A nuclear power steam generator control method, device, medium and product

Abstract

The application discloses a nuclear power steam generator control method, device, medium and product, relates to the technical field of nuclear power control, and comprises the following steps: constructing a multi-objective optimization model according to real-time operation data; determining a control target set value by using a data assimilation technology and a deep reinforcement learning algorithm improved based on an attention mechanism according to the multi-objective optimization model; determining an actuator control signal by using a self-disturbance control technology according to the control target set value; determining a real-time disturbance feature by using a deep neural network according to real-time operation data after the actuator control signal; and dynamically optimizing the weight distribution of each objective function of the multi-objective optimization model and the actuator control signal by using a fuzzy logic control technology according to the real-time disturbance feature and a historical fault library. The application can improve the control precision of the nuclear power steam generator, enhance the stability of the nuclear power steam generator, and adapt to complex working conditions.

Description

Technical Field

[0001] This application relates to the field of nuclear power control technology, and in particular to a nuclear power steam generator control method, equipment, medium and product. Background Technology

[0002] The nuclear power steam generator is the core equipment for energy conversion between the nuclear island and the conventional island. Its control accuracy is directly related to the safety and power generation efficiency of the nuclear power plant. Traditional PID control methods have the following problems: (1) Poor multi-objective coupling: parameters such as water level, temperature, and pressure affect each other, and single-objective control is prone to suboptimal solutions; (2) Insufficient disturbance resistance: external disturbances such as steam flow fluctuations and blockage of heat transfer tubes can easily lead to control instability; (3) Strong model dependence: existing methods rely on accurate mathematical models and are difficult to adapt to the dynamic changes of complex working conditions.

[0003] In recent years, multi-objective optimization and active disturbance rejection control (ADRC) technologies have been gradually applied to the nuclear power field. However, traditional nuclear power steam generator control methods have problems such as poor multi-objective coupling, insufficient disturbance rejection capability, and strong model dependence.

[0004] Based on the above problems, there is an urgent need to provide a new control method for nuclear power steam generators, which can improve the control accuracy of nuclear power steam generators, enhance the stability of nuclear power steam generators, and adapt to complex operating conditions. Summary of the Invention

[0005] The purpose of this application is to provide a control method, equipment, medium, and product for nuclear power steam generators, which can improve the control accuracy of nuclear power steam generators, enhance the stability of nuclear power steam generators, and adapt to complex operating conditions.

[0006] To achieve the above objectives, this application provides the following solution:

[0007] In a first aspect, this application provides a nuclear power steam generator control method, the nuclear power steam generator control method comprising:

[0008] Acquire real-time operating data of the nuclear power plant steam generator; the real-time operating data includes: water level, temperature, pressure, and steam flow rate data;

[0009] Based on real-time operational data, a multi-objective optimization model is constructed with the objectives of minimizing water level deviation, maximizing heat transfer efficiency, and suppressing overheating risk. Based on the multi-objective optimization model, data assimilation technology and a deep reinforcement learning algorithm based on an attention mechanism are used to determine the control target setpoint.

[0010] Based on the control target setpoint, active disturbance rejection control technology is used to determine the control signals of the actuators of the nuclear power steam generator;

[0011] Based on the real-time operating data after the execution control signal, a deep neural network is used to determine the real-time disturbance characteristics; and based on the real-time disturbance characteristics and the historical fault database, fuzzy logic control technology is used to dynamically optimize the weight allocation of each objective function of the multi-objective optimization model and the control signal of the actuator.

[0012] Optionally, acquiring real-time operating data of the nuclear power steam generator specifically includes:

[0013] The water level in a nuclear power plant steam generator is obtained using a capacitive water level sensor.

[0014] Temperatures at different key locations in a nuclear power plant steam generator are obtained using armored thermocouple temperature sensors; these key locations include: steam outlet, feedwater inlet, and the surface of heat transfer tubes.

[0015] The pressure on the steam side and water side inside a nuclear power plant steam generator is obtained using strain gauge pressure sensors.

[0016] Vortex flow meters installed on steam pipelines are used to obtain steam flow rate change data of nuclear power steam generators; and a flow totalizer is used to process and accumulate the steam flow rate change data in real time.

[0017] Optionally, the step of constructing a multi-objective optimization model based on real-time operating data, with the objective functions of minimizing water level deviation, maximizing heat transfer efficiency, and suppressing overheating risk, specifically includes:

[0018] The multi-objective optimization model J is determined using the formula J=a×A+b×B+c×C;

[0019] Where A is the objective function for minimizing the water level deviation. , Let t be the actual water level height at time t. Let be the set water level at time t, T be the total time, and B be the objective function for maximizing heat transfer efficiency. η(t) is the heat transfer efficiency at time t, and C is the objective function for suppressing the risk of overheating, C = ,in, The actual temperature threshold at time t. The safe temperature threshold is defined as a, b, and c, which are weights of the corresponding objective function.

[0020] Optionally, based on the control target setpoint, active disturbance rejection control technology is used to determine the actuator control signals of the nuclear power steam generator, specifically including:

[0021] The total disturbance is determined using an extended state observer based on the control target setpoint.

[0022] Based on the total disturbance, a nonlinear combination function is used for dynamic compensation.

[0023] Optionally, the step of using a nonlinear combination function for dynamic compensation based on the total disturbance specifically includes:

[0024] Using formula Determine the compensation signal at time t ;

[0025] in, and These are the compensation gains, It is a nonlinear function. These are nonlinear coefficients. denoted as the threshold for the linear interval, e is the error value, and de / dt is the error derivative.

[0026] Optionally, the step of determining the actuator control signal of the nuclear power steam generator using active disturbance rejection control technology based on the control target setpoint further includes:

[0027] When steam pressure Exceeding the preset threshold When necessary, it will automatically switch to the backup water supply valve;

[0028] When the heat transfer tube vibrates at frequency Continuously exceeding the safe range At that time, the foreign object blockage compensation algorithm is activated to correct the total perturbation of the expanded state observer.

[0029] Secondly, this application provides a nuclear power steam generator control device, the nuclear power steam generator control device comprising:

[0030] The data acquisition module is used to acquire real-time operating data of the nuclear power steam generator; the real-time operating data includes: water level, temperature, pressure and steam flow rate data;

[0031] The multi-objective optimization module is used to construct a multi-objective optimization model based on real-time operating data, with the objectives of minimizing water level deviation, maximizing heat transfer efficiency, and suppressing overheating risk. Based on the multi-objective optimization model, data assimilation technology and a deep reinforcement learning algorithm based on an attention mechanism are used to determine the control target setpoint.

