottom of the tube bundle. The working fluid in this branch is mainly water (possibly containing a small amount of steam), and it mainly simulates the downward flow process of the working fluid under the action of gravity or circulation head, including velocity distribution, pressure loss, and heat exchange with the surrounding structure (such as weak heat transfer with the ascending channel). It is an important component of the secondary natural circulation loop.

[0107] The fifth branch characterizes the physical branch of the rising channel in the secondary loop, corresponding to the channel through which the working fluid flows upward after absorbing heat from the primary side in the tube bundle region. Within this branch, the working fluid undergoes a phase change process from water to a steam-water mixture, with the steam fraction gradually increasing as the heat absorption process progresses. It mainly simulates the flow, heat absorption, phase change, and acceleration processes of the working fluid outside the heat transfer tubes. Its thermo-hydraulic parameters (such as flow velocity, void fraction, and pressure drop) directly reflect the steam generation efficiency of the steam generator and represent the core region of secondary energy conversion.

[0108] Optionally, in one specific implementation of this application, such as Figure 2 As shown, the secondary fluid in the steam generator generates a driving head based on the density difference between the descending and ascending channels, forming a natural circulation loop. This allows for the construction of... Figure 3 The second physical model is shown.

[0109] Specifically, branch ① (i.e., the first branch): the steam space between the top of the steam-water separator and the outlet pipe of the steam generator; Branch ② (i.e., the second branch): the water supply chamber section outside the separator, between the top of the separator and the top of the tube bundle sleeve; Branch ③ (i.e., the third branch): The rising channel from the top of the separator to the top of the inner sleeve; Branch ④ (i.e., the fourth branch) and Branch ⑤ (i.e., the fifth branch): represent the descending channel and the ascending channel, respectively.

[0110] Each branch consists of several control volumes, which are used to describe changes in pressure, temperature, specific enthalpy, and mass, while the flow channels are used to describe changes in mass flow rate and various head losses in the circulation loop.

[0111] Optionally, the volumetric properties, pressure, and specific enthalpy within the body are determined according to the equation of state: v=v(P,h), ρ=ρ(P,h), T=T(P,h) (5) in, v=v(P,h) In this equation, v represents the specific volume of the working fluid, P represents the pressure of the working fluid, and h represents the specific enthalpy of the working fluid. This equation shows that the specific volume of the working fluid is a function of pressure and specific enthalpy, and its value is determined by the state of the working fluid under the current pressure and specific enthalpy.

[0112] ρ=ρ(P,h) In ρ The density of the working fluid is expressed as P, and the meanings of P and h are the same as above. Density and specific volume are reciprocals of each other, so this formula shows that density is also determined by pressure and specific enthalpy, reflecting the mass distribution characteristics of the working fluid under specific conditions.

[0113] T=T(P,h) In this equation, T represents the temperature of the working fluid, and P and h have the same meaning as above. This equation shows that the temperature of the working fluid is a function of pressure and specific enthalpy, and its value depends on the thermal equilibrium level of the working fluid under the current pressure and energy state, and is used to characterize the hotness or coldness of the working fluid.

[0114] In these alternative embodiments, the secondary-side loop is functionally divided into five physical branches. A model is constructed by combining the secondary-side control equations, enabling the control volume to accurately describe the working fluid state of each branch, and the flow channels to reflect the connection characteristics. This not only reproduces the real processes of secondary-side steam-water flow and heat transfer, but also simplifies the complex system through branch division, improving the model's simulation accuracy and computational efficiency for thermal-hydraulic parameters.

[0115] In one embodiment, the step of inputting the key operating parameters into the digital twin model, and using the digital twin model to perform joint calculations based on the key operating parameters to obtain the thermal-hydraulic parameters of the steam generator for a future time period, includes: By substituting the key operating parameters as boundary and initial conditions into the primary and secondary control equations, the thermal-hydraulic parameters of the steam generator in the future time period can be obtained.

[0116] Optionally, in one specific implementation of this application, the key operating parameters of the nuclear power plant obtained are first set as the boundary conditions and initial conditions of the primary side control equation (formula (1)) and the secondary side control equation (including the mass equation formula (2), the internal energy equation formula (3), and the flow rate equation formula (4)).

[0117] Next, these parameters are substituted into formulas (1)-(4) to initiate the joint solution process of the digital twin model: the thermal parameters of each control body on the primary side are calculated using formula (1), and the mass change, internal energy change, and flow rate on the secondary side are solved simultaneously using formulas (2)-(4). During the process, the heat transfer data of the primary and secondary sides are exchanged in real time to ensure the parameter coupling and matching between the equations. The collaborative solution of multiple sets of equations is completed through iterative calculation, and the thermal-hydraulic parameters of the steam generator for the future time period are finally output.

[0118] In these alternative embodiments, using key operating parameters as the boundaries and initial conditions of the equations allows the calculations to closely reflect the actual operating conditions of the steam generator, avoiding deviations from idealized assumptions. By jointly solving the primary and secondary control equations, the parameters of both loops are calculated collaboratively, accurately outputting future thermal-hydraulic parameters. This provides reliable, up-to-date data support for subsequent intelligent decision-making, ensuring the accuracy and safety of the decisions.

[0119] In one embodiment, the secondary-side control equations include the mass equation and the internal energy equation; The process involves inputting the key operating parameters into the digital twin model, and then using the digital twin model to perform joint calculations based on the key operating parameters to obtain the thermal-hydraulic parameters of the steam generator for a future time period, including: For the target secondary control body with water level among the plurality of secondary control bodies, the target secondary control body is divided into an upper steam region and a lower water region. Based on the key operating parameters corresponding to the target secondary control body, mass conservation equations and energy conservation equations are established for the upper steam region and the lower water region respectively through the mass equation and the internal energy equation; the mass conservation equation and the energy conservation equation are used to describe the differentiated thermo-hydraulic characteristics of the steam-water two-phase region; By jointly solving the mass conservation equation and the energy conservation equation, the thermal-hydraulic parameters corresponding to the differentiated thermal-hydraulic characteristics of the steam-water two-phase region in the future time period are obtained.

[0120] Optionally, in this embodiment, the upper steam region is the space above the water level in the target secondary control body (the control body with a water level), where the working fluid is mainly saturated steam or superheated steam. Its core characteristics are low working fluid density, high fluidity, and existence in steam form, with relatively uniform thermo-hydraulic parameters.

[0121] The lower water region is the space below the water level in the target secondary control volume. The working medium in this region is mainly saturated or unsaturated water. Its characteristics are that the working medium density is much higher than that of the upper steam region, its fluidity is relatively weak, and its thermo-hydraulic parameters are relatively stable. It mainly achieves temperature rise by absorbing heat from the primary side, providing energy for phase change (water evaporation into steam).

[0122] The differentiated thermo-hydraulic characteristics of the steam-water two-phase region refer to the significant differences in thermo-hydraulic parameters and physical processes between the upper steam region and the lower water region in the target secondary control volume. Specifically, in terms of working fluid state, the upper region is gaseous and the lower region is liquid; in terms of parameters, the steam region has lower density and higher flow velocity, while the water region has higher density and lower flow velocity; in terms of physical processes, the steam region is dominated by flow and a small amount of heat exchange, while the water region is dominated by endothermic heating and evaporation phase change. These differentiated characteristics necessitate the establishment of separate equations for each region to accurately describe their respective parameter variations.

