Simulation and actual measurement data fused explosion valve pipeline loop pressure detection method and device

Abstract

The invention discloses a simulation and actual measurement data fused explosion valve pipeline loop pressure detection method and device, and relates to the technical field of explosion valves. The problem that a traditional experimental measurement method can only obtain pressure data of a test point and cannot obtain pressure data of any point in a pipeline is solved. Through node division of a pipeline loop, simulation data and actual measurement data are obtained; the method comprises the following steps: inputting measured data as high-fidelity data, inputting simulation data as low-fidelity data, dividing and building a multi-fidelity agent model of a pipeline loop according to a proportion, processing the multi-fidelity agent model to obtain a pressure curve of the pipeline loop, and obtaining pressure data of any point in the pipeline loop through the pressure curve. According to the method, the proportion of the actual measurement group data to the simulation group data is designed to be between 1: 5 and 1: 10, so that the precision of the pipeline loop multi-fidelity agent model is ensured, and the cost is controlled to be within a certain range. The method is suitable for pressure detection of the explosion valve pipeline loop.

Description

Technical Field

[0001] The invention relates to the technical field of bursting valves, and in particular to pressure simulation of a bursting valve pipeline loop and construction of an agent model to realize pipeline pressure calculation. Background Art

[0002] The bursting valve is an important component of the safety system of a nuclear power plant. Its main function is to trigger the bursting unit through a signal when a serious accident occurs in the nuclear power plant, generate high-pressure gas to push the piston in the valve to break the shear cover, thereby quickly opening the valve and protecting the nuclear island.

[0003] The safety of the bursting valve pipeline loop is the premise for ensuring the normal operation of the bursting valve. In order to measure the pressure of the bursting valve pipeline loop, it is necessary to perform pressure detection on the monitoring points under different amounts of gunpowder. At present, the pressure test of the bursting valve pipeline loop usually adopts the experimental method. However, the experimental method has the following problems: 1. Before the experiment, a test platform needs to be built, and the test platform needs to prepare the main control module, adjustable resistance box, signal conditioning module and multiple sets of test loop terminals and other components, which makes the test platform complex. 2. It takes a lot of time to adjust the parameters of each instrument during the experiment, resulting in a long preparation time, and the parameter adjustment is inaccurate, which easily leads to large measurement errors. 3. The bursting valve pipeline loop is long, and more test elements need to be deployed, usually between 200 and 250, resulting in high costs. 4. The traditional experimental measurement method can only obtain the pressure data of the test point, and cannot obtain the pressure data of any point in the pipeline. 5. Placing a test element at any point in the pipeline loop will be affected by the actual environment and cannot be operated, which will lead to the staff being unable to know the pressure data and unable to guide the design of the bursting valve pipeline loop. Summary of the invention

[0004] The present invention provides a bursting valve pipeline loop pressure measurement method that integrates simulation and measured data, in order to solve the technical problems mentioned in the above background technology.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] The present invention provides a bursting valve pipeline loop pressure detection method by integrating simulation and measured data, the method comprising the following steps:

[0007] Step S1: simulating the burst valve pipeline loop and processing the measured data to obtain predicted experimental data;

[0008] Step S2: dividing the predicted experimental data according to the proportion, and constructing a multi-fidelity proxy model of the pipeline loop according to the divided data;

[0009] Step S3: Calculate the pressure curve of the pipeline loop according to the multi-fidelity proxy model;

[0010] Step S4: Input the node position into the pressure curve to obtain the pressure data in the pipeline loop.

[0011] Furthermore, in another preferred embodiment, the above step S1 specifically includes the following steps:

[0012] Step S11: In the 20-meter-long burst valve pipeline loop, a node is selected every 25 cm, and a total of 80 nodes are used as nodes for extracting data;

[0013] Step S12: From the 80 nodes, one node is selected every 8 nodes, and 10 nodes are selected as sample points, and experimental tests are performed to obtain the pressure data of each node as the experimental group data;

[0014] Step S13: taking 80 nodes as simulation points, and obtaining pressure data of each node through simulation as simulation data;

[0015] Step S14: 80 groups of predicted experimental data are obtained through the data relationship between the same nodes in the experimental group data and the simulation group data.