[0032] The active disturbance rejection control module is used to determine the control signals of the actuators of the nuclear power steam generator based on the control target setpoint and employing active disturbance rejection control technology.

[0033] The dynamic parameter adjustment module is used to determine the real-time disturbance characteristics based on the real-time operating data after the execution control signal using a deep neural network; and based on the real-time disturbance characteristics and the historical fault database, it uses fuzzy logic control technology to dynamically optimize the weight allocation of each objective function of the multi-objective optimization model and the control signal of the actuator.

[0034] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the nuclear power steam generator control method described above.

[0035] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the nuclear power steam generator control method described above.

[0036] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the nuclear power steam generator control method.

[0037] According to the specific embodiments provided in this application, this application has the following technical effects:

[0038] This application provides a control method, equipment, medium, and product for a nuclear power plant steam generator. Based on real-time operating data, a multi-objective optimization model is constructed with the objectives of minimizing water level deviation, maximizing heat transfer efficiency, and suppressing overheating risk. This model incorporates safety, economic, and environmental objectives. Data assimilation technology and a deep reinforcement learning algorithm based on an attention mechanism are employed to determine the control target setpoint. This comprehensively considers multiple interacting parameters such as water level, temperature, and pressure, avoiding suboptimal solutions caused by single-objective control and generating control target setpoints that better meet actual needs, thereby improving the overall performance of the nuclear power plant. The weights of the objective function are dynamically adjusted through the attention mechanism. Based on the control target setpoint, active disturbance rejection control (ADRC) technology is used to determine the control signals of the nuclear power plant steam generator's actuators, estimate the total disturbance in real time, and compensate for it, improving control accuracy. Finally, deep neural networks and fuzzy logic control technology are used to intelligently and precisely adjust parameters based on real-time disturbance characteristics. This dynamic adjustment mechanism enables the system to better adapt to complex and changing operating conditions, further improving the control performance and stability of the nuclear power plant steam generator. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 This is a schematic flowchart of a nuclear power steam generator control method according to one embodiment of this application;

[0041] Figure 2 This is a schematic diagram of the overall process of a nuclear power steam generator control method according to an embodiment of this application;

[0042] Figure 3 This is a schematic diagram of a nuclear power steam generator control device according to one embodiment of this application. Detailed Implementation

[0043] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0044] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0045] In one exemplary embodiment, such as Figure 1 As shown, a control method for a nuclear power steam generator is provided, comprising the following steps S101 to S104: Wherein:

[0046] S101, acquire real-time operating data of the nuclear power steam generator; the real-time operating data includes: water level, temperature, pressure and steam flow rate data;

[0047] In a real nuclear power plant environment, a high-precision, high-reliability data acquisition system is required for the monitoring and control of nuclear steam generators. This system consists of various types of sensors used to acquire key operating parameters of the nuclear steam generator in real time. Specifically:

[0048] S11, using a capacitive water level sensor to obtain the water level height inside the nuclear power steam generator;

[0049] The installation location of the capacitive water level sensor is determined through CFD simulation, avoiding eddies and steam-water mixing fluctuations, and selecting the middle section of a straight section with stable water flow (typically 1 / 3 to 1 / 2 of the height from the bottom). Furthermore, at least four capacitive water level sensors are installed, symmetrically distributed and with redundancy, to improve reliability. Further, to ensure accurate measurement of the water level within the nuclear power plant steam generator, the installation location must be above the minimum safe water level line, away from the feedwater inlet and downcomer, to avoid temperature / flow interference. The capacitive water level sensor used in this application measures water level by detecting changes in capacitance, featuring high accuracy and fast response. It can capture minute fluctuations in water level in real time, and the measurement error can be controlled within a very small range, meeting the stringent requirements for data accuracy in minimizing water level deviation.

[0050] S12, using armored thermocouple temperature sensors to obtain the temperature at different key locations of the nuclear power steam generator; the different key locations include: steam outlet, feedwater inlet and heat transfer tube surface.

[0051] Among them, the armored thermocouple temperature sensor possesses excellent high-temperature resistance and rapid temperature response characteristics. Temperature sensors are installed at various critical locations within the nuclear power steam generator, such as the steam outlet, feedwater inlet, and the surface of heat transfer tubes, to comprehensively monitor the temperature distribution within the generator. The armored thermocouple temperature sensor converts the temperature signal into an electrical signal and transmits it to the data acquisition unit via a shielded cable, effectively reducing the impact of electromagnetic interference on the signal.

[0052] S13, using strain gauge pressure sensors to obtain the pressure on the steam side and water side inside the nuclear power plant steam generator;

[0053] The measurement principle of strain gauge pressure sensors is based on the change in resistance of a strain gauge under pressure. To ensure the accuracy and reliability of the measurement, the pressure sensor must undergo rigorous calibration and testing before installation. Pressure sensors are installed on both the steam and water sides of the nuclear power plant steam generator to monitor the steam and water pressures in real time, providing crucial data support for the stable operation of the nuclear power plant steam generator.

[0054] S14. The steam flow rate change data of the nuclear power steam generator is obtained by using a vortex flow meter installed on the steam pipeline; and the steam flow rate change data is processed and accumulated in real time by a flow totalizer.

[0055] The working principle of a vortex flow meter is to measure the flow rate by utilizing the relationship between the frequency of vortices generated during fluid flow. Vortex flow meters have advantages such as high accuracy, wide measuring range, and low pressure loss, and can accurately measure changes in steam flow rate in nuclear power plant steam generators.

[0056] The collected data on water level, temperature, pressure, and steam flow are transmitted to the central processing unit of the control system via a high-speed data transmission bus. To ensure the stability and accuracy of data transmission, a redundant communication network, such as industrial Ethernet, is used, equipped with data verification and error correction mechanisms to detect and correct data errors that may occur during transmission in real time.

[0057] S102, based on real-time operating data, a multi-objective optimization model is constructed with the objectives of minimizing water level deviation, maximizing heat transfer efficiency, and suppressing overheating risk; and based on the multi-objective optimization model, data assimilation technology and a deep reinforcement learning algorithm based on an attention mechanism are used to determine the control target setpoint.