[0123] Optionally, in one specific implementation of this application, the target secondary control body with water level in the secondary side circuit is first identified, and it is divided into an upper steam region and a lower water region according to the working fluid form.

[0124] Next, using the key operating parameters corresponding to the target control volume as input, based on the secondary side mass equation and internal energy equation, dedicated mass conservation equations and energy conservation equations are established for the two regions respectively. The equation for the upper steam region focuses on describing the mass and energy changes of steam flow, pressure loss and a small amount of phase change (such as condensation), while the equation for the lower water region focuses on characterizing the heat absorption, evaporation and mass exchange process of water with the steam region, thereby reflecting the differentiated characteristics of the steam-water two phases.

[0125] During the solution phase, considering the rigidity of the equation system, the Gear implicit variable-order multi-step method is selected, combined with an adaptive integration step size adjustment strategy for joint solution. Simultaneously, through model simplification, millisecond-level real-time calculations are achieved, outputting parameters such as water level, steam pressure, water temperature, and steam dryness in the two regions for future time periods. Furthermore, the model possesses adaptive parameter correction capabilities, allowing online adjustment of key coefficients to adapt to changes in equipment status, ensuring prediction accuracy and reliability.

[0126] In these alternative embodiments, by dividing the steam-water region and establishing conservation equations separately, the differential characteristics of the two phases are accurately captured, avoiding the loss of accuracy in overall modeling. Jointly solving the equations can take into account the coupling relationships between regions. Combined with key parameter inputs, future thermal-hydraulic parameters can be accurately predicted, improving the simulation capability for complex phase change processes on the secondary side and providing precise data support for water level control, etc.

[0127] In one embodiment, the step of inputting the key operating parameters and the thermal-hydraulic parameters into an intelligent decision-making model, and making a decision through a pre-learned decision table in the intelligent decision-making model to determine the optimal adjustment action of the steam generator includes: A state space is constructed based on the key operating parameters and the thermal-hydraulic parameters. The value network of the intelligent decision-making model is used to evaluate different adjustment actions performed in the state space, and the reward values ​​corresponding to the different adjustment actions are obtained. The optimal adjustment action is determined based on the adjustment action corresponding to the maximum reward value.

[0128] Optionally, in this embodiment, the state space is a multi-dimensional parameter set constructed based on key operating parameters and thermo-hydraulic parameters, used to comprehensively describe the operating state of the steam generator at a certain moment. Different combinations of parameters form specific state points.

[0129] The reward value is an evaluation index of the effectiveness of different adjustment actions in the current state space by the value network of the intelligent decision-making model. It is used to quantify the merits of the actions. Its value is determined according to the contribution of the action to the operational goal (e.g., the reward value is positive when the water level is closer to the set value or when the thermal efficiency is improved, and negative when it deviates from the goal or causes risks).

[0130] Optionally, in one specific implementation of this application, key operating parameters and thermal-hydraulic parameters are first extracted, and these parameters are quantified and combined to construct a multi-dimensional state space, with each dimension corresponding to a parameter, which fully characterizes the current operating state of the steam generator.

[0131] Next, the intelligent decision-making model invokes a pre-trained value network. This network, based on a decision table learned from historical data, evaluates possible adjustment actions in the state space (such as adjusting the water supply valve opening or changing the circulation pump speed). By inputting the current state parameters, the network calculates the degree of deviation from the target value after each action is executed (such as water level deviation or energy consumption change), converting it into a reward value (positive reward for approaching the target, negative reward for deviation). Finally, the model compares the reward values ​​of all actions and selects the action with the maximum reward value as the optimal adjustment action.

[0132] In these alternative embodiments, by calculating the reward value and selecting the action corresponding to the maximum reward value as the optimal adjustment action, it is possible to ensure that the decision-making meets the goals of safe and efficient operation and can quickly adapt to changes in different working conditions.

[0133] In one embodiment, evaluating different adjustment actions performed in the state space through the value network of the intelligent decision-making model to obtain reward values ​​corresponding to the different adjustment actions includes: The reward function of the value network is used to evaluate the different adjustment actions performed in the state space, and the reward values ​​corresponding to the different adjustment actions are obtained. The reward function is determined based on at least one evaluation item, which includes at least one of the following: water level and pressure setpoint tracking reward item, adjustment action smoothness reward item, safety constraint violation penalty reward item, and resource utilization rate reward item.

[0134] Optionally, in this embodiment, the reward function is a mathematical function in the value network of the intelligent decision-making model used to quantify the merits of adjustment actions. Its core function is to transform the impact of adjustment actions on the operation of the steam generator into a calculable reward value. The sign and magnitude of the reward value directly reflect whether the action meets the operational objective and are the core basis for the model to select the optimal adjustment action.

[0135] The water level and pressure setpoint tracking reward is an evaluation item in the reward function used to assess the effectiveness of adjustment actions in controlling key parameters. Its calculation logic is as follows: compare the deviations of the actual water level and pressure of the steam generator after the action is executed with the preset setpoints. The smaller the deviation, the higher the reward value; if the actual values ​​perfectly match the setpoints, this item receives the highest reward. The core objective of this evaluation item is to ensure that the adjustment actions can accurately control key parameters such as water level and pressure, maintaining stable equipment operation.

[0136] The adjustment smoothness bonus is an evaluation item in the reward function used to assess the stability of the adjustment action execution process. It primarily scores the magnitude of changes in the adjustment action (such as the single adjustment amount of the water supply valve opening or the rate of change of the circulating pump speed). The smoother the action change and the less drastic the fluctuations, the higher the bonus value for this item; if the action involves frequent and large adjustments, the bonus value will decrease or even become negative. Its purpose is to reduce the impact on equipment caused by sudden changes in action and extend the equipment's service life.

[0137] The safety constraint violation penalty reward item is an evaluation item in the reward function used to avoid safety risks caused by adjustment actions; it is essentially a "punitive" indicator. When an adjustment action causes operating parameters to exceed safety thresholds (such as water level falling below the minimum safety line or pressure exceeding the warning value), this item will output a negative reward value (i.e., penalty), and the more severe the violation, the larger the absolute value of the negative reward. If no safety constraint is violated, this item may have zero or a small positive reward. It is a key evaluation item for ensuring the safe operation of the steam generator.

[0138] The resource utilization rate reward item is an evaluation item in the reward function used to assess the efficiency of adjustment actions in utilizing resources such as energy and working fluids. Its calculation is based on indicators including the thermal efficiency of the steam generator, feedwater utilization rate, and fuel consumption rate. If the adjustment action improves resource utilization efficiency (such as reducing steam waste and lowering energy consumption), the reward value for this item is higher; if it causes resource waste, the reward value will decrease. The core objective of this evaluation item is to promote adjustment actions that balance equipment stability and economical operation.

