Method and device for determining valve element structure of regulating valve
By optimizing the valve core structure of the control valve through BP neural network, the problem of poor flow characteristic matching in the existing technology was solved, realizing the design of a highly efficient and accurate control valve for uranium enrichment systems, and improving design efficiency and accuracy.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- CHINA NAT NUCLEAR URANIUM ENRICHMENT
- Filing Date
- 2025-12-31
- Publication Date
- 2026-05-08
AI Technical Summary
The existing control valve core is not well matched to the target flow characteristics, resulting in the inability to meet extreme performance requirements in terms of control accuracy, stability and dynamic response, making it difficult to adapt to the high requirements of uranium enrichment systems.
By acquiring a pre-trained valve core structure design model, a BP neural network is used to establish a mapping relationship between the flow coefficient and the valve core radius. Combined with parametric simulation and model training, the valve core structure is optimized to match the target flow characteristics.
It achieves efficient and precise matching of valve core structure to target flow characteristics, improves the design efficiency and accuracy of control valve, meets the different operating conditions of uranium enrichment system, and controls the simulation flow coefficient error within 10%.
Smart Images
Figure CN121997808A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of uranium enrichment technology, specifically relating to a method and apparatus for determining the structure of a control valve core. Background Technology
[0002] Control valves are critical regulating devices in uranium enrichment systems, used for precise control of pressure and flow. Their performance directly affects the operational stability and safety of the entire system. Due to the special nature of the uranium enrichment process media and the high sensitivity of the main unit to pressure fluctuations, the system imposes extreme performance requirements on control valves that far exceed ordinary industrial standards.
[0003] Currently, the valve cores of specialized control valves designed to meet extreme performance requirements are poorly matched to the target flow characteristics. This results in key performance aspects such as control accuracy, stability, and dynamic response failing to fully meet control demands and making it difficult to adapt to the higher control performance requirements of future cascaded system expansion. Traditional valve core design methods are mainly based on the flow area principle for iterative development, heavily reliant on manual experience and trial-and-error adjustments. This leads to long design cycles, high costs, and difficulty in finding the globally optimal solution in complex multi-parameter design spaces, becoming a bottleneck restricting the independent development of high-performance valves for uranium enrichment plants. Summary of the Invention
[0004] The purpose of this application is to provide a method and apparatus for determining the valve core structure of a control valve, thereby solving the problem in the prior art where the valve core of a special control valve that meets extreme performance requirements does not match the target flow characteristics well.
[0005] The technical solution to achieve the purpose of this application is as follows:
[0006] The first aspect of this application provides a method for determining the valve core structure of a control valve, the method comprising:
[0007] Obtain a pre-obtained valve core structure design model; the valve core structure design model is trained based on a pre-obtained sample set.
[0008] The correspondence between the target opening degree and the target flow coefficient of the regulating valve is input into the valve core structure design model to obtain the valve core radius of the regulating valve under the target opening degree;
[0009] The sample set includes a one-to-one correspondence between the valve opening degree, flow coefficient, and valve core radius; the sample set is obtained by simulation based on the three-dimensional model of the regulating valve.
[0010] Optionally, the sample set is obtained according to the following steps:
[0011] Obtain a three-dimensional model of the control valve; the three-dimensional model is determined based on the correspondence between the desired opening degree and the flow coefficient;
[0012] Obtain the sample opening degree and sample valve core radius;
[0013] The sample opening degree and the sample valve core radius are input into the three-dimensional model to derive the corresponding flow coefficient, thus obtaining the sample set.
[0014] Optionally, the three-dimensional model of the regulating valve is obtained according to the following steps:
[0015] The stroke [0, L] of the regulating valve core is divided into n discrete opening degrees l. i And obtain each opening degree l i Corresponding flow coefficient Kv i L represents the maximum stroke of the valve core, and n is a positive integer; i = 1, 2, ..., n;
[0016] Based on the physical relationship between the flow coefficient and the throttling area, the opening l can be deduced. i The corresponding valve core radius r i ;
[0017] According to each opening degree l i The lower valve core radius r i The three-dimensional model is then established.
