Regulator spray valve control method

Abstract

The application discloses a pressure stabilizer spray valve control method, comprising: acquiring valve rod angle data and spray flow data, taking the valve rod angle data corresponding to the first time when the spray flow data exceeds the preset flow threshold as the physical zero position data; obtaining the real opening of the valve based on the physical zero position data, and performing characteristic calibration on the corresponding relationship between the real opening and the spray flow data to obtain mapping data of the real opening and the spray flow data; based on the mapping data, predicting the system state under different openings to obtain a prediction result, and determining the minimum opening instruction that meets the preset operation condition according to the prediction result. The application introduces the physical zero position calibration and the real opening-flow mapping mechanism, eliminates the control blind area caused by the mechanical dead zone, realizes the quantization and prediction of the real mechanical state of the valve, and can directly determine the minimum opening instruction that meets the preset operation condition according to the prediction result, thereby improving the control efficiency of the continuous spray regulation test of the pressure stabilizer.

Description

Technical Field

[0001] This invention relates to the field of valve control technology, and more specifically, to a method for regulating a pressure regulator spray valve. Background Technology

[0002] The pressurizer spray valve (RCP001 / 002VP) in a nuclear power plant is a key device for maintaining the stability of the primary loop pressure. Its RCP003 test (pressurizer continuous spray regulation test) aims to determine and set the minimum spray opening of the valve to reduce thermal stress and maintain water chemical properties. The test requires that the temperature difference between the two spray lines be less than 5°C and that the power fluctuation of the proportional heater be controlled within (50±5)%. However, because this spray valve is an angular stroke regulating valve, the linear motion of the pneumatic rod is converted into the rotational motion of the valve stem through the crank arm and universal joint. Moreover, the valve disc size is slightly larger than the valve seat, resulting in a "mechanical dead zone" in the closed position: within this dead zone, although the valve stem generates stroke according to the control system command (instrument signal opening), the valve disc still covers the valve seat, and the actual mechanical opening is zero. This leads to three technical drawbacks: First, the true opening degree is unknowable. Traditional methods relying on instrument signals cannot perceive the true mechanical state within the dead zone, making the RCP003 (continuous spray regulation of the voltage regulator) test entirely dependent on the operator's experience for repeated adjustments, with tests taking 3-5 hours and easily occupying the critical path of major overhauls. Second, the zero-point reference is not unified. Valve maintenance and debugging lack a unified physical zero-point reference, and the "zero point" setting varies from person to person, introducing human error. Third, the control is blind. There is no dynamic correlation model between valve opening, pipeline flow, pipeline temperature, and proportional heater power fluctuations, and the test control lacks data support, making accurate prediction and control impossible. Currently, there is no complete technical solution that comprehensively solves the above-mentioned problems of "dead zone measurement," "zero-point calibration," and "intelligent control under data constraints." Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a method for regulating the spray valve of a voltage regulator, addressing the problems existing in the prior art.

[0004] The technical solution adopted by this invention to solve its technical problem is: a method for regulating a voltage regulator spray valve, comprising the following steps: Step S1: Obtain valve stem angle data and spray flow rate data, and take the valve stem angle data corresponding to the first time the spray flow rate data exceeds the preset flow rate threshold as the physical zero point data; Step S2: Obtain the actual valve opening based on the physical zero-position data, and perform characteristic calibration on the correspondence between the actual opening and the spray flow data to obtain the mapping data between the actual opening and the spray flow data; Step S3: Based on the mapping data, predict the system state under different opening degrees, obtain the prediction results, and determine the minimum opening degree command that satisfies the preset operating conditions according to the prediction results.

[0005] Furthermore, the voltage regulator spray valve includes a valve stem and a valve disc; the valve stem is used to drive the valve disc to complete the opening adjustment, and a concentric annular target is provided on the cylindrical surface of the valve stem; The valve stem angle data is obtained in the following way: The image of the concentric ring target is acquired by the image acquisition device, and the valve stem angle data is calculated based on the displacement of the center of the ellipse formed by the target projection; wherein the image acquisition device is located next to the valve stem, and its lens is aimed at the concentric ring target; Alternatively, the radial displacement data of the valve stem cylindrical surface collected by the laser displacement sensor can be obtained, and the valve stem angle data can be calculated. The laser displacement sensor is installed on the outer periphery of the valve stem, with its detection end facing the valve stem surface.

[0006] Furthermore, the calculation of the valve stem angle data based on the displacement of the center of the ellipse formed by the target projection includes: After preprocessing the image, the edge points of the target projection are extracted; Ellipse fitting is performed on the edge points to obtain the coordinates of the ellipse center; The center coordinates of the ellipse are located at the sub-pixel level; The horizontal displacement of the ellipse center coordinates within the image plane is calculated. The pixel equivalent is obtained through camera calibration. Combined with the valve stem radius, the horizontal displacement is converted into the valve stem angle data.

[0007] Furthermore, the laser beam of the laser displacement sensor is perpendicular to the axial direction of the valve stem and illuminates a preset position on the cylindrical surface of the valve stem, the preset position having a preset horizontal distance from the rotation center of the valve stem; The step of obtaining the radial displacement data of the valve stem cylindrical surface collected by the laser displacement sensor and converting it into the valve stem angle data includes: Acquire radial displacement data collected by the laser displacement sensor, wherein the radial displacement data includes the change in radial distance from the laser displacement sensor to the valve stem surface; The valve stem angle data is obtained by converting the radial distance change to the preset horizontal distance through arctangent calculation.