[0058] The multi-objective optimization model J is determined using the formula J=a×A+b×B+c×C;

[0059] Where A is the objective function for minimizing the water level deviation. , Let t be the actual water level height at time t. Let be the set water level at time t, T be the total time, and B be the objective function for maximizing heat transfer efficiency. η(t) is the heat transfer efficiency at time t, and C is the objective function for suppressing the risk of overheating, C = ,in, The actual temperature threshold at time t. The safe temperature threshold is defined as a, b, and c, which are weights of the corresponding objective function.

[0060] As a specific example, minimizing water level deviation in a multi-objective optimization model requires accurate water level data and efficient control algorithms. In practical applications, in addition to using high-precision water level sensors, digital filtering techniques (Kalman filtering, median filtering, FIR low-pass filtering, and moving average filtering) are employed to process the collected water level data, removing noise interference and improving the stability and accuracy of the water level readings. The deep reinforcement learning algorithm improved based on the attention mechanism employs a proximal policy optimization (PPO) algorithm based on policy gradients, combined with an experience replay mechanism. The experience replay mechanism stores the experience samples generated during the algorithm's training process in an experience replay pool, randomly selecting samples for learning in subsequent training. This breaks down the correlation between samples, improving the algorithm's training efficiency and stability. To accelerate the solution speed for the Pareto optimal solution set, a multi-core CPU cluster is used for parallel computation, distributing the optimization task across multiple computing nodes for simultaneous processing, significantly shortening the solution time.

[0061] Achieving the goal of maximizing heat transfer efficiency requires comprehensive consideration of the heat transfer process and operating parameters of the nuclear power plant steam generator (feedwater flow rate, steam discharge rate, and heat transfer tube cleaning cycle); heat transfer efficiency η , Where is the theoretical maximum heat transfer (calculated based on the limiting values ​​of the fluid inlet and outlet temperatures), Q is the actual heat transfer, and U is the overall heat transfer coefficient, which is related to the heat transfer tube material, fouling thermal resistance, fluid flow velocity, etc. For heat transfer area, The logarithmic mean temperature difference is used to reflect the temperature difference characteristics between hot and cold fluids. It can be seen that the calculation of heat transfer efficiency η depends not only on conventional measurement parameters such as temperature and flow rate, but also on the heat transfer model inside the nuclear power steam generator. Corrections are made accordingly. In actual operation, the performance of the nuclear power steam generator is regularly tested to obtain key parameters such as the fouling thermal resistance and heat transfer coefficient of the heat transfer tubes, and these parameters are then substituted into the heat transfer model for calculation. Simultaneously, heat transfer efficiency is improved by optimizing the operating parameters of the nuclear power steam generator, such as adjusting the feedwater flow rate, controlling the steam discharge rate, and optimizing the cleaning cycle of the heat transfer tubes.

[0062] S103, based on the control target setpoint, uses active disturbance rejection control technology to determine the control signals of the actuators of the nuclear power steam generator; S103 plays a key role in disturbance rejection and control.

[0063] S103 specifically includes:

[0064] S31, Based on the control target setpoint, the total disturbance is determined using an extended state observer (ESO);

[0065] The primary function of the Observational Observer (ESO) is to uniformly estimate the unmodeled dynamics within the nuclear power steam generator and external disturbances into a total disturbance, and to eliminate its impact on the control output through compensation signals. In actual operation, nuclear power steam generators face various complex disturbance factors, such as steam flow fluctuations, scaling of heat transfer tubes, and equipment vibration. To accurately estimate the total disturbance, the ESO employs an adaptive parameter adjustment strategy, dynamically adjusting the observation gain parameter based on the real-time dynamic characteristics of the nuclear power steam generator, such as the rate of change of the internal medium's flow and the amplitude of pressure fluctuations. In implementation, the observation gain parameter of the ESO is adjusted online using real-time monitored data on the rate of change of flow and the amplitude of pressure fluctuations, according to pre-set parameter adjustment rules. For example, when the rate of change of flow is large, the observation gain parameter is appropriately increased, enabling the ESO to track disturbance changes more quickly and accurately; when the amplitude of pressure fluctuations is small, the observation gain parameter is appropriately decreased to improve the observer's stability. To further improve the estimation accuracy of the ESO, sliding mode observer technology is also employed for improvement. Sliding mode observers are highly robust to disturbances and uncertainties. By introducing a sliding surface into the observer, the observation error can converge to zero in a finite time, thereby improving the estimation accuracy of the total disturbance.

[0066] S32 uses a nonlinear combination function for dynamic compensation based on the total disturbance. S32 can be used as a nonlinear feedback unit.

[0067] Specifically, using formulas Determine the compensation signal at time t ;

[0068] in, and These are the compensation gains, It is a nonlinear function. These are nonlinear coefficients. Here, e is the threshold for the linear interval, e is the error value, and de / dt is the error derivative. In actual adjustments, The initial values ​​of adjustable parameters are determined through a combination of theoretical analysis and simulation experiments. For example, firstly, based on the dynamic characteristics and control requirements of the nuclear power steam generator, the range of these parameters is initially calculated using relevant formulas in control theory. Then, through numerous simulation experiments, these parameters are optimized and adjusted under different operating conditions, and the response performance of the nuclear power steam generator, such as overshoot, settling time, and steady-state error, is observed to finally determine a set of suitable initial values. During the operation of the nuclear power steam generator, these parameters are adjusted online using an adaptive control algorithm based on the actual control effect. For example, when system oscillations occur, the values ​​are appropriately reduced. The value, while adjusting The value of [value] is used to improve the stability of the nuclear power steam generator; when the response speed of the nuclear power steam generator is slow, the value should be appropriately increased. The value of increases the sensitivity of control.

[0069] To prevent control failure caused by a single sensor malfunction, S103 also includes: triggering bypass logic protection based on steam pressure threshold and heat transfer tube vibration frequency signal, which serves as a redundant decision unit.

[0070] Specifically, in practical applications, the steam pressure threshold... and the safe range of heat transfer tube vibration frequency The setting of thresholds and safety ranges requires comprehensive consideration of the nuclear power steam generator's design specifications, historical operating data, and safety standards. Through statistical analysis of extensive historical operating data, combined with the design parameters and safety regulations of the nuclear power steam generator, reasonable thresholds and safety ranges are determined. For example, by statistically analyzing pressure variation data of the nuclear power steam generator under different operating conditions, patterns and anomalies in pressure changes are identified, and this serves as the basis for determining the steam pressure threshold. Simultaneously, considering the close relationship between heat transfer tube vibration frequency and equipment operating status, the safe range of heat transfer tube vibration frequency is determined through monitoring and analysis of heat transfer tube vibration data, combined with the equipment's mechanical performance and safety requirements.