[0018] Optionally, obtaining the sample opening degree and sample valve core radius specifically includes:
[0019] According to each opening degree l i The lower valve core radius r i Determine the opening degree l i The sampling range for the valve core radius;
[0020] Each opening l i As the sample opening, Latin hypercube sampling is used to sample from the corresponding valve core radius sampling interval, which is then used as the corresponding sample valve core radius.
[0021] Optionally, the valve core structure design model is obtained through the following steps:
[0022] A BP neural network surrogate model is trained using the sample set to establish a fast mapping relationship from valve core radius to flow coefficient, thereby obtaining the valve core structure design model.
[0023] Optionally, during the training of the BP neural network agent model, the optimal fitness particles of each generation are selected according to the criterion of elite retention, and the three-dimensional model is used for verification.
[0024] Optionally, the verification using the three-dimensional model further includes:
[0025] Based on the verified data, update the sample set and the BP neural network proxy model.
[0026] A second aspect of this application provides an apparatus for determining the structure of a control valve core, the apparatus comprising:
[0027] The model acquisition module is used to acquire a pre-obtained valve core structure design model; the valve core structure design model is trained based on a pre-obtained sample set.
[0028] The radius acquisition module is used to input the correspondence between the target opening degree and the target flow coefficient of the regulating valve into the valve core structure design model to obtain the valve core radius of the regulating valve under the target opening degree;
[0029] The sample set includes a one-to-one correspondence between the valve opening degree, flow coefficient, and valve core radius; the sample set is obtained by simulation based on the three-dimensional model of the regulating valve.
[0030] A third aspect of this application provides a computer-readable storage medium storing computer code that, when executed, implements any one of the methods for determining the valve core structure of a control valve as provided in the first aspect of this application.
[0031] A fourth aspect of this application provides a controller, including a memory and a processor; the memory stores computer code, and when the processor executes the computer code, it implements any one of the methods for determining the valve core structure of a control valve as provided in the first aspect of this application.
[0032] The beneficial technical effects of this application are as follows:
[0033] This application provides a method and apparatus for determining the valve core of a control valve. The method includes: acquiring a pre-obtained valve core structure design model; training the valve core structure design model based on a pre-obtained sample set; inputting the correspondence between the target opening degree and the target flow coefficient of the control valve into the valve core structure design model to obtain the valve core radius of the control valve at the target opening degree; wherein, the sample set includes a one-to-one correspondence between the opening degree, the flow coefficient, and the valve core radius; the sample set is obtained by simulation based on a three-dimensional model of the control valve. This application combines parametric simulation and model training to achieve efficient and accurate matching of the valve core structure to the target flow characteristics while ensuring optimization accuracy, thereby improving the design capability of a series of control valves and meeting the regulation needs of different system scenarios. This application significantly improves design efficiency while ensuring calculation accuracy, thus effectively supporting the professional and serialized design needs of control valves in uranium enrichment plants under different operating conditions. Application results show that the error between the simulated flow coefficient and the target value of the control valve determined by this application can be controlled within 10%, significantly improving the matching accuracy of valve flow characteristics and overall design efficiency. Attached Figure Description
[0034] Figure 1 A flowchart illustrating a method for determining the valve core structure of a control valve, provided as an embodiment of this application;
[0035] Figure 2 A schematic diagram of the valve core stroke in a method for determining the valve core structure of a regulating valve, provided in an embodiment of this application;
[0036] Figure 3 This is a schematic diagram of the flow area corresponding to a certain opening degree of the valve core in a method for determining the valve core structure of a regulating valve provided in an embodiment of this application. Detailed Implementation
[0037] To enable those skilled in the art to better understand this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the embodiments described below are only a part of the embodiments of this application, and not all of them. Based on the embodiments described in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0038] See Figure 1 The figure is a flowchart illustrating a method for determining the valve core structure of a regulating valve according to an embodiment of this application.
[0039] This application provides a method for determining the valve core structure of a regulating valve, comprising:
[0040] Step S101: Obtain the pre-obtained valve core structure design model.