[0008] Further, in step S2, obtaining the actual valve opening based on the physical zero-position data includes: Subtract the physical zero-position data from the valve stem angle data to obtain the actual opening degree; When the valve stem angle data is less than the physical zero position data, the actual opening degree is set to zero.

[0009] Further, in step S2, the step of performing characteristic calibration on the correspondence between the actual opening degree and the spray flow rate data to obtain the mapping data between the actual opening degree and the spray flow rate data includes: Obtain the fluid dynamics model of the valve, perform flow field simulation under different actual opening degrees, and obtain the first spray flow rate data corresponding to the actual opening degree; And / or, acquire second spray flow data collected on the test bench based on different actual opening degrees; When the first spray flow rate data and the second spray flow rate data are obtained, a correspondence is established between the first spray flow rate data and the second spray flow rate data and the actual opening degree to obtain the mapping data between the actual opening degree and the spray flow rate data.

[0010] Further, the step of establishing a correspondence between the first spray flow rate data and the second spray flow rate data and the actual spray opening, to obtain the mapping data between the actual spray opening and the spray flow rate data, includes: Set the prior distribution of the parameters to be calibrated; Construct a likelihood function based on the second spray flow rate data; The posterior sampling method is used to perform posterior sampling, which is iterated multiple times and aging samples are discarded to obtain the posterior distribution of the parameter to be calibrated; The posterior mode is taken as the calibration parameter value, and the first spray flow rate data is corrected based on the calibration parameter value to obtain the calibrated spray flow rate data; Establish the correspondence between the actual opening degree and the calibrated spray flow rate data to obtain the mapping data.

[0011] Further, in step S3, the prediction of the system state under different opening degrees based on the mapping data, and the resulting prediction, includes: Obtain the cold section temperature, the initial water temperature of the voltage regulator, and the control parameters of the heater; The actual opening degree, the spray flow rate data, the cold section temperature, the initial water temperature of the pressure regulator, and the heater control parameters are input into the thermal hydraulic physical model to predict the spray pipeline temperature difference and the heater power, thus obtaining the prediction results.

[0012] Furthermore, the step of inputting the actual opening degree, the spray flow rate data, the cold section temperature, the initial water temperature of the pressure regulator, and the heater control parameters into the thermal-hydraulic physical model to predict the spray pipeline temperature difference and heater power, and obtaining the prediction results, also includes: The prediction results of the thermal-hydraulic physical model are corrected using a long short-term memory neural network. The long short-term memory neural network uses historical experimental data as training samples, and takes the actual opening degree, spray flow rate data, initial temperature, historical power and historical temperature difference within the time window as inputs, and outputs the prediction residual. The sum of the predicted value from the thermal-hydraulic physical model and the predicted residual is taken as the final prediction result.

[0013] Further, in step S3, determining the minimum opening instruction that satisfies the preset operating conditions based on the prediction result includes: With minimizing the actual opening degree as the objective function, and with the constraints that the absolute value of the temperature difference in the spray pipeline does not exceed a first threshold and the ratio of the heater power to the maximum power is between a first ratio and a second ratio, a constrained optimization problem is constructed. The optimization problem is solved using an optimization algorithm to obtain the minimum true opening that satisfies the constraints, which is then used as the minimum opening instruction.

[0014] The beneficial effects of this invention are as follows: This invention provides a method for regulating a pressure regulator spray valve, comprising: acquiring valve stem angle data and spray flow data; using the valve stem angle data corresponding to the first time the spray flow data exceeds a preset flow threshold as physical zero-position data; acquiring the actual valve opening based on the physical zero-position data, and performing characteristic calibration on the correspondence between the actual opening and the spray flow data to obtain mapping data between the actual opening and the spray flow data; predicting the system state under different openings based on the mapping data, obtaining prediction results, and determining the minimum opening command that satisfies preset operating conditions based on the prediction results. This invention, by introducing physical zero-position calibration and an actual opening-flow mapping mechanism, eliminates the control blind zone caused by mechanical dead zones, realizes the quantification and prediction of the actual mechanical state of the valve, thereby significantly improving the regulation efficiency and accuracy of the RCP003 test, shortening the test time, reducing reliance on operator experience, and effectively ensuring the stability of the primary circuit pressure and the safety of equipment operation. Attached Figure Description

[0015] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings: Figure 1 This is a logic flowchart of the voltage regulator spray valve control method; Figure 2 This is a schematic diagram of an ellipse formed by the projection of a ring target in optical vision measurement. Figure 3 This is a schematic diagram of a laser beam illuminating a preset position on the cylindrical surface of a valve stem in a laser ranging method. Detailed Implementation

[0016] To provide a clearer understanding of the technical features, objectives, and effects of the present invention, specific embodiments of the invention are now described in detail with reference to the accompanying drawings. In the following description, specific details such as particular structures and techniques are set forth for illustrative purposes and not for limitation, so as to provide a thorough understanding of the embodiments of the invention. However, those skilled in the art will understand that the invention can also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known devices, circuits, and methods are omitted so as not to obscure the description of the invention with unnecessary detail.