[0016] Furthermore, in a preferred embodiment, the ratio in the above step S2 is between 1:5 and 1:10.

[0017] Furthermore, there is a preferred embodiment in which the ratio is designed to be 1:7.

[0018] Furthermore, in another preferred embodiment, when the ratio is designed to be 1:7, step S2 specifically includes the following steps:

[0019] Step S21: The node positions of 10 sets of experimental data are taken as input, which is high-fidelity data and is expressed as 10 sets of experimental data are output, represented by y j e =(j=1,...,m);

[0020] Step S22: The node positions of 70 sets of prediction data are taken as input, which is low-fidelity data and is expressed as 70 sets of prediction data are output, represented by y i c =(j=1,...,n);

[0021] Step S23: construct a kernel function, and use the kernel function to map the high-fidelity data input and the low-fidelity data input to a high-dimensional space;

[0022] Step S24: Optimizing the hyperparameters in the kernel function to obtain an effective mapping of the high-fidelity data input and the low-fidelity data input to the high-dimensional space;

[0023] Step S25: construct a loss function and solve the loss function to obtain a regression model;

[0024] Step S26: using the root mean square error of the high-fidelity data as the training error of the model to obtain the pressure curve of the pipeline loop.

[0025] Furthermore, there is a preferred embodiment, the above kernel function is expressed as:

[0026]

[0027] Furthermore, there is a preferred embodiment, the above regression model is expressed as:

[0028]

[0029] Among them, y represents the target data, α i Represents the predicted data node position, α i * represents predicted data, x i represents the node position of the experimental data, x represents the experimental data, and b represents the bias.

[0030] The bursting valve pipeline pressure detection method of the present invention by fusion of simulation and measured data can be fully implemented by computer software. Therefore, correspondingly, the present invention also provides a bursting valve pipeline pressure detection system by fusion of simulation and measured data. The system includes a storage device, and the storage device performs the following steps:

[0031] Simulate the burst valve pipeline loop and process the measured data to obtain predicted experimental data;

[0032] The predicted experimental data are divided according to proportion, and a multi-fidelity proxy model of the pipeline loop is constructed based on the divided data;

[0033] The pressure curve of the pipeline loop is calculated based on the multi-fidelity proxy model;

[0034] By inputting the node locations into the pressure curve, pressure data within the piping loop can be obtained.

[0035] The present invention also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the bursting valve pipeline pressure detection method integrating simulation and measured data as described above is executed.

[0036] The present invention also provides a computer device, which includes a memory and a processor, wherein the memory stores a computer program, and when the processor runs the computer program stored in the memory, the processor executes any one of the above-mentioned methods for detecting bursting valve pipeline pressure by fusing simulation and measured data.

[0037] The beneficial effects of the present invention are:

[0038] 1. The present invention provides a bursting valve pipeline loop pressure detection method that integrates simulation and measured data. The simulation data and measured data are acquired by dividing the nodes of the pipeline loop. The measured data is input as high-fidelity data, and the simulation data is input as low-fidelity data. A multi-fidelity proxy model of the pipeline loop is built by proportional division. The multi-fidelity proxy model is processed to obtain a pressure curve of the pipeline loop, and the pressure data of any point in the pipeline loop can be obtained through the pressure curve.

[0039] Furthermore, the present invention uses the distance from the selected sample point to the burst valve as input and the pressure data measured by the pressure sensor as output as high-fidelity data for building a multi-fidelity proxy model, while using simulation data as low-fidelity data to achieve accurate prediction of the input at any node position in the pipeline loop.