[0071] When steam pressure Exceeding the preset threshold When necessary, the system automatically switches to the standby feedwater valve. To avoid excessive impact on the system during the switching process, a fuzzy control-based switching buffer mechanism is employed. This mechanism uses fuzzy inference to control the opening speed of the standby valve and the closing speed of the original valve based on the rate of change of steam pressure and the difference between the current pressure and a threshold. For example, when the steam pressure rises rapidly and approaches the threshold, the standby valve opens slowly, while the original valve closes quickly; when the steam pressure rises slowly and approaches the threshold, the operating speed of both the standby and original valves is relatively slow, thus achieving a smooth switching and ensuring stable water level and pressure in the nuclear power steam generator.

[0072] When the heat transfer tube vibrates at frequency Continuously exceeding the safe range When the foreign object blockage compensation algorithm is activated, the ESO disturbance estimate is corrected. This algorithm, based on the heat transfer tube blockage model and vibration spectrum analysis, accurately corrects the ESO disturbance estimate. In practical implementation, a heat transfer tube blockage model is first established, considering factors such as the geometry of the heat transfer tube, fluid flow characteristics, and the nature of the blockage. By analyzing the vibration spectrum, characteristic information of the vibration frequency is obtained to determine whether the heat transfer tube is blocked and the degree of blockage. Then, based on the blockage model and vibration spectrum analysis results, the ESO disturbance estimate is corrected. For example, if characteristic frequency components related to heat transfer tube blockage appear in the vibration spectrum, and the vibration frequency continuously exceeds the safe range, the impact of the blockage on system disturbance is calculated according to the blockage model and incorporated into the ESO disturbance estimate, thereby improving the accuracy of the ESO estimate of the total disturbance of the nuclear power steam generator and ensuring the accuracy of control.

[0073] S104 uses a deep neural network to determine real-time disturbance characteristics based on real-time operating data after the execution control signal; and uses fuzzy logic control technology based on real-time disturbance characteristics and historical fault database to dynamically optimize the weight allocation of each objective function of the multi-objective optimization model and the control signal of the actuator, thereby adapting to complex and ever-changing operating conditions.

[0074] The historical fault database is a comprehensive and detailed system built upon long-term accumulation of operational data and fault diagnosis records from nuclear power steam generators. The database includes information such as the time of the fault, fault type, operating parameters at the time of the fault, and fault handling measures. To facilitate data management and retrieval, distributed database technology is used for storage and management of the historical fault database. Distributed databases offer advantages such as high reliability, scalability, and high performance, meeting the data storage and retrieval needs of the historical fault database. Simultaneously, data mining techniques are used to analyze and extract patterns and characteristics of fault occurrences, providing strong support for dynamic parameter adjustments.

[0075] In practical applications, deep neural networks play a crucial role in dynamic parameter tuning, extracting dynamic features of nuclear power steam generators, generating parameter tuning commands for ADRC technology, and compensating for gain. In practical applications, a Convolutional Neural Network (CNN) architecture is employed, which possesses powerful feature extraction capabilities and can automatically learn complex features from the operating data of nuclear power steam generators. The input data for the CNN includes time-series data of multiple real-time operating parameters of the nuclear power steam generator, such as water level, temperature, pressure, and steam flow rate. Through processing with multiple convolutional and pooling layers, feature extraction and dimensionality reduction are performed on the input data, ultimately outputting a dynamic feature vector of the nuclear power steam generator. Then, a fully connected layer is used to map the dynamic feature vector into the space of parameter adjustment instructions and compensation gain for ADRC (Advanced Dynamic Response) technology, generating corresponding adjustment instructions and gain values.

[0076] To improve the training performance and generalization ability of deep neural networks, various training optimization techniques were employed, such as stochastic gradient descent, batch normalization, and Dropout regularization. Simultaneously, the deep neural network was periodically updated and optimized, adjusting its structure and parameters based on new operational data and fault information to adapt to changes in the operating conditions of the nuclear power steam generator. Based on an event-triggered mechanism, multi-objective optimization calculations were initiated only when the nuclear power steam generator's state deviated from the preset safe range, reducing the real-time computational load. The boundary values ​​of the preset safe range were determined through system stability analysis and risk assessment, and were dynamically updated during operation based on the real-time performance indicators of the nuclear power steam generator.

[0077] In practical implementation, the operating state of the nuclear power steam generator is simulated and analyzed using a system dynamics model. Combined with safety standards and operational experience, the initial boundary values ​​of the preset safety range are determined. Then, by real-time monitoring of system operating parameters, such as water level deviation, heat transfer efficiency fluctuations, and overheating risk indicators, the real-time performance indicators of the nuclear power steam generator are calculated. When the real-time performance indicators of the nuclear power steam generator (determined based on raw operating data) approach the boundary of the preset safety range, a risk assessment program is initiated, and the boundary values ​​of the preset safety range are dynamically adjusted based on the current operating conditions and potential risks.

[0078] For example, if the water level deviation of a nuclear power steam generator approaches the upper limit of a preset safety range, and the heat transfer efficiency fluctuates significantly, a risk assessment will indicate a high safety risk. In this case, the upper limit of the preset safety range will be appropriately lowered, and multi-objective optimization calculations will be initiated to adjust the weights of the multi-objectives and the parameters of the ADRC (Advanced Dynamic Response Control) technique to ensure the safe and stable operation of the nuclear power steam generator. Dynamic parameter adjustment, combined with fuzzy logic control technology, uses multiple state variables such as the water level deviation, heat transfer efficiency fluctuation, and overheating risk indicators as fuzzy inputs. Based on pre-set fuzzy rules, the weights of the objective function of the multi-objective model and the parameters of the ADRC technique are adjusted more intelligently and precisely.