[0041] In this embodiment, the valve core structure design model is trained based on a pre-obtained sample set; the sample set includes a one-to-one correspondence between opening degree, flow coefficient, and valve core radius; the sample set is simulated based on the three-dimensional model of the regulating valve. The specific methods for obtaining the sample set and the three-dimensional model will be explained in detail below, but will not be elaborated upon here.
[0042] Step S102: Input the correspondence between the target opening degree and the target flow coefficient of the regulating valve into the valve core structure design model to obtain the valve core radius of the regulating valve under the target opening degree.
[0043] It is understandable that determining the valve core radius at each opening degree yields the specific structure of the valve core. This application's embodiment combines parametric simulation and model training to achieve efficient and precise matching of the valve core structure to the target flow characteristics while ensuring optimization accuracy. This enhances the design capabilities of a series of control valves and meets the regulation needs of different system scenarios.
[0044] In some possible implementations of the embodiments of this application, the sample set can be obtained according to the following steps:
[0045] Obtain a three-dimensional model of the control valve; the three-dimensional model is determined based on the correspondence between the desired opening degree and the flow coefficient;
[0046] Obtain the sample opening degree and sample valve core radius;
[0047] The sample opening degree and the sample valve core radius are input into the three-dimensional model to derive the corresponding flow coefficient, thus obtaining the sample set.
[0048] In some possible implementations of the embodiments of this application, the three-dimensional model of the regulating valve can be obtained according to the following steps:
[0049] The stroke [0, L] of the regulating valve core is divided into n discrete opening degrees l. i ,like Figure 2 As shown, obtain each opening l i Corresponding flow coefficient Kv i L represents the maximum stroke of the valve core, and n is a positive integer; i = 1, 2, ..., n;
[0050] Based on the physical relationship between the flow coefficient and the throttling area, the opening l can be deduced. i The corresponding valve core radius r i ;
[0051] According to each opening degree l i The lower valve core radius r iThe three-dimensional model is then established.
[0052] Figure 2 An example is shown with an opening degree l i The division method is as follows: L represents the valve core stroke, and D represents the valve passage diameter. Each opening degree l... i Corresponding flow coefficient Kv i The design goal for the valve core design.
[0053] In one example, obtaining the sample opening degree and sample valve core radius may specifically include:
[0054] According to each opening degree l i The lower valve core radius r i Determine the opening degree l i The sampling range for the valve core radius;
[0055] Each opening l i As the sample opening, Latin hypercube sampling is used to sample from the corresponding valve core radius sampling interval, which is then used as the corresponding sample valve core radius.
[0056] In some possible implementations of the embodiments of this application, the valve core structure design model can be obtained through the following steps:
[0057] A BP neural network surrogate model is trained using the sample set to establish a fast mapping relationship from valve core radius to flow coefficient, thereby obtaining the valve core structure design model.
[0058] In one example, during the training of the BP neural network agent model, the optimal fitness particles of each generation can be selected according to the criterion of elite retention, and the three-dimensional model can be used for verification.
[0059] In another example, the verification using the three-dimensional model may further include:
[0060] Based on the verified data, update the sample set and the BP neural network proxy model.
[0061] The following describes in detail, with a specific example, a method for determining the valve core structure of a regulating valve provided in the embodiments of this application.
[0062] This application provides a method for determining the valve core structure of a control valve. The designed control valve has the following requirements: channel diameter D = 30 mm, valve core stroke L = 18 mm, maximum flow coefficient Kvmax = 13.5, turnability ratio R = 50, and is a single-seat plunger type control valve with an inherent flow characteristic of equal percentage. The optimization design process of the valve core parameters is described step by step below:
[0063] The flow characteristic curve is uniformly discretized into 11 openings at intervals of 0.1 openings from 0 to 1 opening. The Kv corresponding to each opening is... i for:
[0064] i is 0, 1, ...
[0065] like Figure 3 As shown in Table 1, the initial valve core radius for each opening degree is calculated according to the principle of equivalent flow area of the control valve.
[0066] Table 1. Opening Degree and Corresponding Valve Core Radius
[0067]
[0068] Figure 3 In this context, D is the diameter of the valve flow channel cross section, d is the valve core diameter corresponding to the point with the minimum lateral area at that opening degree, m is the generatrix length of the frustum of the fluid flow area, and point (x,y) is the point on the valve core profile corresponding to the minimum flow area at a certain opening degree.