[0017] like Figure 1 As shown, Figure 1 This is a logic flowchart of the voltage regulator spray valve control method.

[0018] This invention provides a method for regulating a voltage regulator spray valve, comprising: Step S1: Obtain valve stem angle data and spray flow rate data. Use the valve stem angle data corresponding to the first time the spray flow rate data exceeds the preset flow rate threshold as the physical zero point data. In this step, it should be noted that due to the mechanical dead zone of the pressure regulator spray valve, traditional methods relying on instrument signals for valve opening cannot perceive the valve's true mechanical state. To solve this technical problem, this invention synchronously collects the actual movement data of the valve stem and the flow data of the downstream spray pipeline, defining the valve stem position corresponding to the first time the spray flow data exceeds a preset flow threshold as the "physical zero point." This physical zero point establishes a unified and objective benchmark for all subsequent control operations, eliminating installation and wear differences caused by different valves and different maintenance batches.

[0019] Specifically, the valve stem angle data can be obtained through non-contact measurement. The criterion for determining when the spray flow rate data first exceeds a preset flow threshold can be a comprehensive judgment combining the preset flow threshold with the duration and valve stem movement trend, to eliminate false triggering caused by signal noise. More specifically, the flow threshold can be set to 0.05%-0.5% of the fully open valve flow rate. The valve is considered truly open only when the flow rate exceeds the threshold and remains stable for more than 0.2-0.5 seconds, while the valve stem angle continues to increase slowly and continuously.

[0020] Step S2: Obtain the actual valve opening based on the physical zero-point data, and perform characteristic calibration on the correspondence between the actual opening and the spray flow data to obtain the mapping data between the actual opening and the spray flow data; In this step, it should be noted that the true opening degree refers to the effective mechanical stroke obtained by subtracting the calibrated physical zero point from the current valve stem angle. In the low-opening region of the valve, the flow rate and opening degree exhibit a complex nonlinear relationship, requiring precise calibration to establish a reliable mapping to support subsequent system state prediction.

[0021] Step S3: Based on the mapping data, predict the system state under different opening degrees, obtain the prediction results, and determine the minimum opening degree command that meets the preset operating conditions based on the prediction results.

[0022] In this step, it should be noted that the system state includes at least the temperature difference between the two spray lines and the power fluctuation of the proportional heater. This invention maps the actual opening degree to the spray flow rate, further predicts the system response under different opening degrees, and obtains the minimum opening degree command that satisfies the operating conditions by solving an optimization problem.

[0023] Furthermore, the voltage regulator spray valve includes a valve stem and a valve disc; the valve stem is used to drive the valve disc to complete the opening adjustment, and concentric annular targets are set on the cylindrical surface of the valve stem; the valve stem angle data is obtained in the following way: Optical vision method: The image of the concentric ring target is acquired by the image acquisition device, and the valve stem angle data is calculated based on the displacement of the center of the ellipse formed by the target projection; wherein, the image acquisition device is located next to the valve stem, and its lens is aimed at the concentric ring target. It should be noted that the core of this method lies in using the geometric transformation relationship of the cylindrical target projection to inversely solve the valve stem rotation angle. The annular target on the valve stem surface forms an ellipse on the imaging plane, and its center position shifts horizontally as the valve stem rotates. This shift has a definite mapping relationship with the rotation angle. In addition to the above-mentioned scheme based on the displacement of the ellipse center, other angle measurement methods based on image analysis can also be used, such as analyzing the changes in ellipse shape parameters (such as the ratio of major to minor axes), or setting multiple feature points on the target and solving the rotation angle through pose calculation algorithms.

[0024] Alternatively, laser ranging method: acquire radial displacement data of the valve stem cylindrical surface collected by laser displacement sensor, and calculate the valve stem angle data. The laser displacement sensor is installed on the outer periphery of the valve stem with its detection end facing the valve stem surface.

[0025] It should be noted that the core of this method lies in utilizing the triangular relationship between the change in radial distance on the valve stem surface and the rotation angle during valve stem rotation. In addition to the single-point laser displacement sensor solution, alternative solutions such as multi-sensor fusion measurement, different types of displacement sensors (e.g., triangular reflective, confocal, or fiber optic), or Hall effect-based magnetic angle encoders can be used, depending on the on-site installation conditions and accuracy requirements. These all fall within the scope of the non-contact measurement concept of this invention.

[0026] Furthermore, the valve stem angle data is calculated based on the displacement of the center of the ellipse formed by the target projection, including: extracting the edge points of the target projection after image preprocessing; performing ellipse fitting on the edge points to obtain the coordinates of the ellipse center; performing sub-pixel-level positioning on the ellipse center coordinates; calculating the horizontal displacement of the ellipse center coordinates in the image plane, obtaining the pixel equivalent through camera calibration, and converting the horizontal displacement into valve stem angle data by combining it with the valve stem radius.

[0027] like Figure 2 As shown, specifically, a high-contrast annular target 2 is attached or engraved on the cylindrical surface in the middle of the valve stem 1. This target contains at least two concentric rings to improve the positioning accuracy and anti-interference capability during image recognition. An image acquisition device (such as an industrial camera) is fixedly installed on the side of the valve stem 1, with its lens aimed at the annular target 2, and provides uniform illumination in conjunction with a ring light source to enhance the visibility of the target in complex industrial environments.