[0079] In practical applications, each fuzzy input variable is first fuzzified, transforming it into fuzzy linguistic variables such as "large," "medium," and "small." Then, based on expert experience and extensive simulation data, a fuzzy rule base is developed. Each rule in the fuzzy rule base describes the relationship between the input and output variables. For example, "If the water level deviation is large and the heat transfer efficiency fluctuation is small, then appropriately increase the weight of the water level deviation minimization objective, and simultaneously adjust the proportional gain of ADRC." "Increase".

[0080] During fuzzy inference, based on the current input variable values, fuzzy values ​​of the output variables are calculated using fuzzy matching and inference algorithms. Through defuzzification, the fuzzy output is transformed into weights of the objective function of a specific multi-objective model and parameter adjustment values ​​for ADRC (Adaptive Fuzzy Logic Control) technology. To improve the effectiveness of fuzzy logic control, adaptive fuzzy logic control technology can also be employed to automatically adjust fuzzy rules and membership functions based on the operating conditions of the nuclear power steam generator, further optimizing control performance.

[0081] The process of determining the boundaries of the preset safety zone specifically includes:

[0082] (1) Initial boundary determination: The operating state of the nuclear power steam generator is simulated and analyzed using the system dynamics model. Combined with safety standards (such as safety thresholds in equipment design specifications) and historical operating experience, the initial boundary values ​​of the preset safety range are determined (the dynamic parameter adjustment part is clear).

[0083] (2) Dynamic updates: During operation, the real-time performance indicators of the nuclear power steam generator (such as water level deviation, heat transfer efficiency fluctuation, etc.) are dynamically adjusted. When the real-time performance indicators approach the initial boundary, the boundary values ​​are further optimized through a risk assessment procedure to adapt to changes in operating conditions.

[0084] The risk assessment procedure specifically involves combining real-time performance indicators (such as water level deviation approaching the upper limit, large fluctuations in heat transfer efficiency, etc.) and a historical fault database to assess the current safety risk level of the system; based on the risk assessment results, dynamically adjusting the boundary values ​​of the preset safety range (such as lowering the upper limit of the safety range to improve safety); triggering multi-objective optimization calculations, and readjusting the objective function weight allocation of the multi-objective optimization model and the parameters of Active Disturbance Rejection Control (ADRC) technology to ensure stable system operation.

[0085] Based on adaptive fuzzy logic control technology, and combined with real-time operational data feedback and historical fault database analysis, the core parameters of fuzzy control are dynamically optimized. The specific method is as follows:

[0086] 1. Feedback Adjustment Based on Real-Time Operational Data: Real-time performance indicators of the nuclear power steam generator (such as water level deviation, heat transfer efficiency fluctuation, and overheating risk indicators) are used as feedback signals to evaluate the current fuzzy control effect. For example, if the water level deviation remains "large" (exceeding the expected range) after control, it is identified that the existing fuzzy rule "large water level deviation → adjustment strategy" is insufficient, and the consequent of the rule is modified (such as increasing the adjustment range of ADRC compensation gain). If the control response of the same input variable (such as pressure fluctuation) differs significantly under different operating conditions (such as when steam flow changes abruptly), the applicable conditions of the fuzzy rule are adjusted through online learning (such as adding the antecedent constraint of the rule "when steam flow changes abruptly").

[0087] 2. Pattern Mining Based on Historical Fault Database: Utilize the fault types, operating parameters at the time of the fault, and corresponding control effect data stored in the historical fault database to extract patterns through data mining (such as the optimal control strategy under the "high pressure + high water level deviation" condition), and update the fuzzy rule database accordingly. For example, if historical data shows that "when the heat transfer tube vibration frequency exceeds the limit, the original rules are insufficient to suppress the over-temperature risk," then add or strengthen the rule "vibration frequency exceeding the upper limit → prioritize increasing the target weight of over-temperature risk."

[0088] 3. Dynamic optimization of membership functions: Adjust the parameters (such as range, shape, and center point) of membership functions based on the distribution characteristics of real-time operating data. For example, if the actual water level fluctuation range expands (e.g., from ±50mm to ±80mm), the membership function range corresponding to "large water level deviation" is expanded from ">50mm" to ">80mm" to avoid increased fuzziness error due to changes in operating conditions; if the temperature response speed slows down, the membership function range for "small temperature deviation" can be narrowed (e.g., adjusted from "±2℃" to "±1℃") to improve sensitivity to small temperature changes.

[0089] Among them, the membership function is the core tool in fuzzy logic control for transforming precise input variables into fuzzy linguistic variables, and its characteristics are as follows:

[0090] (1) Fuzzy partitioning of input variables

[0091] For input variables such as water level deviation, heat transfer efficiency fluctuation, and overheat risk index, they are divided into several fuzzy sets (linguistic variables). For example, water level deviation is divided into "negative large (NL)", "negative small (NS)", "zero (Z)", "positive small (PS)" and "positive large (PL)"; heat transfer efficiency fluctuation is divided into "small (S)", "medium (M)" and "large (L)"; and overheat risk index is divided into "low (L)", "medium (M)" and "high (H)".

[0092] (2) The form and function of membership function

[0093] Membership functions describe the degree to which a precise input value belongs to a certain fuzzy set (ranging from 0 to 1). Common forms include triangular, trapezoidal, or Gaussian functions (the document does not explicitly limit the specific form; the form should be adapted to the control requirements). For example, for "water level deviation = positive (PL)", its membership function might be defined as follows: when the actual water level deviation > 50 mm, membership = 1; when the deviation is between 30 and 50 mm, membership increases linearly from 0 to 1 as the deviation increases; when the deviation < 30 mm, membership = 0 (example of triangular function).

[0094] The membership function transforms precise data such as "water level deviation = 40mm" into fuzzy information such as "the membership degree of belonging to the large (PL) is 0.5, and the membership degree of belonging to the small (PS) is 0.5", thus providing a basis for subsequent fuzzy inference.

[0095] (3) Synergy with fuzzy rules

[0096] The output of the membership function serves as the input to the fuzzy rule. For example, in the rule "If the water level deviation is positive (PL) and the overheating risk is low (L), then increase the target weight of the water level deviation (value a)", the membership values ​​of "positive (PL)" and "low (L)" will determine the triggering strength of the rule. Finally, the specific control parameter adjustment value (such as the increment of the value a) is output through defuzzification.