[0139] Optionally, in one specific implementation of this application, the state space design includes: real-time sensor-collected water level, pressure, flow rate, and temperature data (i.e., key operating parameters); a future N-step system state sequence predicted by the digital twin module (thermal and hydraulic parameters for future time periods); historical state data (time window T); and equipment state information including valve opening and pump status.

[0140] Action space design: The output is a continuous variable, including suggested changes in the opening of the feedwater regulating valve, the steam discharge valve, and the feedwater pump speed. It can also include discrete commands, such as feedwater pump start / stop switching and standby regulating valve activation, to address equipment state switching needs under extreme operating conditions.

[0141] Reward function design: (6) in: R tracking Water level and pressure setpoint tracking reward items; R smooth Adjust the motion smoothness reward item; R safety Safety constraint violation penalty and reward items; R efficiency Resource utilization rate incentive item.

[0142] R t This is the reward value; w1, w2, w3, and w4 are the weights corresponding to each reward item.

[0143] In these alternative embodiments, these steps integrate multi-dimensional evaluation items through a reward function, comprehensively measuring the effectiveness of the adjustment action: ensuring that water level and pressure conform to set values, balancing action smoothness and resource efficiency, and mitigating risks through safety penalties. The synergy of multiple evaluation items makes the reward value more accurate, enabling the model to select the optimal action that balances safety, stability, and economy, thus improving the reliability and practicality of intelligent decision-making.

[0144] In one embodiment, determining the optimal adjustment action based on the adjustment action corresponding to the maximum reward value includes: The adjustment action corresponding to the maximum reward value is subjected to a security check, and if the check passes, the adjustment action corresponding to the maximum reward value is determined as the optimal adjustment action.

[0145] Optionally, in one specific implementation of this application, the maximum reward value and its corresponding action are first selected from the reward values ​​corresponding to each adjustment action; then the safety verification module is started, and the action is substituted into the steam generator safety rule base (such as verifying whether the water level and pressure exceed the safety threshold after the action is executed); if the verification result shows that the action meets the safety constraints, it is directly determined as the optimal adjustment action; if it fails, the action corresponding to the second largest reward value is selected again for repeated verification until the optimal adjustment action that has passed the safety verification is output.

[0146] In these alternative embodiments, adding a safety check after selecting the action with the maximum reward value can eliminate actions that seem optimal but pose safety risks. Determining the optimal action only after the check passes ensures that the decision balances effectiveness and safety, avoids equipment malfunctions caused by adjustments, and improves decision reliability.

[0147] In one embodiment, after determining the optimal adjustment action of the steam generator by making a decision using a decision table pre-learned in the intelligent decision model, the method further includes: Based on the optimal adjustment action, operation guidance information is determined, including the control command corresponding to the optimal adjustment action and the basis for determining the optimal adjustment action; The operation guidance information is displayed.

[0148] Optionally, in this embodiment, the operation guidance information is specific information generated based on the optimal adjustment action to guide the operation and adjustment of the steam generator. It includes two core parts: first, the control command corresponding to the optimal adjustment action (such as adjusting the opening of a valve, changing the pump speed, etc.); second, the basis for determining the action (such as decision logic based on the current water level deviation, pressure status, and reward value evaluation). Its purpose is to enable operators to clearly understand the operation to be performed and the reasons behind it, so as to accurately execute the adjustment and understand the rationality of the decision.

[0149] Optionally, in one specific implementation of this application, after the intelligent decision-making model outputs the optimal adjustment action, it transmits it to the operation guidance generation module. This module includes an instruction conversion unit and a basis processing unit: the instruction conversion unit converts the optimal action into a specific control instruction (such as "adjust the water supply valve opening to 50%)"; the basis processing unit extracts decision-making process data (such as key parameter deviations and reward value calculation results) to form the basis for action determination. Subsequently, the module integrates the control instruction and the determination basis into operation guidance information and transmits it to the human-computer interaction unit for display in a visual interface (such as a pop-up window or dashboard), allowing operators to intuitively obtain the operation content and decision logic while retaining their final operation decision-making authority.

[0150] In these optional embodiments, generating and displaying operational guidance information containing control commands and decision-making basis allows operators to quickly understand the operation content and reasons, and accurately execute adjustments. At the same time, it preserves human decision-making authority, balancing the efficiency of intelligent decision-making with the safety of manual control, thereby improving operational accuracy and reliability.

[0151] Secondly, such as Figure 4 As shown, this application also provides a method for training an intelligent decision-making model, which specifically includes the following steps: S401, Obtain a training sample set, which includes sample key operating parameters, sample thermal-hydraulic parameters, and adjustment action labels of the nuclear power plant; the sample thermal-hydraulic parameters are the predicted thermal-hydraulic parameters of the nuclear power plant's steam generator in the future time period based on the sample key operating parameters. S402, the key operating parameters of the sample and the thermal-hydraulic parameters of the sample are input into the intelligent decision-making model to be trained, and the decision is made through the decision table in the intelligent decision-making model to be trained to determine the predicted adjustment action of the steam generator; S403, Calculate the loss value based on the predicted adjustment action and the corresponding adjustment action label; S404, if the loss value does not meet the preset conditions, the decision table is iteratively updated until the loss value meets the preset conditions, and a trained intelligent decision model is obtained. The decision table includes a mapping relationship between different decision actions and reward values ​​for the nuclear power plant and the steam generator under different operating conditions. The operating conditions are determined based on the key operating parameters of the nuclear power plant and the thermal-hydraulic parameters of the steam generator. The reward values ​​are determined based on the safety levels corresponding to the liquid level and pressure of the steam generator.

[0152] The explanation of the relevant terms can be found in the description of the foregoing embodiments, and will not be repeated here.

[0153] Optionally, in one specific implementation of this application, the training sample set is constructed from historical operating data, wherein the sample thermal-hydraulic parameters are generated by the thermal-hydraulic prediction model based on the key operating parameters of the samples and associated with adjustment action labels (such as expert operation records); the key operating parameters of the samples and the thermal-hydraulic parameters are fused into working state features, which are input into the model to be trained. The model queries the candidate actions and reward values ​​corresponding to the current state through the decision table, and outputs the predicted adjustment actions based on the principle of maximizing reward values.

[0154] Subsequently, the cross-entropy loss function can be used to calculate the difference between the predicted action and the label to obtain the loss value. If the loss does not meet the convergence threshold, the state-action reward value mapping in the decision table can be iteratively updated using the policy gradient method until the loss converges, thus obtaining a trained intelligent decision model that can output decisions that meet both safety and efficiency requirements under new working conditions.