[0028] After preprocessing the acquired target images (including grayscale conversion and Gaussian filtering for noise reduction), the edge points of the target projection are extracted using the Canny operator. The Gaussian filter kernel size is set to 5×5, and the high / low threshold ratio of the Canny operator is set to 1:2 to obtain clear, continuous edge contours with minimal noise interference.

[0029] For the extracted edge points, the general equation of the ellipse is fitted using the least squares method: in, , C, D, E, and F are the coefficients of the equation of the ellipse to be fitted, and (x, y) are the pixel coordinates of the edge points in the image plane. By solving this equation, the coordinates of the ellipse center can be obtained. Since valve stem 1 is a cylinder, the projection of the annular target 2 on its surface onto the image plane is an ellipse, and the center of this ellipse shifts horizontally as the valve stem rotates.

[0030] To further improve measurement accuracy, the following method is adopted: The moment method is used to locate the center of the ellipse at the sub-pixel level. Specifically, by matching the grayscale distribution within a 7×7 neighborhood with the moments of an ideal step edge model, the edge localization accuracy is improved to 0.1 pixels, thereby obtaining a higher resolution displacement change.

[0031] Let the horizontal displacement of the center of the ellipse in the image plane be... (Unit: pixels), obtain pixel equivalent k (unit: mm / pixel) through camera calibration, given the valve stem radius is... (Unit: mm), then the actual rotation angle of the valve stem Calculate using the following formula: When the turning angle is small, a small angle approximation can be used to simplify the calculation, that is... The radian value is then converted to degrees for output. Using this algorithm, the system can output valve stem angle data in real time. The measurement resolution can reach within 0.05°, meeting the requirements for high-precision monitoring.

[0032] Furthermore, the laser beam of the laser displacement sensor is perpendicular to the axis of the valve stem 1 and illuminates a preset position 3 on the cylindrical surface of the valve stem. The preset position 3 has a preset horizontal distance from the rotation center of the valve stem 1. The radial displacement data of the cylindrical surface of the valve stem collected by the laser displacement sensor is acquired and converted into valve stem angle data through calculation. This includes: acquiring the radial displacement data collected by the laser displacement sensor, which includes the change in radial distance from the laser displacement sensor to the valve stem surface; and converting the valve stem 1 angle data through arctangent calculation based on the ratio of the change in radial distance to the preset horizontal distance.

[0033] like Figure 3 As shown, specifically, the laser displacement sensor is fixedly installed on the side of the valve stem 1. The laser beam is perpendicular to the axis of the valve stem 1 and illuminates a specific position 3 on the cylindrical surface of the valve stem 1. The horizontal distance between this position and the center of rotation of the valve stem 1 is... (Requires initial calibration). When valve stem 1 rotates, the measuring point on the surface moves circumferentially, causing a change in the radial distance from the sensor to that point. Let the initial sensor distance be... The distance measured after rotation is Then the change in radial displacement This change can be positive or negative, depending on the direction of rotation of valve stem 1. The angle conversion formula is: The sensor parameters can be selected using the sampling rate. linearity Resolution The model number is specified to ensure the real-time performance and accuracy of dynamic measurements. This conversion model is sensitive to sensor installation angle deviations; in practical applications, it can be corrected by combining multi-point calibration or polynomial fitting.

[0034] Furthermore, in step S2, obtaining the true opening degree of the valve based on the physical zero-position data includes: subtracting the physical zero-position data from the valve stem angle data to obtain the true opening degree; when the valve stem angle data is less than the physical zero-position data, the true opening degree is set to zero.

[0035] Specifically, to calculate the actual valve opening, the physical zero point needs to be determined first through dynamic zero-point calibration. This calibration process must be linked to flow monitoring. Continuous acquisition of valve stem angle is also required. and spray flow rate The sampling frequency is not lower than 10Hz. The valve disc is determined to have disengaged from the valve seat when all three of the following conditions are met simultaneously: spray flow rate... Exceeding the preset threshold (This threshold is taken as the valve's fully open flow rate, for example:) 0.1%, (This can be obtained through factory data or full opening test); the flow rate exceeding the threshold remains stable for more than 3 sampling cycles (e.g., 0.3 seconds) to eliminate false triggering caused by signal noise or instantaneous disturbances; valve stem angle The trend of a continuous, slow increase indicates that the valve is actively opening rather than being stationary or vibrating. The valve stem angle at the moment of triggering is recorded as... Considering the hysteresis error during valve opening and closing, the hysteresis error was statistically analyzed through multiple opening and closing tests (e.g., at least 10 times). And take their average value, the final physical zero point Based on the physical zero position obtained from calibration Actual opening When the calculation result is less than zero, the value is zero. Dynamic zero-point calibration can be performed automatically once before each RCP003 test, or it can be manually triggered by the operator after valve maintenance, to ensure that each test is adjusted based on the latest physical zero point of the current valve state.