[0097] like Figure 2 As shown, this application provides a cascaded control method for multi-objective optimization and ADRC of nuclear power steam generators. The specific implementation steps are as follows:

[0098] S1 Data Acquisition: The nuclear power plant's steam generator continuously collects real-time data on water level, temperature, pressure, and steam flow rate using various sensors. During acquisition, to ensure accuracy and reliability, each sensor is regularly calibrated and maintained, and preprocessing techniques such as data filtering and noise reduction are employed to process the acquired data. Simultaneously, a data backup mechanism is established to back up the acquired data to multiple storage devices in real time to prevent data loss. The acquired data is transmitted to the central processing unit of the control system via a high-speed data transmission bus, providing data support for subsequent control decisions.

[0099] S2, Target Setting: After receiving the collected real-time operational data, an improved deep reinforcement learning algorithm is used to solve for the Pareto optimal solution set and generate control target setpoints. During the solution process, weights are dynamically allocated using an attention mechanism based on the varying importance of safety, economic, and environmental objectives at different operational stages of the nuclear power plant's steam generator. For example, during the nuclear power plant startup phase, since system stability and safety are paramount, more attention weights are allocated to the objectives of minimizing water level deviation and suppressing overheating risks; during the stable operation phase, to improve power generation efficiency and economic benefits, the weight of the objective of maximizing heat transfer efficiency is appropriately increased. This dynamic weight allocation method generates control target setpoints that better meet actual operational needs. To ensure algorithm convergence and computational efficiency, reasonable learning rates, discount factors, and other parameters are set during the training of the deep reinforcement learning algorithm, and parallel computing technology is employed to accelerate the solution process.

[0100] S3, Control Signal Calculation and Execution: Based on the control target setpoint, the adaptive parameter adjustment strategy of the extended state observer in ADRC technology is adopted. According to the real-time dynamic characteristics of the nuclear power steam generator operation, such as the flow rate change rate and pressure fluctuation amplitude of the internal medium, the observation gain parameters are dynamically adjusted to accurately estimate the total system disturbance. During the estimation process, the ESO feedback mechanism is used to continuously optimize the observer parameters and improve the estimation accuracy of the total disturbance. The nonlinear feedback unit calculates the control signal based on the estimation results, driving the actuators such as feedwater valves. During execution, considering the dynamic response characteristics of the actuators, such as valve opening and closing times and flow characteristics, the control signal is appropriately compensated and adjusted to ensure that the actuators can accurately execute control commands and achieve precise control of the nuclear power steam generator.

[0101] S4, Parameter Adjustment: Dynamic parameter adjustment combined with fuzzy logic control technology analyzes the control effect. Multiple state variables of the nuclear power steam generator operation, such as water level deviation, heat transfer efficiency fluctuation, and overheating risk indicators, are used as fuzzy inputs. Based on pre-set fuzzy rules, the weights of multiple objectives and the parameters of ADRC technology are intelligently and precisely adjusted. During the adjustment process, the operating status of the nuclear power steam generator is continuously monitored, and the fuzzy rules and membership functions are dynamically updated based on feedback information from the control effect to adapt to complex and changing operating conditions. For example, if it is found that the water level deviation of the nuclear power steam generator remains large for a period of time, while the heat transfer efficiency fluctuation is small, the weight of the water level deviation minimization objective is appropriately increased according to the fuzzy rules, and the integral gain of ADRC is increased to enhance the ability to regulate the water level deviation. At the same time, to avoid over-adjustment of parameters leading to instability of the nuclear power steam generator, boundary and step size limits for parameter adjustment are set. After each adjustment, the effectiveness of the adjustment is determined using performance evaluation indicators of the nuclear power steam generator, such as control accuracy (quantified by the degree of deviation of key parameters from the target setpoint), stability (assessed by the fluctuation amplitude, duration, and disturbance immunity of key parameters), and response speed (measured by the adjustment time of the system to changes or disturbances in the target setpoint). If the performance of the nuclear power steam generator does not improve or even deteriorates after the adjustment, the adjustment is revoked, and other adjustment strategies are tried.

[0102] In addition, if any abnormalities are found in the nuclear power steam generator, such as sudden changes in steam flow or blockage of heat transfer tubes by foreign objects, the relevant information should be reported in a timely manner.

[0103] S5, Disturbance Compensation: When abnormal conditions such as foreign object blockage in the heat transfer tube or sudden changes in steam flow are detected, the redundancy judgment unit is triggered. The redundancy judgment unit makes judgments based on steam pressure thresholds, heat transfer tube vibration frequency signals, etc. If the steam pressure exceeds the preset threshold, in addition to automatically switching to the backup feedwater valve, it will also record the time of the fault, pressure value, and other information, and issue an alarm to the operator through the alarm mechanism of the nuclear power steam generator, prompting them to pay attention to the operating status of the nuclear power steam generator. After switching valves, the changes in parameters such as water level and pressure of the nuclear power steam generator are closely monitored to ensure stable operation of the nuclear power steam generator.

[0104] If the vibration frequency of the heat transfer tubes continues to exceed the safe range, the foreign object blockage compensation algorithm is activated. This algorithm not only corrects the ESO disturbance estimate but also combines historical vibration data and blockage cases of the heat transfer tubes to make a preliminary judgment on the possible location and extent of blockage. Then, based on the judgment results, corresponding measures are taken, such as adjusting the operating parameters of the nuclear power steam generator and reducing the load, to reduce the vibration and wear of the heat transfer tubes. At the same time, maintenance personnel are notified to inspect and clean the heat transfer tubes to prevent the blockage from worsening. After maintenance is completed, the vibration of the heat transfer tubes and the operating status of the system are reassessed to ensure that the system returns to normal operation.

[0105] Among them, the foreign object blockage compensation algorithm is based on the vibration frequency of the heat transfer tube. Continuously exceeding the preset safety range When the system determines that the heat transfer tube may be blocked by foreign objects (such as mechanical impurities, scale buildup, etc.), it automatically activates the compensation algorithm. The vibration frequency is monitored in real time by sensors; "continuously exceeding" typically refers to the vibration frequency exceeding a preset threshold for a duration exceeding the safe range (e.g., 30 consecutive seconds) to avoid false triggering caused by momentary interference. The operation of the foreign object blockage compensation algorithm relies on a pre-established heat transfer tube blockage model and real-time vibration spectrum analysis.