[0155] In this embodiment, key operating parameters of the nuclear power plant are acquired to ensure accurate perception of the current operating status of the steam generator. Based on this, the thermal-hydraulic parameters of the steam generator for future periods are predicted using these key operating parameters. This prediction process allows decision-making to move beyond passive responses to current conditions and instead provides a forward-looking understanding of future conditions, reducing decision-making biases caused by the inability to adapt to dynamic changes in operating conditions. Finally, the key operating parameters and the predicted thermal-hydraulic parameters are jointly input into an intelligent decision-making model, and the optimal adjustment action is determined through a pre-learned decision table. The decision table constructs a "decision action - reward value" mapping relationship based on different operating states of the nuclear power plant and the steam generator, with the reward value determined by the safety of the liquid level and pressure, ensuring that the decision action always prioritizes the safe and stable operation of the steam generator. This process, through multi-parameter fusion to define operating conditions and safety-oriented reward value constraints on the selection of adjustment actions, enables the intelligent decision-making model to adapt to different complex operating conditions. The output adjustment actions are more closely aligned with actual operating requirements, significantly reducing decision-making bias and thus improving the reliability of steam generator decision-making.

[0156] In one embodiment, the step of inputting the key operating parameters and the thermal-hydraulic parameters of the sample into the intelligent decision-making model to be trained, and making a decision through the decision table in the intelligent decision-making model to determine the predicted adjustment action of the steam generator, includes: Based on the key operating parameters and thermo-hydraulic parameters of the sample, the sample state space is constructed. The value network of the intelligent decision-making model to be trained is used to evaluate the different adjustment actions performed in the sample state space, and the reward values ​​corresponding to the different adjustment actions are obtained. The prediction adjustment action is determined based on the adjustment action corresponding to the maximum reward value.

[0157] In one embodiment, the step of evaluating different adjustment actions performed in the sample state space through the value network of the intelligent decision-making model to be trained, and obtaining reward values ​​corresponding to the different adjustment actions, includes: The reward function of the value network is used to evaluate the different adjustment actions performed in the state space, and the reward values ​​corresponding to the different adjustment actions are obtained. The reward function is determined based on at least one evaluation item, which includes at least one of the following: water level and pressure setpoint tracking reward item, adjustment action smoothness reward item, safety constraint violation penalty reward item, and resource utilization rate reward item.

[0158] It should be noted that the explanation of the relevant terms can be found in the description of the foregoing embodiments, and will not be repeated here.

[0159] It should be noted that the various optional implementation methods described in the embodiments of this application can be combined with each other or implemented individually without conflict, and the embodiments of this application do not limit this.

[0160] To facilitate understanding of the decision-making method for the steam generator and the training method for the intelligent decision-making model provided in the above embodiments, the following describes the decision-making method for the steam generator and the training method for the intelligent decision-making model using a specific scenario embodiment.

[0161] Optionally, in this embodiment, a millisecond-level response device for intelligent auxiliary decision-making in steam generator water level control of a nuclear power plant is provided. The core of its technical solution lies in constructing an auxiliary control system integrating real-time sensing, intelligent evaluation, and decision support. This device is implemented through an edge computing unit deployed at the equipment site, forming a millisecond-level auxiliary decision-making closed loop that works collaboratively with the upper-level DCS system.

[0162] 1. System Architecture This device adopts a three-layer distributed auxiliary decision-making architecture: Sensing layer: Composed of a high-speed sensor array and a data acquisition unit, responsible for real-time acquisition of multi-source operating parameters of the steam generator system.

[0163] Decision-making layer: Deployed within an edge computing unit, it includes a digital twin prediction module and an intelligent assessment and decision-making module. The edge computing unit is deployed near the steam generator system to form an independent millisecond-level auxiliary decision-making closed loop. The edge computing unit communicates asynchronously with the power plant's distributed control system (DCS) via an industrial bus, uploading decision-making suggestions to the DCS and receiving system status and setpoint information from the DCS. Its auxiliary decision-making closed loop operates independently of the DCS's main control cycle.

[0164] Operation recommendation layer: Composed of instruction encoding unit and human-machine interaction unit, responsible for providing operation guidance to DCS system or operator.

[0165] This device, through a clearly defined data flow, transmits decision-making information from sensors via the data acquisition unit to the edge computing unit, and finally to the DCS system or operator. It also achieves asynchronous communication and data exchange with the power plant's DCS system via an industrial bus. The system also includes an independent safety verification unit to ensure the safety and rationality of all decision recommendations. This architecture guarantees millisecond-level real-time response while ensuring the operator's central role in the decision-making loop, achieving an intelligent human-machine collaborative operation mode.

[0166] 2. Perception Layer (Environmental Perception System) This device uses a high-speed sensor array (including but not limited to high-frequency pressure sensors, differential pressure transmitters, temperature sensors, flow meters, etc.) deployed on the steam generator system as a "LiDAR" and "camera". It synchronously and in real time collects multi-source operating parameters, including the primary side working pressure and inlet and outlet temperatures of the steam generator, and the secondary side water level, steam pressure, steam flow, feedwater flow, feedwater temperature, etc., at a sampling frequency of not less than 1kHz, so as to achieve millisecond-level global situational awareness of the system status.

[0167] The data acquisition unit employs synchronous sampling technology to ensure the time consistency of all parameters, providing an accurate time-series data foundation for subsequent analysis. All sensor signals are directly connected to the edge computing unit deployed in the field via hardwiring, minimizing signal transmission delay.

[0168] 3. Decision-making level (central decision-making brain) The decision-making layer is the intelligent core of this invention. It is deployed within an edge computing unit and contains two core modules: 3.1 Digital Twin Prediction Module (High-Precision Map and Prediction) This module runs a real-time, high-precision physical model of the steam generator, much like a high-precision map for an autonomous driving system. It can make millisecond-level advance predictions of changes in the system within the next few seconds based on the current state, and anticipate the generation and evolution of "false water levels".

[0169] This module establishes a high-precision real-time physical model of the steam generator, and its basic principle is as follows: like Figure 2 As shown, the secondary fluid of the steam generator relies on the density difference between the descending and ascending channels to form a driving pressure head, thus creating a natural circulation loop.

[0170] To accurately establish a high-precision real-time physical model, the two-loop system is divided into 5 branches (e.g., Figure 3 (as shown) Branch 1: Steam space between the top of the steam-water separator and the outlet pipe of the steam generator Branch 2: The water supply chamber section outside the separator, between the top of the separator and the top of the tube bundle sleeve. Branch 3: The rising channel from the top of the separator to the top of the inner sleeve Branch 4 and Branch 5: Represent the descending channel and the ascending channel, respectively. Each branch consists of several control volumes, which are used to describe changes in pressure, temperature, specific enthalpy, and mass, while the flow channels are used to describe changes in mass flow rate and various head losses in the circulation loop.

[0171] The mathematical model is as follows: Primary side governing equations (single-phase incompressible fluid assumption): (1) In the above formula, Let i be the mass and internal energy of the primary control volume. . , and Let be the density, volume, and internal energy of the primary control volume i, respectively. and These are the enthalpy values ​​of the fluid entering and leaving the control volume, respectively. For primary side flow, To transfer the heat power to the secondary side, This refers to the heat capacity of the tube sheet and lower head metal.