[0036] Further, in step S2, the correspondence between the actual opening degree and the spray flow rate data is characterized and the mapping data between the actual opening degree and the spray flow rate data is obtained by: acquiring the fluid dynamics model of the valve, performing flow field simulation under different actual opening degrees, and obtaining the first spray flow rate data corresponding to the actual opening degree; and / or, acquiring the second spray flow rate data collected on the test bench based on different actual opening degrees; when the first spray flow rate data and the second spray flow rate data are acquired, the correspondence between the first spray flow rate data, the second spray flow rate data and the actual opening degree is established to obtain the mapping data between the actual opening degree and the spray flow rate data.

[0037] Specifically, as an optional implementation, fluid dynamics simulation (CFD) can be used to create a refined fluid model of the valve. In terms of geometric simplification, all major features of the valve flow path (valve disc, valve seat, inlet and outlet pipes) are retained, while details with minimal impact on flow, such as bolt holes and sealing grooves, are ignored. A polyhedral mesh is used for mesh generation, with boundary layer refinement at the valve disc-seat gap (e.g., a design value of 0.1 mm). The first mesh layer has a height of 0.005 mm, a growth factor of 1.2, and a total of 15 layers are used to ensure... ,in This represents the dimensionless distance to the wall, i.e., the ratio of the distance from the first-layer mesh node to the wall to the wall friction velocity and kinematic viscosity. It is used to determine whether the near-wall mesh resolution is sufficient to capture the flow characteristics of the viscous sublayer. The total mesh size is 8-10 million. Boundary conditions are set as follows: inlet is a pressure inlet (15.5 MPa), temperature 300℃; outlet is a pressure outlet (15.45 MPa, simulating a 50 kPa pressure difference); the wall is non-slip and adiabatic. The turbulence model uses... The model incorporates reinforced wall treatment. The solver employs... The steady-state solution based on pressure yields a convergence residual of less than 1e-5. The input parameter is the actual opening degree. (In increments of 0.1°, from 0 to 4.5°), output the mass flow rate at each opening degree. (kg / s). This simulation model is applicable to this type of valve under operating conditions of 0-20% opening, 15-16MPa pressure, and 280-320℃ temperature. It is understood that the simulation parameters mentioned above (such as mesh size, step size, pressure value, etc.) are illustrative and can be adjusted according to the specific valve model and operating conditions in actual applications.

[0038] As an alternative implementation method, the experiment calibration was conducted on a high-pressure, high-temperature test bench, using a non-contact measurement system as the opening reference and synchronously acquiring flow data. The bench employed a full-pressure loop, with a pressure of 20 MPa and a temperature of 320°C, and deionized water as the medium. A Coriolis mass flow meter (accuracy 0.35%FS, repeatability 0.1%) was used. Data acquisition employed a NIPXIe system with a sampling rate of 1 kHz, synchronized with the vision system via the PTP protocol (time difference less than 1 ms). The test procedure was as follows: after stabilizing at each opening point for 3 minutes, data was collected for the following 2 minutes, and the average was taken, repeated 3 times. Similarly, the above bench parameters and test procedures are merely examples, and those skilled in the art can make appropriate adjustments according to actual experimental conditions and accuracy requirements.

[0039] Furthermore, the mapping data between the actual spray opening and the actual spray flow rate is obtained by establishing a correspondence between the first and second spray flow rate data and the actual spray opening, including: setting the prior distribution of the parameter to be calibrated; constructing a likelihood function based on the second spray flow rate data; performing posterior sampling using a posterior sampling method, iterating multiple times and discarding aging samples to obtain the posterior distribution of the parameter to be calibrated; taking the posterior mode as the calibration parameter value, correcting the first spray flow rate data based on the calibration parameter value to obtain the calibrated spray flow rate data; and establishing a correspondence between the actual spray opening and the calibrated spray flow rate data to obtain the mapping data.

[0040] Specifically, as a preferred fusion method, Bayesian calibration can be used to fuse simulation and experimental data. The model form is as follows: ,in These are CFD simulation predictions. For parameters to be calibrated (such as flow coefficient, local resistance coefficient). To account for model bias (modeled using Gaussian processes). To observe the noise, among which Indicates observation noise Follows a mean of 0 and a variance of normal distribution The execution steps are as follows: Set the prior distribution. Values ​​are determined based on engineering experience and experimental data. Construct the likelihood function; use the Markov chain Monte Carlo method (No-U-TurnSampler) for posterior sampling, iterating 10,000 times, discarding the first 2,000 iterations as the aging period; take the posterior mode as... Calculate the calibrated flow rate Based on the 2.5% and 97.5% quantiles of the posterior sample, provide 95% confidence intervals. Input is CFD simulation data. and experimental data The output is the calibrated CV curve and the overall expanded uncertainty (U=1.8%, k=2 in the 0-4.5° opening range). It should be noted that the above values ​​(such as the number of iterations, the number of samples during the aging period, and the uncertainty) are specific examples and not limitations of the invention. In other embodiments, other data fusion methods such as weighted averaging, neural network fitting, or Gaussian process regression can be used instead of Bayesian calibration, as long as a mapping relationship between the true opening and the spray flow rate can be established.

[0041] This mapping data forms a high-confidence conversion benchmark from the actual valve opening to the spray flow rate, providing input for the subsequent intelligent decision-making and analysis module. Based on this mapping data, the intelligent decision-making and analysis module will construct a chain-like transmission model of "valve position, flow rate, and system state" (combining physical model and data-driven approach) to predict the system state under different opening degrees and optimize the solution of the minimum opening command.