[0106] (1) Heat transfer tube blockage model: The heat transfer tube blockage model incorporates key parameters such as the geometry of the heat transfer tube (e.g., tube diameter, length, support plate spacing), fluid flow characteristics (e.g., steam / water flow rate, velocity distribution) and blockage properties (e.g., blockage type, accumulation thickness, distribution location) to quantify the impact of blockage on system disturbance (e.g., the correlation between vibration frequency shift and pressure drop change when the blockage rate increases by 10%).

[0107] (2) Vibration spectrum analysis: By extracting the spectral features of the heat transfer tube vibration signal (such as abnormal frequency components and amplitude changes), and combining the correspondence between "foreign object blockage - vibration features" in the historical fault database, the severity (such as mild, moderate, and severe) and approximate location of the blockage can be accurately identified.

[0108] After activating the heat transfer tube blockage model, the total perturbation estimate of ESO is corrected through the following steps:

[0109] 1. Based on the vibration spectrum analysis results and combined with the heat transfer tube blockage model, calculate the actual disturbance amplitude of the system caused by foreign object blockage (such as the disturbance caused by increased fluid resistance and decreased heat transfer efficiency due to blockage).

[0110] 2. Use the amplitude of this disturbance as a compensation amount to dynamically correct the ESO's estimate of the "total disturbance" (the original ESO may not have fully captured the nonlinear disturbance caused by congestion).

[0111] 3. The corrected total disturbance is incorporated into the dynamic compensation stage of ADRC, enabling the actuator control signals (such as the adjustment of the water supply valve opening) to more accurately offset the effects of blockage, suppress abnormal vibration, and restore stable system operation.

[0112] Throughout the cascaded control process, a comprehensive system monitoring and recording mechanism was established. Through the monitoring interface, operators can view various operating parameters of the nuclear power plant steam generator, control target setpoints, control signals, and the working status of each module in real time. Simultaneously, the system's operating data is recorded in real time, including historical fault information, parameter adjustment records, and control effect evaluation data. This recorded data is used for subsequent system analysis, performance optimization, and fault diagnosis, providing strong support for continuous improvement and refinement of control methods and the system. For example, by analyzing historical fault data, the patterns and trends of fault occurrence can be summarized, allowing for preventative measures to be taken in advance and reducing the probability of faults. Furthermore, by studying parameter adjustment records and control effect evaluation data, fuzzy rules and ADRC parameters can be further optimized to improve the system's control performance.

[0113] This application uses data assimilation technology (such as Kalman filter) to fuse real-time data (water level, temperature, pressure, steam flow, etc.) collected by sensors with the predicted values ​​of the mechanism model to analyze the changing trends of parameters such as current heat transfer efficiency and fouling thermal resistance.

[0114] By utilizing the attention mechanism in deep reinforcement learning algorithms, the weights of safety and economic objectives are dynamically allocated according to the operating phase of the nuclear power steam generator (such as startup, stable operation, and shutdown). For example, during the stable operation phase, the focus is on maximizing heat transfer efficiency.

[0115] Based on the Pareto optimal solution set, adjustment instructions are generated for real-time operating data, such as:

[0116] When heat transfer efficiency decreases, the system automatically calculates the optimal feedwater flow rate increment or steam discharge rate adjustment value to increase the heat transfer temperature difference or flow rate, thereby improving U and ;

[0117] Based on the monitoring data of fouling thermal resistance, the cleaning cycle of the heat transfer tube is dynamically optimized, and a cleaning command is triggered when the fouling thermal resistance exceeds the threshold.

[0118] The adjusted control strategy is executed using ADRC technology (e.g., driving feedwater valves), and the control effect is monitored in real time (e.g., whether heat transfer efficiency recovers, whether water level / pressure stabilizes). Based on the control effect and combined with fuzzy logic technology, water level deviation and heat transfer efficiency fluctuations are used as inputs to adjust the weights of the objective function of the multi-objective model and the parameters in the ADRC technology in real time, forming a closed-loop process of "data acquisition-analysis-adjustment-verification". For example, if the heat transfer efficiency does not meet expectations after adjusting the feedwater flow rate, the system will automatically increase the adjustment weight of the steam emission rate and re-optimize the strategy; achieving the over-temperature risk suppression target requires accurate temperature monitoring and effective control measures. In practical applications, the safe temperature threshold... The settings are not fixed but dynamically adjusted based on the material properties of the nuclear power steam generator, its operating history, and current operating conditions. For example, after prolonged operation, the high-temperature resistance of the nuclear power steam generator may decrease due to equipment aging and corrosion. In this case, it is necessary to appropriately lower the safe temperature threshold to ensure the safe operation of the equipment. When the actual temperature... When the safe temperature threshold is exceeded, the system will immediately activate over-temperature risk mitigation measures, such as increasing the cooling medium flow rate, adjusting the steam emission rate, and reducing the load on the nuclear power steam generator. The implementation of these measures is coordinated and controlled by a multi-objective optimization module based on the calculation results of a deep reinforcement learning algorithm, to ensure that the over-temperature risk is mitigated while minimizing the impact on the system's economy and environmental performance.

[0119] The multi-objective optimization module integrates data assimilation technology, with the crucial step of using a Kalman filter to fuse sensor data and mechanistic model predictions to correct the heat transfer efficiency calculation parameters. The Kalman filter is an optimal filter based on linear minimum mean square error estimation. In practical applications, the parameters of the Kalman filter need to be precisely set according to the specific characteristics of the nuclear power steam generator and the statistical characteristics of the measurement noise. For example, by statistically analyzing historical sensor measurement data, the covariance matrix of the measurement noise is obtained and used as one of the input parameters of the Kalman filter. Simultaneously, combined with the mechanistic model of the nuclear power steam generator, the uncertainty of the model is quantified and incorporated into the Kalman filter calculation process, thereby achieving precise correction of the heat transfer efficiency calculation parameters.

[0120] To accelerate Pareto front convergence using initial value optimization methods, a particle swarm optimization (PSO) initial population generation strategy based on cluster analysis is adopted. First, historical operating data from the nuclear power plant's steam generator is clustered, dividing the data under different operating conditions into multiple categories. Then, based on the characteristics and distribution patterns of each category, the initial population of the PSO algorithm is rationally distributed in the solution space. This makes the initial population more representative, covering a wider range of solution spaces, thereby accelerating the convergence speed of the Pareto front and improving the efficiency and quality of multi-objective optimization.

[0121] Based on the same inventive concept, this application also provides a nuclear power steam generator control device for implementing the aforementioned nuclear power steam generator control method. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations in one or more embodiments of the nuclear power steam generator control device provided below can be found in the limitations of the nuclear power steam generator control method described above, and will not be repeated here.