[0172] Secondary side control equations: (2) (3) (4) in, , and These represent the mass, internal energy, and flow rate of the secondary control unit, respectively. The subscripts *in* and *out* indicate entry into and exit from the control unit. The subscript *ex* indicates exit from the steam generator body, such as feedwater and steam connections. Generalized source term. This includes heat transferred through the sleeve wall in the descending and ascending channels, heat transferred by the thermal components, heat transferred by thermal diffusion, and heat transferred by steam condensation upon encountering the wall. L and S represent the equivalent length and cross-sectional area of ​​the flow channel k. The pressure is represented by the subscripts f, s, g, and a, which respectively indicate the frictional pressure drop, local pressure drop, heavy pressure drop, and acceleration pressure drop occurring in the flow channel.

[0173] The physical properties such as specific volume, pressure, and specific enthalpy within the body are determined based on the equation of state: v=v(P,h),ρ=ρ(P,h),T=T(P,h)(5) For a control body with a water level, mass and energy conservation equations need to be established for the upper steam region and the lower water region respectively.

[0174] This model has several key technical features: it adopts a multi-region partitioning strategy and models differently based on physical characteristics; to deal with rigid equations, it uses the Gear implicit variable-order multi-step method for solving and adaptively adjusts the integration step size; through model simplification and code optimization, it achieves millisecond-level real-time solutions and can effectively predict the dynamic changes in water level, pressure, and temperature; in addition, the model also has parameter adaptive capabilities and can correct model parameters online to adapt to changes in equipment status, thereby ensuring prediction accuracy and reliability.

[0175] 3.2 Intelligent Assessment and Decision-Making Module This module runs an intelligent evaluation model trained using deep reinforcement learning. Its core function is to simulate advanced decision-making logic, deeply integrate real-time sensing data with digital twin prediction results, and achieve a fundamental understanding of phenomena such as "false water levels" and intelligent identification of disturbance types. The model can generate optimized operational suggestions within milliseconds, and its decision-making objective is to maintain stable water levels across all operating conditions, not just limited to setpoint tracking.

[0176] This module employs an intelligent control algorithm based on deep reinforcement learning (DRL), specifically implemented as follows: 1) State-space design: The system status includes real-time sensor data on water level, pressure, flow rate, and temperature; a sequence of system statuses for the next N steps predicted by the digital twin module; historical status data (time window T); and equipment status information including valve opening and pump status.

[0177] 2) Motion space design: The output is a continuous variable, including suggested values ​​for changes in the opening degree of the feedwater regulating valve, changes in the opening degree of the steam discharge valve, and adjustments to the feedwater pump speed.

[0178] 3) Reward function design:

[0179] in: R tracking Rewards for tracking water level and pressure setpoints; R smooth : Rewards for smooth control of actions; R safety Penalties for violating security constraints; R efficiency Rewards based on economic indicators.

[0180] 4) Network structure: The Actor-Critic framework is adopted. The Actor network (action network) receives the state input and outputs the mean vector of the proposed action, while the Critic network (evaluation network) evaluates the state-action pair by Q-value (reward value). A target network is introduced to improve training stability.

[0181] 5) Training process: The behavior cloning model is pre-trained based on historical operating data; offline reinforcement learning is performed in a digital twin environment; online safety fine-tuning is carried out using constraint strategy optimization; and the model adapts to equipment aging and changes in operating conditions through a continuous learning mechanism.

[0182] 6) Reasoning process: The process includes real-time state coding and preprocessing, forward propagation of the policy network to generate preliminary action suggestions, a safety verification module to verify feasibility, and finally outputting verified decision suggestions to the operation recommendation layer. When false water level initiation characteristics are identified, the abnormal change trend of water level measurement values ​​is analyzed, and targeted operation suggestions are generated based on changes in steam flow and feedwater flow.

[0183] 4. Operation Recommendation Layer The operational suggestions generated by the decision-making layer are transmitted to the operational recommendation layer, which consists of an instruction encoding unit and a human-machine interaction unit. The instruction encoding unit converts the suggestions into signal formats or operational instructions recognizable by the DCS, while the human-machine interaction unit intuitively displays the decision suggestions and their basis to the operator. The entire auxiliary decision-making process is completed in the edge computing unit deployed at the equipment site, forming a millisecond-level auxiliary decision-making closed loop that works in collaboration with the upper-level DCS system. This ensures both real-time response and maintains the operator's final decision-making authority in the decision-making loop.

[0184] 5. System Integration and Interfaces The entire system is integrated into an industrial-grade edge computing platform and has the following characteristics: 1) Hardware platform: Industrial computers using x86 or Advanced RISC Machines (ARM) architecture, equipped with Field-Programmable Gate Array (FPGA) acceleration cards to meet millisecond-level computing requirements; 2) Real-time operating system: guarantees deterministic response performance; 3) Communication interface: Hardwired input / output (I / O): Used for connecting sensors and actuators. Industrial Ethernet: Communicating with DCS systems Time synchronization: Supports the Institute of Electrical and Electronics Engineers (IEEE) 1588 precision time protocol to ensure timing consistency across all nodes. 4) Data Flow: Sensor data → Data acquisition card → Shared memory area; The digital twin module reads real-time data and performs millisecond-level predictions of system status; The intelligent assessment module integrates real-time data and prediction results to generate decision recommendations; Operation suggestions are sent to the DCS system or operator interface via the output interface.

[0185] 6. Computer-based implementation: The computer device of the present invention includes a memory and a processor. The memory stores a computer program, and the processor executes all the steps of the above-mentioned intelligent auxiliary decision-making for the water level of the steam generator by calling the computer program stored in the memory. The processor can be a general-purpose or special-purpose processor such as a CPU or FPGA, and the software module can be stored in RAM, hard disk, or other computer-readable storage media.

[0186] System workflow: 1) System power-on initialization, loading digital twin model and intelligent evaluation model; 2) Collect sensor data in real time, and perform data preprocessing and quality verification; 3) The digital twin module performs state prediction and outputs the system state prediction results for the next N steps (i.e., the thermal and hydraulic parameters of the steam generator in the future time period). 4) The intelligent evaluation module combines the current state and prediction results to generate operation suggestions (i.e., the optimal adjustment action). 5) The security verification module verifies the rationality and security of the operation suggestions; 6) Display decision suggestions to the operator through the human-machine interface, or convert them into DCS executable instructions; 7) Repeat steps 2-6 to form a millisecond-level auxiliary decision-making closed loop.

[0187] Through the above-mentioned technical solution, this invention realizes intelligent millisecond-level auxiliary decision-making for steam generator water level, effectively improving the operator's ability to recognize and handle complex dynamic processes such as "false water level", and significantly enhancing the safety and reliability of nuclear power plant operation.

[0188] In these alternative embodiments, the millisecond-level response device for intelligent auxiliary decision-making in steam generator water level control of nuclear power plants provided by the present invention brings the following significant technical effects: This system achieves a fundamental improvement in millisecond-level real-time response and control quality: through the synergistic integration of high-speed sensing, edge computing, and high-frequency actuators, it shortens the system response time and reduces the control lag problem caused by timing mismatch in traditional control systems. The system can quickly track and suppress fluctuations caused by "false water levels," significantly improving the accuracy and stability of water level control, effectively reducing overshoot, and shortening settling time, thereby ensuring the safe and stable operation of the steam generator under transient and disturbance conditions.