[0042] Further, in step S3, based on the mapping data, the system state under different opening degrees is predicted, and the prediction results are obtained by: acquiring the cold section temperature, the initial water temperature of the pressure regulator and the heater control parameters; inputting the actual opening degree, spray flow rate data, cold section temperature, initial water temperature of the pressure regulator and the heater control parameters into the thermal hydraulic physical model to predict the spray pipeline temperature difference and the heater power, and obtaining the prediction results.

[0043] Specifically, as an optional implementation, a thermal-hydraulic physical model can describe the thermal mixing process after the sprayed cold water is injected into the pressurizer and the dynamic response of the proportional heater based on the energy conservation equation. The input parameters of this model include: real-time actual opening degree. Instantaneous spray flow rate Q(t) obtained by interpolation of mapped data (CV curve); cold section temperature T_cold (e.g., 293℃); initial water temperature of the pressure regulator. (Available from DCS); PID control parameters K_p, Ki, and K_d for the proportional heater (available from the power plant's DCS configuration). The core equation of the model can be expressed as: in, Where c is the water volume of the pressure regulator, and c_p is the specific heat capacity. U is the density, and U·A is the heat loss coefficient (which can be obtained by identifying historical data). For ambient temperature, This represents a function that describes the change in water temperature within the voltage regulator over time.

[0044] The heater power is calculated based on the PID control law: ; Where T_set is the set temperature and P_max is the heater's maximum power, the model outputs the predicted regulator water temperature. Predicted temperature difference between the two spray lines and the predicted heater power . This indicates integration over time. It should be noted that the specific parameter values ​​in the above equations (such as cold section temperature, water loading, etc.) are illustrative examples and can be set according to the specific operating parameters of the nuclear power unit in practical applications. In other embodiments, the thermal-hydraulic physics model can also use a multi-node lumped parameter model or a one-dimensional distributed parameter model to replace the above lumped parameter model to improve prediction accuracy.

[0045] Furthermore, the actual opening degree, spray flow rate data, cold section temperature, initial water temperature of the pressure regulator, and heater control parameters are input into the thermal-hydraulic physical model to predict the spray pipeline temperature difference and heater power. The prediction results also include: using a long short-term memory neural network to perform residual correction on the prediction results of the thermal-hydraulic physical model; the long short-term memory neural network uses historical experimental data as training samples, and takes the actual opening degree, spray flow rate data, initial temperature, historical power, and historical temperature difference within the time window as inputs, and outputs the prediction residuals; the sum of the predicted values ​​of the thermal-hydraulic physical model and the prediction residuals is used as the final prediction result.

[0046] Specifically, due to the existence of nonlinear, time-varying, or unmodeled dynamics in actual systems that the physical model cannot fully capture (such as thermal stratification effects, sensor response delays, uneven distribution of heat loss in pipelines, etc.), relying solely on the physical model may produce systematic prediction biases. As an optional implementation, a long short-term memory neural network can be used to perform residual correction on the prediction results of the physical model. For example, the training data can be collected from RCP003 test data (such as at least 30 valid test sets) from multiple past overhauls (e.g., 5 times), including θ(t), Q(t), and measured temperature difference. ,power Initial conditions, etc., with a sampling rate of, for example, 1Hz. The input dimension can be set to a time window length of 60 steps, with features for each time step as follows. That is, the input tensor shape is The network structure could be, for example, a two-layer LSTM with 50 and 25 hidden nodes respectively. The output layer is fully connected (1 node). Training parameters can be set as follows: optimizer Adam, learning rate 0.001, batch size 32, loss function is root mean square error (RMSE), 5-fold temporal cross-validation is used, training epochs are 500, and early stopping is employed to prevent overfitting. The residual correction logic is as follows: ,in, The residuals of the network prediction (i.e., the deviation between the physical model and the actual system) will ultimately be Substitute these into subsequent optimization constraints. It should be noted that the above network structure and training parameters are for illustrative purposes only. In other implementations, alternative models such as gated recurrent units (GRU), temporal convolutional networks (TCN), or Transformers can be used to achieve residual correction. The amount of training data and the sampling frequency can also be adjusted according to the actual data situation.

[0047] Further, in step S3, determining the minimum opening instruction that meets the preset operating conditions based on the prediction results includes: constructing a constrained optimization problem with the objective function of minimizing the actual opening and the constraints of the absolute value of the temperature difference in the spray pipeline not exceeding a first threshold and the ratio of the heater power to the maximum power being between a first ratio and a second ratio; solving the optimization problem using an optimization algorithm to obtain the minimum actual opening that meets the constraints, which is then used as the minimum opening instruction.