[0069] A three-dimensional model of the control valve is established, and the valve core radius, channel diameter, and total stroke are defined parametrically. The model is then imported into fluid simulation software, a parametric simulation project is established, and the flow rate in the simulation results is defined parametrically. One parametric simulation yields one flow characteristic curve.
[0070] Latin hypercube sampling was performed on 12 design parameters of the valve core, including 11 valve core radius parameters r. i and 1 length parameter l i Each radius parameter is uniformly distributed within its given range, and the length parameter is l in Table 1. i The discrete points are selected. The sampling count is set to 400, generating 400 feature sample points that uniformly cover the design space. CFD simulation is performed on the valve core geometry corresponding to each sample point to obtain its flow characteristic data. Finally, a sample library D containing 400 sets of valve core radius and flow characteristic data is constructed.
[0071] This step requires updating startup parameters using a batch processing method. The key pseudocode is as follows:
[0072] *****************************************************************************************************
[0073] Simulation software batch startup code:
[0074] "%AWP_ROOT241%\Framework\bin\win64\runwb2.exe" -F ".\Simulation_Project_Path" -BIR ".\Execution_Path" #Simulation_Project_Path and Execution_Path are the path to the parametric simulation project and the path to the parameters, respectively.
[0075] Parametric simulation driver pseudocode:
[0076] Parameters_Radius = [(14.944,14.923,14.885,14.831,14.732,14.625,14.563,14.256,13.795,13.235.11.179)] # Valve structure parameter group for a certain transmission
[0077] Parameters_Names = [] # Names of parametric simulation design points, used for parametric driving.
[0078] Parameters_Positions = [(10.8, 0.6)] # Valve core position, corresponding to valve opening degree
[0079] for position in Parameters_Positions: # Complete the plotting of the valve flow coefficient curve in one loop
[0080] Parameters_Radius=[(14.944,14.923,14.885,14.831,14.732,14.625,14.563,14.256,13.795,13.235.11.179)]
[0081] Design_Point = Parameters_Radius.append(position) # Valve structural parameters + valve core position
[0082] dp = Parameters.CreateDesignPoint("DesignPoint") # Create a design point
[0083] for iin range(length(Design_Point)):
[0084] dp.Set_Parameter(Design_Point,Parameters_Names[i]) # Parametric simulation design point
[0085] UpdateAllDesignPoints(DesignPoints = [dp]) # Parametric Simulation Driver
[0086] **********************************************************************************************************************
[0087] Combining batch parametric simulation with particle swarm optimization (PSO) algorithm can optimize valve core parameters. The fitness function used in the algorithm is:
[0088]
[0089] Where, ω i It is a normalized constraint weighted value, and the product of the square of the weighted coefficient for each opening and the corresponding design flow rate is 1; Q is the simulated flow rate value corresponding to the i-th opening in the t-th generation; i It is the flow rate calculated based on the design values.
[0090] However, this method requires a large number of flow field simulation calculations, which is time-consuming. A neural network flow prediction proxy model can be established to replace a large number of flow field simulation calculations.
[0091] A neural network prediction model is built based on a standardized dataset. A backpropagation neural network traffic prediction model is constructed using sklearn and keras. The model structure is as follows:
[0092] • Neural network model structure;
[0093] • Input layer: The valve parameter dataset with 13 features is normalized before being input;
[0094] • Hidden layers: The first layer has 64 neurons, the second layer has 32 neurons, and the ReLU function is used.
[0095] • Output layer: A linear neuron used for direct output;
[0096] • Optimizer: Adam adaptive moment estimator;
[0097] • Loss function: Mean Square Error (MSE);
[0098] • Prediction and inverse normalization.
[0099] To verify the accuracy of the prediction model, a portion of non-sample space data was selected for comparison, with the prediction loss value being less than 10.-3 The model matches the expected predictions.
[0100] To ensure the accuracy of the optimization results, simulation-based verification was performed at the global optimal solution of each generation of the particle swarm optimization algorithm. The maximum error between the simulated flow coefficient and the design value was 9.72%. This method successfully optimized the valve core profile of the control valve with high efficiency and high precision.