[0189] It provides intelligent decision support with forward-looking and adaptive capabilities: by introducing a joint decision-making mechanism of "digital twin + reinforcement learning," the system can deeply integrate multi-source sensing information and real-time simulation prediction to achieve accurate identification and intelligent compensation for complex phenomena such as "false water levels." This capability greatly enhances the understanding and decision-making level of nonlinear and time-varying operating conditions, providing operators with advanced and reliable operational guidance and suggestions, and significantly reducing operational risks caused by misjudgment or delayed response.

[0190] The system's overall safety and operational reliability are enhanced: As an intelligent auxiliary decision-making device deployed on-site, it can achieve millisecond-level closed-loop control independently of the DCS, and can also operate collaboratively with the upper-level system, effectively reducing the cascading risks caused by network latency or central system failures. The system's rapid disturbance suppression and abnormal operating condition preprocessing capabilities add a strong technical defense line to nuclear power plant operations, improving the power plant's defense-in-depth level.

[0191] The system improves the automation level and economy of power plants: stable and precise water level control reduces mechanical wear on actuators and related equipment, extending equipment lifespan. Simultaneously, by providing clear and reliable decision support, the system reduces over-reliance on operator experience, decreases the probability of human error, and enhances the overall level of automation, providing a technical foundation for power plants to achieve more efficient, safe, and stable operation.

[0192] Figure 5 A schematic diagram of the decision-making device for a steam generator provided in another embodiment of this application is shown. For ease of explanation, only the parts related to the embodiments of this application are shown.

[0193] Reference Figure 5 The decision-making device of the steam generator may include: The first acquisition module 501 is used to acquire key operating parameters of the nuclear power plant; The prediction module 502 is used to predict the thermal-hydraulic parameters of the steam generator of the nuclear power plant in the future time period based on the key operating parameters. The first decision module 503 is used to input the key operating parameters and the thermal-hydraulic parameters into the intelligent decision model, and make decisions through the pre-learned decision table in the intelligent decision model to determine the optimal adjustment action of the steam generator. The decision table includes a mapping relationship between different decision actions and reward values ​​for the nuclear power plant and the steam generator under different operating conditions. The operating conditions are determined based on the key operating parameters of the nuclear power plant and the thermal-hydraulic parameters of the steam generator. The reward values ​​are determined based on the safety levels corresponding to the liquid level and pressure of the steam generator.

[0194] Figure 6 A schematic diagram of the structure of a training device for an intelligent decision-making model provided in another embodiment of this application is shown. For ease of explanation, only the parts related to the embodiments of this application are shown.

[0195] Reference Figure 6 The training device for the intelligent decision-making model may include: The second acquisition module 601 is used to acquire a training sample set, which includes key operating parameters, thermal-hydraulic parameters, and adjustment action labels. The thermal-hydraulic parameters are the predicted thermal-hydraulic parameters of the steam generator of the nuclear power plant in the future time period based on the key operating parameters. The second decision module 602 is used to input the key operating parameters of the sample and the thermal-hydraulic parameters of the sample into the intelligent decision model to be trained, and to make a decision through the decision table in the intelligent decision model to be trained, and determine the predicted adjustment action of the steam generator. Calculation module 603 is used to calculate the loss value based on the predicted adjustment action and the corresponding adjustment action label; Training module 604 is used to iteratively update the decision table when the loss value does not meet the preset conditions, until the loss value meets the preset conditions, so as to obtain a trained intelligent decision model. The decision table includes a mapping relationship between different decision actions and reward values ​​for the nuclear power plant and the steam generator under different operating conditions. The operating conditions are determined based on the key operating parameters of the nuclear power plant and the thermal-hydraulic parameters of the steam generator. The reward values ​​are determined based on the safety levels corresponding to the liquid level and pressure of the steam generator.

[0196] It should be noted that the information interaction and execution process between the above-mentioned devices / units are based on the same concept as the method embodiments of this application, and are devices corresponding to the above-mentioned methods. All implementation methods in the above-mentioned method embodiments are applicable to the embodiments of this device. For details on its specific functions and the technical effects it brings, please refer to the method embodiment section, which will not be repeated here.

[0197] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0198] Figure 7 A schematic diagram of the hardware structure of the electronic device provided in an embodiment of this application is shown.

[0199] The device may include a processor 701 and a memory 702 storing program instructions.

[0200] When processor 701 executes the program, it implements the steps in any of the above method embodiments.

[0201] For example, the program can be divided into one or more modules / units, one or more of which are stored in memory 702 and executed by processor 701 to complete this application. The one or more modules / units can be a series of program instruction segments capable of performing a specific function, which describe the execution process of the program in the device.

[0202] Specifically, the processor 701 may include a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of this application.

[0203] Memory 702 may include mass storage for data or instructions. For example, and not limitingly, memory 702 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 702 may include removable or non-removable (or fixed) media. Where appropriate, memory 702 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 702 is non-volatile solid-state memory.

[0204] Memory may include read-only memory (ROM), random access memory (RAM), disk storage media devices, optical storage media devices, flash memory devices, and electrical, optical, or other physical / tangible memory storage devices. Therefore, typically, memory includes one or more tangible (non-transitory) readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the methods according to one aspect of this disclosure.

[0205] The processor 701 implements any of the methods described in the above embodiments by reading and executing program instructions stored in the memory 702.

[0206] In one example, the electronic device may also include a communication interface 703 and a bus 710. The processor 701, memory 702, and communication interface 703 are connected via the bus 710 and communicate with each other.

[0207] The communication interface 703 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of this application.

[0208] Bus 710 includes hardware, software, or both, that couples components of an online data traffic metering device together. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 710 may include one or more buses. Although specific buses are described and illustrated in embodiments of this application, any suitable bus or interconnect is contemplated herein.

[0209] Furthermore, in conjunction with the methods in the above embodiments, this application embodiment can provide a storage medium for implementation. This storage medium stores program instructions; when these program instructions are executed by a processor, they implement any of the methods in the above embodiments.

[0210] This application also provides a chip, which includes a processor and a communication interface. The communication interface and the processor are coupled. The processor is used to run programs or instructions to implement the various processes of the above method embodiments and achieve the same technical effect. To avoid repetition, it will not be described again here.

[0211] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0212] This application provides a computer program product, which is stored in a storage medium and executed by at least one processor to implement the various processes of the above method embodiments and achieve the same technical effects. To avoid repetition, it will not be described again here.

[0213] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0214] The functional modules shown in the above block diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on machine-readable media or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable media" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer grids such as the Internet, intranets, etc.

[0215] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0216] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to create a machine such that these instructions, executable via the processor of the computer or other programmable data processing apparatus, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0217] The above are merely specific embodiments of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A decision method of a steam generator, characterized by, The method comprises: obtaining key operation parameters of a nuclear power plant; based on the key operation parameters, predicting thermal-hydraulic parameters of a steam generator of the nuclear power plant in a future time period; inputting the key operation parameters and the thermal-hydraulic parameters into an intelligent decision-making model, making decisions through a pre-learned decision table in the intelligent decision-making model, and determining optimal adjustment actions of the steam generator; wherein the decision table comprises a mapping relationship between different decision actions and reward values of the nuclear power plant and the steam generator in different working states, the working states are determined according to the key operation parameters of the nuclear power plant and the thermal-hydraulic parameters of the steam generator, and the reward values are determined according to safety degrees corresponding to liquid level heights and pressures of the steam generator, respectively.