[0048] Specifically, as an optional implementation, a single-objective constrained optimization problem can be constructed as follows. The decision variable is the spray valve opening. (Continuous variables, e.g., range 0°-4.5°). The objective function is to minimize... ,Right now The constraint can be set as follows: the absolute value of the temperature difference does not exceed a first threshold (e.g., 5℃), i.e. The heater power as a percentage of the maximum power is between the first and second percentages (e.g., 45% to 55%), i.e. , ; and opening boundary constraints , ( (≤4.5°). Since the objective function is a single-variable minimization problem, and the prediction model may have nonlinear, nonconvex, or even discontinuous characteristics, the penalty function method can be used to transform the constrained problem into an unconstrained problem: ,in The penalty factor is set to 1000 (e.g., 1000). Then, an optimization algorithm is used to solve the problem. For example, NSGA-II (a non-dominated sorting genetic algorithm with an elitist strategy) can be used for multi-objective optimization. Although this algorithm is often used for multi-objective problems, it is also suitable for the single-objective constrained optimization scenario in this scheme. The parameters can be set as follows: population size 100, number of generations 500, crossover probability 0.8, mutation probability 0.1, and elitist retention number 10. The optimal solution in the Pareto front is then output. The recommended opening degree refers to the smallest possible opening degree among all feasible opening degrees that satisfy the temperature difference ≤ 5℃ and power fluctuation within the range of 45%-55%. It should be noted that the above thresholds (e.g., 5℃, 45%, 55%), opening degree range (0°-4.5°), and algorithm parameters are illustrative examples and can be adjusted according to the technical specifications and specific operating conditions of the nuclear power plant in actual applications. In other implementations, the golden section search, Newton's method, simulated annealing, or particle swarm optimization can be used to replace the genetic algorithm for optimization. Taking the optical vision solution as a specific example, a ring target can be attached to the middle of the RCP001VP valve stem, a camera and a ring light source can be installed, and connected to an edge computer. After starting the system, the valve is slowly opened, and the system automatically performs dynamic zero-point calibration and records the physical zero point. =0.15°. Input the initial parameters for this experiment (e.g., cold section temperature 293°C, initial temperature difference 0°C), and the system will call the chain model to pre-generate the opening degree-feasibility table to obtain the recommended opening degree. =1.8°. The operator set the opening according to the recommended setting. After stabilizing for 5 minutes, the measured temperature difference was 3.2°C and the power was 48%, meeting the requirements. The system automatically saved this data for the next model iteration.

[0049] This invention provides a method for regulating the spray valve of a pressurizer, which offers significant advantages over existing technologies. This application employs a diversified, non-contact measurement approach (including vision, laser, eddy current, and other methods), enabling flexible adaptation to different field conditions and significantly broadening the scope of application. It covers all technical routes based on the core concept of "non-contact measurement + dynamic zero-point calibration." Regarding standardization and reliability, this application uses a data-driven approach with unified standards. Regardless of the specific sensor used, the output "physical zero point" and "actual opening" benchmarks remain consistent, achieving objective quantification of maintenance standards. In terms of technology lifecycle, the core of this application lies in its methodology. Specific sensing technologies can be iteratively updated with technological advancements (e.g., from 720p cameras to 4K cameras). This method can be widely applied in multi-site nuclear power units, offering significant economic benefits, reduced operation and maintenance costs, and improved safety and reliability.

[0050] The main contributions of this application to power plant productivity are as follows: In terms of direct economic benefits, the current RCP003 test currently occupies an average of 3-5 hours of the critical path. After applying this technology, the test time can be stably controlled within 2 hours. A single overhaul can release at least 2 hours of critical path window, providing a valuable buffer for other critical path activities and significantly reducing the risk of overall overhaul delays due to test delays, resulting in huge economic value. In terms of reducing operation and maintenance costs, this technology does not require multiple senior operators or maintenance engineers to repeatedly debug for a long time, reducing manpower consumption. At the same time, precise control avoids unnecessary frequent and large-scale valve movements, extending the service life of the spray valve and its actuator, and reducing spare parts procurement and replacement costs. In terms of safety and reliability, accurate understanding of valve characteristics provides a more solid equipment-level guarantee for the stability of the primary loop pressure boundary.

[0051] It is understood that the above embodiments only illustrate preferred embodiments of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can freely combine the above technical features without departing from the concept of the present invention, and can also make several modifications and improvements, all of which fall within the protection scope of the present invention. Therefore, all equivalent transformations and modifications made with respect to the scope of the claims of the present invention should fall within the scope of the claims of the present invention.

Claims

1. A method for regulating a voltage regulator spray valve, characterized in that, Includes the following steps: Step S1: Obtain valve stem angle data and spray flow rate data, and take the valve stem angle data corresponding to the first time the spray flow rate data exceeds the preset flow rate threshold as the physical zero point data; Step S2: Obtain the actual valve opening based on the physical zero-position data, and perform characteristic calibration on the correspondence between the actual opening and the spray flow data to obtain the mapping data between the actual opening and the spray flow data; Step S3: Based on the mapping data, predict the system state under different opening degrees, obtain the prediction results, and determine the minimum opening degree command that satisfies the preset operating conditions according to the prediction results.

2. The method for regulating the spray valve of a voltage regulator according to claim 1, characterized in that, The voltage regulator spray valve includes a valve stem and a valve disc; the valve stem is used to drive the valve disc to complete the opening adjustment, and a concentric annular target is provided on the cylindrical surface of the valve stem; The valve stem angle data is obtained in the following way: The image of the concentric ring target is acquired by the image acquisition device, and the valve stem angle data is calculated based on the displacement of the center of the ellipse formed by the target projection; wherein the image acquisition device is located next to the valve stem, and its lens is aimed at the concentric ring target; Alternatively, the radial displacement data of the valve stem cylindrical surface collected by the laser displacement sensor can be obtained, and the valve stem angle data can be calculated. The laser displacement sensor is installed on the outer periphery of the valve stem, with its detection end facing the valve stem surface.