[0101] This application combines parametric simulation and model training to achieve efficient and accurate matching of the valve core structure to the target flow characteristics while ensuring optimization accuracy. This enhances the design capability of a series of control valves and meets the regulation needs of different system scenarios. This application significantly improves design efficiency while maintaining computational accuracy, thus effectively supporting the specialized and serialized design needs of control valves in uranium enrichment plants under different operating conditions. Application results show that the error between the simulated flow coefficient and the target value of the control valve determined by this application can be controlled within 10%, significantly improving the matching accuracy of valve flow characteristics and overall design efficiency.
[0102] Based on the method for determining the valve core structure of a control valve provided in the above embodiments, this application also provides an apparatus for determining the valve core structure of a control valve.
[0103] This application provides an embodiment of a device for determining the valve core structure of a regulating valve, comprising:
[0104] The model acquisition module is used to acquire a pre-obtained valve core structure design model; the valve core structure design model is trained based on a pre-obtained sample set.
[0105] The radius acquisition module is used to input the correspondence between the target opening degree and the target flow coefficient of the regulating valve into the valve core structure design model to obtain the valve core radius of the regulating valve under the target opening degree;
[0106] The sample set includes a one-to-one correspondence between the valve opening degree, flow coefficient, and valve core radius; the sample set is obtained by simulation based on the three-dimensional model of the regulating valve.
[0107] In one example, the sample set can be obtained according to the following steps:
[0108] Obtain a three-dimensional model of the control valve; the three-dimensional model is determined based on the correspondence between the desired opening degree and the flow coefficient;
[0109] Obtain the sample opening degree and sample valve core radius;
[0110] The sample opening degree and the sample valve core radius are input into the three-dimensional model to derive the corresponding flow coefficient, thus obtaining the sample set.
[0111] In another example, the three-dimensional model of the control valve can be obtained according to the following steps:
[0112] The stroke [0, L] of the regulating valve core is divided into n discrete opening degrees l. i And obtain each opening degree l i Corresponding flow coefficient Kv i L represents the maximum stroke of the valve core, and n is a positive integer; i = 1, 2, ..., n;
[0113] Based on the physical relationship between the flow coefficient and the throttling area, the opening l can be deduced. i The corresponding valve core radius r i ;
[0114] According to each opening degree l i The lower valve core radius r i The three-dimensional model is then established.
[0115] As an example, obtaining the sample opening degree and sample valve core radius may specifically include:
[0116] According to each opening degree l i The lower valve core radius r i Determine the opening degree l i The sampling range for the valve core radius;
[0117] Each opening l i As the sample opening, Latin hypercube sampling is used to sample from the corresponding valve core radius sampling interval, which is then used as the corresponding sample valve core radius.
[0118] In some possible implementations of the embodiments of this application, the valve core structure design model can be obtained through the following steps:
[0119] A BP neural network surrogate model is trained using the sample set to establish a fast mapping relationship from valve core radius to flow coefficient, thereby obtaining the valve core structure design model.
[0120] In one example, during the training of the BP neural network agent model, the optimal fitness particles of each generation are selected according to the criterion of elite retention, and the three-dimensional model is used for verification.
[0121] In another example, the verification using the 3D model further includes:
[0122] Based on the verified data, update the sample set and the BP neural network proxy model.
[0123] This application combines parametric simulation and model training to achieve efficient and accurate matching of the valve core structure to the target flow characteristics while ensuring optimization accuracy. This enhances the design capability of a series of control valves and meets the regulation needs of different system scenarios. This application significantly improves design efficiency while maintaining computational accuracy, thus effectively supporting the specialized and serialized design needs of control valves in uranium enrichment plants under different operating conditions. Application results show that the error between the simulated flow coefficient and the target value of the control valve determined by this application can be controlled within 10%, significantly improving the matching accuracy of valve flow characteristics and overall design efficiency.
[0124] Based on the method and apparatus for determining the valve core structure of a control valve provided in the above embodiments, this application also provides a computer-readable storage medium storing computer code thereon. When the computer code is executed, it implements any one of the methods for determining the valve core structure of a control valve provided in the above embodiments.