2. The method of claim 1, wherein, The method comprises: constructing a digital twin model of the steam generator; inputting the key operation parameters into the digital twin model, and performing joint calculation based on the key operation parameters through the digital twin model to obtain the thermal-hydraulic parameters of the steam generator in the future time period.

3. The method of claim 2, wherein, The steam generator comprises a primary side loop and a secondary side loop. The method comprises: constructing a primary side control equation according to a heat balance relationship of the primary side loop; the primary side control equation is used to solve thermal-hydraulic parameters corresponding to the primary side loop; constructing a first physical model corresponding to the primary side loop according to the primary side control equation; the first physical model comprises a network composed of a plurality of primary side control bodies and a primary side flow channel connecting the primary side control bodies; constructing a secondary side control equation according to a mass conservation relationship, an energy conservation relationship and a momentum conservation relationship of the secondary side loop; the secondary side control equation is used to solve thermal-hydraulic parameters corresponding to the secondary side loop; constructing a second physical model corresponding to the secondary side loop according to the secondary side control equation; the second physical model comprises a network composed of a plurality of secondary side control bodies and a secondary side flow channel connecting the secondary side control bodies; constructing the digital twin model according to the first physical model and the second physical model.

4. The method of claim 3, wherein, The method comprises: substituting the key operation parameters as boundary conditions and initial conditions into the primary side control equation and the secondary side control equation to obtain the thermal-hydraulic parameters of the steam generator in the future time period.

5. The method of claim 3, wherein, The method comprises: calculating a first product according to mass and a first derivative of an i-th primary side control body; the first derivative is a derivative of internal energy of the i-th primary side control body with respect to time; calculating a second product according to a difference between fluid enthalpy values entering and leaving the i-th primary side control body and a flow rate of the primary side loop; calculating a first difference of the second product and a thermal power delivered to the secondary side circuit; calculating a first sum of the first difference and a heat capacity of a tube sheet and a lower head metal of the steam generator; establishing an equation of the first product and the first sum, to obtain the primary side control equation.

6. The method of claim 3, wherein, the secondary side control equation comprises a mass equation, an internal energy equation and a flow equation; the secondary side control equation is constructed according to a mass conservation relation, an energy conservation relation and a momentum conservation relation of the secondary side circuit, comprising: determining the mass equation according to a flow rate of a secondary side flow passage entering and leaving the secondary side control body, and a flow rate exchanged between the steam generator body and an external boundary; determining the internal energy equation according to a flow rate of the secondary side flow passage entering and leaving the secondary side control body, a fluid enthalpy value entering and leaving the secondary side control body, a flow rate exchanged between the steam generator body and the external boundary, a fluid enthalpy value exchanged between the steam generator body and the external boundary, an internal energy of a heat transfer tube metal of the steam generator, and a generalized heat; the generalized heat is used to represent heat exchange of a non-core heat transfer process in the secondary side control body; determining the flow equation according to an equivalent length and a cross-sectional area of the secondary side flow passage, pressures of two secondary side control bodies connected with the secondary side flow passage, a friction pressure drop, a local pressure drop, a reposition pressure drop and an acceleration pressure drop in the secondary side flow passage.

7. The method of claim 3, wherein, the secondary side control equation comprises a mass equation and an internal energy equation; the key operation parameters are input into the digital twin model, and the digital twin model is used to jointly calculate based on the key operation parameters, to obtain thermal-hydraulic parameters of the steam generator in a future time period, comprising: for a target secondary side control body with a water level in the plurality of secondary side control bodies, the target secondary side control body is divided into an upper steam region and a lower water region; based on the key operation parameters corresponding to the target secondary side control body, mass conservation equations and energy conservation equations are respectively established for the upper steam region and the lower water region through the mass equation and the internal energy equation; the mass conservation equations and the energy conservation equations are used to describe differential thermal-hydraulic characteristics of a steam-water two-phase region; by jointly solving the mass conservation equations and the energy conservation equations, thermal-hydraulic parameters corresponding to the differential thermal-hydraulic characteristics of the steam-water two-phase region of the steam generator in the future time period are obtained.

8. The method of claim 6, wherein, determining the mass equation according to a flow rate of a secondary side flow passage entering and leaving the secondary side control body, and a flow rate exchanged between the steam generator body and an external boundary, comprising: calculating a second sum of all flow rates of the secondary side flow passages entering the i th secondary side control body; calculating a third sum of all flow rates of the secondary side flow passages leaving the i th secondary side control body; calculating a second difference between the second sum and the third sum; calculating a third difference between the second difference and the flow rate exchanged between the steam generator body and the external boundary; establishing an equation between the second derivative and the third difference, to obtain the mass equation; the second derivative is the derivative of the mass of the i-th secondary side control volume with respect to time.

9. The method of claim 6, wherein, The internal energy equation is determined according to the flow rate of the secondary side flow passage into and out of the secondary side control volume, the fluid enthalpy value of the secondary side flow passage into and out of the secondary side control volume, the flow rate exchanged between the steam generator body and the external boundary, the fluid enthalpy value exchanged between the steam generator body and the external boundary, the internal energy of the heat transfer tube metal of the steam generator, and the generalized heat, and comprises: a fourth sum of the product between the flow rate of all the secondary side flow passages into the i-th secondary side control volume and the fluid enthalpy value of all the secondary side flow passages into the i-th secondary side control volume is calculated; a fifth sum of the product between the flow rate of all the secondary side flow passages out of the i-th secondary side control volume and the fluid enthalpy value of all the secondary side flow passages out of the i-th secondary side control volume is calculated; a fourth difference between the fourth sum and the fifth sum is calculated; a sixth sum of the product between the flow rate exchanged between the steam generator body and the external boundary and the fluid enthalpy value exchanged between the steam generator body and the external boundary is calculated; a fifth difference between the fourth difference and the sixth sum is calculated; a seventh sum between the fifth difference, the internal energy of the heat transfer tube metal of the steam generator, and the generalized heat is calculated; an equation between the seventh sum and the third derivative is established, to obtain the internal energy equation; the third derivative is the derivative of the internal energy of the i-th secondary side control volume with respect to time.

10. The method of claim 6, wherein, The generalized heat comprises at least one of the following: heat transferred through the sleeve wall surface of the downcomer and the upcomer of the steam generator, heat transferred by the thermal member of the steam generator, heat transferred by thermal diffusion of the steam generator, and heat transferred by condensation of steam encountering the wall surface of the steam generator.