3. The method for regulating the spray valve of a voltage regulator according to claim 2, characterized in that, The valve stem angle data calculated based on the center displacement of the ellipse formed by the target projection includes: After preprocessing the image, the edge points of the target projection are extracted; Ellipse fitting is performed on the edge points to obtain the coordinates of the ellipse center; The center coordinates of the ellipse are located at the sub-pixel level; The horizontal displacement of the ellipse center coordinates within the image plane is calculated. The pixel equivalent is obtained through camera calibration. Combined with the valve stem radius, the horizontal displacement is converted into the valve stem angle data.

4. The method for regulating the spray valve of a voltage regulator according to claim 2, characterized in that, The laser beam of the laser displacement sensor is perpendicular to the axis of the valve stem and illuminates a preset position on the cylindrical surface of the valve stem. The preset position is at a preset horizontal distance from the rotation center of the valve stem. The step of obtaining the radial displacement data of the valve stem cylindrical surface collected by the laser displacement sensor and converting it into the valve stem angle data includes: Acquire radial displacement data collected by the laser displacement sensor, wherein the radial displacement data includes the change in radial distance from the laser displacement sensor to the valve stem surface; The valve stem angle data is obtained by converting the radial distance change to the preset horizontal distance through arctangent calculation.

5. The method for regulating the spray valve of a voltage regulator according to claim 1, characterized in that, In step S2, obtaining the actual valve opening based on the physical zero-position data includes: Subtract the physical zero-position data from the valve stem angle data to obtain the actual opening degree; When the valve stem angle data is less than the physical zero position data, the actual opening degree is set to zero.

6. The method for regulating the spray valve of a voltage regulator according to claim 1, characterized in that, In step S2, the characteristic calibration of the correspondence between the actual opening degree and the spray flow rate data to obtain the mapping data between the actual opening degree and the spray flow rate data includes: Obtain the fluid dynamics model of the valve, perform flow field simulation under different actual opening degrees, and obtain the first spray flow rate data corresponding to the actual opening degree; And / or, acquire second spray flow data collected on the test bench based on different actual opening degrees; When the first spray flow rate data and the second spray flow rate data are obtained, a correspondence is established between the first spray flow rate data and the second spray flow rate data and the actual opening degree to obtain the mapping data between the actual opening degree and the spray flow rate data.

7. The method for regulating the spray valve of a voltage regulator according to claim 6, characterized in that, The step of establishing a correspondence between the first spray flow rate data, the second spray flow rate data, and the actual spray opening to obtain the mapping data between the actual spray opening and the spray flow rate data includes: Set the prior distribution of the parameters to be calibrated; Construct a likelihood function based on the second spray flow rate data; The posterior sampling method is used to perform posterior sampling, which is iterated multiple times and aging samples are discarded to obtain the posterior distribution of the parameter to be calibrated; The posterior mode is taken as the calibration parameter value, and the first spray flow rate data is corrected based on the calibration parameter value to obtain the calibrated spray flow rate data; Establish the correspondence between the actual opening degree and the calibrated spray flow rate data to obtain the mapping data.

8. The method for regulating the spray valve of a voltage regulator according to claim 1, characterized in that, In step S3, the prediction of the system state under different opening degrees based on the mapping data, and the resulting prediction, include: Obtain the cold section temperature, the initial water temperature of the voltage regulator, and the control parameters of the heater; The actual opening degree, the spray flow rate data, the cold section temperature, the initial water temperature of the pressure regulator, and the heater control parameters are input into the thermal hydraulic physical model to predict the spray pipeline temperature difference and the heater power, thus obtaining the prediction results.

9. The method for regulating the spray valve of a voltage regulator according to claim 8, characterized in that, The process of inputting the actual opening degree, the spray flow rate data, the cold section temperature, the initial water temperature of the pressure regulator, and the heater control parameters into the thermal-hydraulic physical model to predict the spray pipeline temperature difference and heater power, and obtaining the prediction results, further includes: The prediction results of the thermal-hydraulic physical model are corrected using a long short-term memory neural network. The long short-term memory neural network uses historical experimental data as training samples, and takes the actual opening degree, spray flow rate data, initial temperature, historical power and historical temperature difference within the time window as inputs, and outputs the prediction residual. The sum of the predicted value from the thermal-hydraulic physical model and the predicted residual is taken as the final prediction result.

10. The method for regulating the spray valve of a voltage regulator according to claim 8, characterized in that, In step S3, determining the minimum opening command that satisfies the preset operating conditions based on the prediction result includes: With minimizing the actual opening degree as the objective function, and with the constraints that the absolute value of the temperature difference of the spray pipeline does not exceed a first threshold and the ratio of the heater power to the maximum power is between a first ratio and a second ratio, a constrained optimization problem is constructed. The optimization problem is solved using an optimization algorithm to obtain the minimum true opening that satisfies the constraints, which is then used as the minimum opening instruction.

Images