[0125] Based on the method and apparatus for determining the valve core structure of a control valve provided in the above embodiments, this application also provides a controller, including a memory and a processor; the memory stores computer code, and when the processor executes the computer code, it implements any one of the methods for determining the valve core structure of a control valve provided in the above embodiments.
[0126] The present application has been described in detail above with reference to the accompanying drawings and embodiments. However, the present application is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present application. All content not described in detail in this application can be derived from existing technology.
Claims
1. A method for determining the valve core structure of a regulating valve, characterized in that, The method includes: Obtain a pre-obtained valve core structure design model; the valve core structure design model is trained based on a pre-obtained sample set. The correspondence between the target opening degree and the target flow coefficient of the regulating valve is input into the valve core structure design model to obtain the valve core radius of the regulating valve under the target opening degree; The sample set includes a one-to-one correspondence between the valve opening degree, flow coefficient, and valve core radius; the sample set is obtained by simulation based on the three-dimensional model of the regulating valve.
2. The method for determining the valve core structure of a regulating valve according to claim 1, characterized in that, The sample set is obtained according to the following steps: Obtain a three-dimensional model of the control valve; the three-dimensional model is determined based on the correspondence between the desired opening degree and the flow coefficient; Obtain the sample opening degree and sample valve core radius; The sample opening degree and the sample valve core radius are input into the three-dimensional model to derive the corresponding flow coefficient, thus obtaining the sample set.
3. The method for determining the valve core structure of a regulating valve according to claim 2, characterized in that, The three-dimensional model of the regulating valve is obtained according to the following steps: The stroke [0, L] of the regulating valve core is divided into n discrete opening degrees l. i And obtain each opening degree l i Corresponding flow coefficient Kv i L represents the maximum stroke of the valve core, and n is a positive integer; i = 1, 2, ..., n; Based on the physical relationship between the flow coefficient and the throttling area, the opening l can be deduced. i The corresponding valve core radius r i ; According to each opening degree l i The lower valve core radius r i Establish the three-dimensional model.
4. The method for determining the valve core structure of a regulating valve according to claim 3, characterized in that, The acquisition of the sample opening and sample valve core radius specifically includes: According to each opening degree l i The lower valve core radius r i Determine the opening degree l i The sampling range of the valve core radius; Each opening l i As the sample opening, Latin hypercube sampling is used to sample from the corresponding valve core radius sampling interval, which is then used as the corresponding sample valve core radius.
5. The method for determining the valve core structure of a regulating valve according to any one of claims 1-4, characterized in that, The valve core structure design model is obtained through the following steps: A BP neural network surrogate model is trained using the sample set to establish a fast mapping relationship from valve core radius to flow coefficient, thereby obtaining the valve core structure design model.
6. The method for determining the valve core structure of a regulating valve according to claim 5, characterized in that, During the training of the BP neural network agent model, the optimal fitness particles of each generation are selected according to the criterion of elite retention, and the three-dimensional model is used for verification.
7. The method for determining the valve core structure of a regulating valve according to claim 6, characterized in that, The verification using the three-dimensional model then includes: Based on the verified data, update the sample set and the BP neural network proxy model.
8. An apparatus for determining the structure of a control valve core, characterized in that, The device includes: The model acquisition module is used to acquire a pre-obtained valve core structure design model; the valve core structure design model is trained based on a pre-obtained sample set. The radius acquisition module is used to input the correspondence between the target opening degree and the target flow coefficient of the regulating valve into the valve core structure design model to obtain the valve core radius of the regulating valve under the target opening degree; The sample set includes a one-to-one correspondence between the valve opening degree, flow coefficient, and valve core radius; the sample set is obtained by simulation based on the three-dimensional model of the regulating valve.
9. A computer-readable storage medium, characterized in that, It stores computer code that, when executed, implements a method for determining the valve core structure of a control valve as described in any one of claims 1-7.
10. A controller, characterized in that, It includes a memory and a processor; the memory stores computer code, and when the processor executes the computer code, it implements a method for determining the valve core structure of a control valve as described in any one of claims 1-7.