[0122] In one exemplary embodiment, such as Figure 3 As shown, a nuclear power steam generator control device is provided, comprising:

[0123] The data acquisition module is used to acquire real-time operating data of the nuclear power steam generator; the real-time operating data includes: water level, temperature, pressure and steam flow rate data;

[0124] The multi-objective optimization module is used to construct a multi-objective optimization model based on real-time operating data, with the objectives of minimizing water level deviation, maximizing heat transfer efficiency, and suppressing overheating risk. Based on the multi-objective optimization model, data assimilation technology and a deep reinforcement learning algorithm based on an attention mechanism are used to determine the control target setpoint.

[0125] The active disturbance rejection control module is used to determine the control signals of the actuators of the nuclear power steam generator based on the control target setpoint and employing active disturbance rejection control technology.

[0126] The dynamic parameter adjustment module is used to determine the real-time disturbance characteristics based on the real-time operating data after the execution control signal using a deep neural network; and based on the real-time disturbance characteristics and the historical fault database, it uses fuzzy logic control technology to dynamically optimize the weight allocation of each objective function of the multi-objective optimization model and the control signal of the actuator.

[0127] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal. The computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is connected to the system bus via the I / O interfaces. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The I / O interfaces of the computer device are used for exchanging information between the processor and external devices. The communication interface of the computer device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a nuclear power steam generator control method.

[0128] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0129] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0130] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0131] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0132] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0133] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0134] In this application, all actions to acquire signals, information, or data are carried out in compliance with the relevant data protection laws and policies of the country where the location is situated, and with the authorization granted by the owner of the relevant device.

[0135] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0136] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A nuclear power steam generator control method, characterized by, The nuclear power steam generator control method comprises: acquiring real-time operation data of the nuclear power steam generator; the real-time operation data comprises water level height, temperature, pressure and steam flow data; constructing a multi-objective optimization model with water level height deviation minimization, heat transfer efficiency maximization and super-temperature risk suppression as objective functions according to the real-time operation data; and determining control target set values according to the multi-objective optimization model by using data assimilation technology and a deep reinforcement learning algorithm improved based on an attention mechanism; determining an actuator control signal of the nuclear power steam generator by using active disturbance rejection control technology according to the control target set values; determining real-time disturbance characteristics by using a deep neural network according to the real-time operation data after the actuator control signal; and dynamically optimizing weight distribution of each objective function of the multi-objective optimization model and the actuator control signal by using fuzzy logic control technology according to the real-time disturbance characteristics and a historical fault library; the multi-objective optimization model with water level height deviation minimization, heat transfer efficiency maximization and super-temperature risk suppression as objective functions is constructed according to the real-time operation data, and specifically comprises: a multi-objective optimization model J is determined by using a formula J = a × A + b × B + c × C; wherein A is an objective function of water level height deviation minimization, , is an actual water level height at time t, is a set water level height at time t, T is a total time, B is an objective function of heat transfer efficiency maximization, , η(t) is a heat transfer efficiency at time t, C is an objective function of over-temperature risk suppression, C= wherein, is an actual temperature threshold at time t, is a safety temperature threshold, a, b, c are weights corresponding to objective functions; the actuator control signal of the nuclear power steam generator is determined by using active disturbance rejection control technology according to the control target set values, and specifically comprises: total disturbance is determined by using an extended state observer according to the control target set values; dynamic compensation is performed by using a nonlinear combination function according to the total disturbance; the dynamic compensation is performed by using the nonlinear combination function according to the total disturbance, and specifically comprises: Using the formula determining the compensation signal at time t ; wherein, and are compensation gains, is a non-linear function, is a non-linear coefficient, is a linear interval threshold, e is an error value, and de / dt is an error derivative.

2. The nuclear power steam generator control method of claim 1, wherein, the real-time operation data of the nuclear power steam generator is acquired, and specifically comprises: a water level height in the nuclear power steam generator is acquired by using a capacitive water level sensor; temperatures of different key positions of the nuclear power steam generator are acquired by using armored thermocouple temperature sensors; the different key positions comprise a steam outlet, a feedwater inlet and a heat transfer pipe surface; pressures of steam sides and water sides in the nuclear power steam generator are acquired by using strain gauge pressure sensors; steam flow change data of the nuclear power steam generator are acquired by using a vortex shedding flowmeter installed on a steam pipeline; and the steam flow change data are processed and accumulated in real time by using a flow totalizer.

3. The nuclear power steam generator control method of claim 1, wherein the actuator control signal of the nuclear power steam generator is determined by using active disturbance rejection control technology according to the control target set values, and specifically further comprises: When the steam pressure exceeds a preset threshold value, automatically switch to a backup water supply valve; When the heat transfer tube vibration frequency Exceeds the safe range The foreign matter clogging compensation algorithm is activated to correct the total disturbance of the expansion state observer.

4. A nuclear power steam generator control apparatus for implementing the nuclear power steam generator control method according to any one of claims 1 to 3, characterized by the nuclear power steam generator control device comprises: a data acquisition module configured to acquire real-time operation data of the nuclear power steam generator; the real-time operation data comprises water level height, temperature, pressure and steam flow data; a multi-objective optimization module configured to construct a multi-objective optimization model with water level height deviation minimization, heat transfer efficiency maximization and super-temperature risk suppression as objective functions according to the real-time operation data; and determine control target set values according to the multi-objective optimization model by using data assimilation technology and a deep reinforcement learning algorithm improved based on an attention mechanism; an active disturbance rejection control module configured to determine an actuator control signal of the nuclear power steam generator by using active disturbance rejection control technology according to the control target set values; The dynamic parameter adjustment module is configured to determine real-time disturbance features by using a deep neural network according to real-time running data after the execution control signal; and dynamically optimize the weight distribution of each objective function of the multi-objective optimization model and the execution mechanism control signal by using a fuzzy logic control technology according to the real-time disturbance features and the historical fault library.

5. A computer device comprising: The memory, the processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the nuclear power steam generator control method of any one of claims 1-3.

6. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the nuclear power steam generator control method of any one of claims 1-3.

7. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the nuclear power steam generator control method of any one of claims 1-3. The computer program is executed by the processor to implement the nuclear power steam generator control method of any one of claims 1-3.

Images