11. The method of claim 6, wherein, The flow equation is determined according to the equivalent length and the cross-sectional area of the secondary side flow passage, the pressure of the two secondary side control volumes connected with the secondary side flow passage, the frictional pressure drop, the local pressure drop, the gravity pressure drop, and the acceleration pressure drop in the secondary side flow passage, and comprises: a sum between the frictional pressure drop, the local pressure drop, the gravity pressure drop, and the acceleration pressure drop in the secondary side flow passage is determined as the resistance loss; a seventh difference between the sixth difference and the resistance loss is determined as the net pressure difference of the secondary side flow passage; the sixth difference is the difference between the pressures of the two secondary side control volumes connected with the secondary side flow passage; a ratio between the equivalent length and the cross-sectional area of the secondary side flow passage is determined as the proportional coefficient of the secondary side flow passage; an equation between the fourth derivative and the third product is established, to obtain the flow equation; the fourth derivative is the derivative of the flow rate of the secondary side flow passage with respect to time; the third product is the product between the proportional coefficient and the net pressure difference.

12. The method of claim 3, wherein, The second physical model corresponding to the secondary side loop is constructed according to the secondary side control equation, and comprises: The secondary side circuit is divided into a plurality of physical branches, including: a first branch for characterizing a steam space between a top of a steam-water separator and a steam outlet connection; a second branch for characterizing a portion of a feedwater cavity between the top of the separator and a top of a tube bundle sleeve; a third branch for characterizing an ascending channel between the top of the separator and a top of an inner sleeve; a fourth branch for characterizing a descending channel; and a fifth branch for characterizing the ascending channel; The second physical model is constructed based on the plurality of physical branches and the secondary side control equations, wherein the secondary side control volumes are used to describe changes in pressure, temperature, specific enthalpy, and mass of the working fluid in each of the physical branches, and the secondary side flow paths are used to describe changes in mass flow of the working fluid and head losses of the circulating loop connecting the control volumes.

13. The method of claim 1, wherein, The method further includes: inputting the key operating parameters and the thermal-hydraulic parameters into an intelligent decision model, making a decision through a pre-learned decision table in the intelligent decision model, and determining an optimal adjustment action of the steam generator. According to the key operating parameters and the thermal-hydraulic parameters, a state space is constructed. The method further includes: evaluating different adjustment actions performed in the state space through a value network of the intelligent decision model to obtain reward values corresponding to the different adjustment actions. According to an adjustment action corresponding to a maximum reward value, the optimal adjustment action is determined.

14. The method of claim 13, wherein, The method further includes: evaluating different adjustment actions performed in the state space through a reward function of the value network to obtain reward values corresponding to the different adjustment actions. The reward function is determined according to at least one evaluation item, and the at least one evaluation item includes at least one of a water level and pressure set value tracking reward item, an adjustment action smoothness reward item, a safety constraint violation penalty reward item, and a resource utilization rate reward item. The method further includes: performing a safety check on the adjustment action corresponding to the maximum reward value, and determining the adjustment action corresponding to the maximum reward value as the optimal adjustment action if the check is passed.

15. The method of claim 13, wherein, After the optimal adjustment action of the steam generator is determined by making a decision through the pre-learned decision table in the intelligent decision model, the method further includes: According to the optimal adjustment action, operation guidance information is determined, the operation guidance information including a control instruction corresponding to the optimal adjustment action and a determination basis for determining the optimal adjustment action.

16. The method of claim 1, wherein, The operation guidance information is displayed. The method includes: A training sample set is obtained, the training sample set including sample key operating parameters, sample thermal-hydraulic parameters, and adjustment action labels of a nuclear power plant; the sample thermal-hydraulic parameters being thermal-hydraulic parameters of the steam generator of the nuclear power plant in a future time period predicted based on the sample key operating parameters.

17. A method for training an intelligent decision model, the method comprising: ​ ​ input the sample key operation parameter and the sample thermal-hydraulic parameter to the intelligent decision-making model to be trained, make a decision through a decision table in the intelligent decision-making model to be trained, and determine a predicted adjustment action of the steam generator; calculate a loss value according to the predicted adjustment action and a corresponding adjustment action label; in a case where the loss value does not satisfy a preset condition, iteratively update the decision table until the loss value satisfies the preset condition, and obtain a trained intelligent decision-making model; wherein the decision table includes a mapping relationship between different decision actions and reward values of the nuclear power plant and the steam generator in different working states, the working state is determined according to the key operation parameter of the nuclear power plant and the thermal-hydraulic parameter of the steam generator, and the reward value is determined according to the safety degree corresponding to the liquid level height and the pressure of the steam generator respectively.

18. The method of claim 17, wherein, The method comprises the following steps: construct a sample state space according to the sample key operation parameter and the sample thermal-hydraulic parameter; evaluate different adjustment actions performed in the sample state space through a value network of the intelligent decision-making model to be trained, and obtain reward values corresponding to the different adjustment actions; determine the predicted adjustment action according to an adjustment action corresponding to the maximum reward value.

19. The method of claim 18, wherein, The method comprises the following steps: evaluate different adjustment actions performed in the state space through a reward function of the value network, and obtain reward values corresponding to the different adjustment actions; wherein the reward function is determined according to at least one evaluation item, and the at least one evaluation item includes at least one of the following: a water level and pressure set value tracking reward item, an adjustment action smoothness reward item, a safety constraint violation penalty reward item, and a resource utilization rate reward item.

20. A decision device for a steam generator, characterized by The device comprises: a first acquisition module configured to acquire a key operation parameter of a nuclear power plant; a prediction module configured to predict a thermal-hydraulic parameter of a steam generator of the nuclear power plant in a future time period based on the key operation parameter; a first decision-making module configured to input the key operation parameter and the thermal-hydraulic parameter to an intelligent decision-making model, make a decision through a pre-learned decision table in the intelligent decision-making model, and determine an optimal adjustment action of the steam generator; wherein the decision table includes a mapping relationship between different decision actions and reward values of the nuclear power plant and the steam generator in different working states, the working state is determined according to the key operation parameter of the nuclear power plant and the thermal-hydraulic parameter of the steam generator, and the reward value is determined according to the safety degree corresponding to the liquid level height and the pressure of the steam generator respectively.

21. A training device for an intelligent decision-making model, characterized in that, The device comprises: The second obtaining module is configured to obtain a training sample set, wherein the training sample set comprises sample key operation parameters, sample thermal-hydraulic parameters and adjustment action labels of a nuclear power plant; the sample thermal-hydraulic parameters are thermal-hydraulic parameters of a steam generator of the nuclear power plant in a future time period, which are predicted based on the sample key operation parameters; The second decision module is configured to input the sample key operation parameters and the sample thermal-hydraulic parameters into an intelligent decision model to be trained, and determine a predicted adjustment action of the steam generator by decision-making in the intelligent decision model to be trained. The calculation module is configured to calculate a loss value according to the predicted adjustment action and a corresponding adjustment action label. The training module is configured to iteratively update the decision table until the loss value meets a preset condition, so as to obtain a trained intelligent decision model, in a case where the loss value does not meet the preset condition. The decision table comprises a mapping relationship between different decision actions and reward values of the nuclear power plant and the steam generator in different working states, wherein the working states are determined according to key operation parameters of the nuclear power plant and thermal-hydraulic parameters of the steam generator, and the reward values are determined according to safety degrees corresponding to liquid level heights and pressures of the steam generator, respectively.

Images