[0040] Furthermore, the present invention designs the ratio of the measured group data to the simulation group data to be between 1:5 and 1:10, thereby ensuring that the accuracy of the pipeline loop multi-fidelity proxy model and the control cost are within a certain range.

[0041] Furthermore, compared with the prior art, the present invention only needs to select 10-15 experimental nodes to complete the pressure prediction of the entire pipeline loop. Compared with the 200-250 test points of the existing experimental method, the accuracy is higher with fewer experimental nodes.

[0042] The invention is suitable for pressure detection of a bursting valve pipeline loop. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0044] Figure 1 The present invention discloses a flow chart of a bursting valve pipeline loop pressure detection method integrating simulation and measured data. DETAILED DESCRIPTION

[0045] The specific embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings. The following embodiments will help those skilled in the art to further understand the present invention, but do not limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several changes and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

[0046] Embodiment 1. This embodiment aims at the existing bursting valve pipeline loop pressure measurement realized by experimental method, which has the problems of only obtaining pressure data of the test point, and being unable to obtain pressure data of any point in the pipeline, and large measurement error and high cost. Thus, a bursting valve pipeline loop pressure detection method integrating simulation and measured data is proposed, and the method comprises the following steps:

[0047] Step S1: simulating the burst valve pipeline loop and processing the measured data to obtain predicted experimental data;

[0048] Step S2: dividing the predicted experimental data according to the proportion, and constructing a multi-fidelity proxy model of the pipeline loop according to the divided data;

[0049] Step S3: Calculate the pressure curve of the pipeline loop according to the multi-fidelity proxy model;

[0050] Step S4: Input the node position into the pressure curve to obtain the pressure data in the pipeline loop.

[0051] The present embodiment provides a bursting valve pipeline loop pressure detection method that integrates simulation and measured data, and acquires simulation data and measured data by dividing the nodes of the pipeline loop; the measured data is input as high-fidelity data, and the simulation data is input as low-fidelity data, and a multi-fidelity proxy model of the pipeline loop is built by proportional division; the multi-fidelity proxy model is processed to obtain a pressure curve of the pipeline loop, and the pressure data of any point in the pipeline loop can be obtained through the pressure curve.

[0052] Implementation Method 2: See Figure 1 This embodiment is to explain the step S1 of the bursting valve pipeline loop pressure detection method of the above embodiment 1 by fusing simulation and measured data:

[0053] Step S1: simulating the burst valve pipeline loop and processing the measured data to obtain predicted experimental data;

[0054] The specific steps include:

[0055] According to the length of the bursting valve pipeline loop and the uncertainty in the actual pipeline loop, the collection density of the bursting valve pipeline loop is formulated. This can ensure the high-density distribution of the collected data in the pipeline loop space, which is conducive to capturing the subtle differences in pressure changes in the pipeline.

[0056] Step S11: In a 20-meter-long burst valve pipeline loop, a node is selected every 25 cm, and a total of 80 nodes are used as nodes for extracting data.

[0057] In order to reduce costs and workload, a strategy for selecting measured samples was developed, and 10 nodes were selected as sample points for experimental testing. This not only ensured the uniform distribution of measured sample points in the entire pipeline, but also reduced experimental costs and workload.

[0058] Step S12: From the 80 nodes, select one node every 8 nodes, select 10 nodes as sample points, and conduct experimental tests to obtain the pressure data of each node as the experimental group data.

[0059] By adding simulation data, the parts that cannot be covered by experimental data are supplemented, ensuring the integrity of the entire pipeline loop data and effectively balancing the experimental cost and efficiency.

[0060] Step S13: 80 nodes are used as simulation points, and pressure data of each node is obtained through simulation as simulation data.

[0061] Step S14: Obtain 80 groups of predicted experimental data through the data relationship of the same nodes in the experimental group data and the simulation group data;

[0062] Step S15: The 80 groups of predicted experimental data are divided according to proportion, and a multi-fidelity proxy model of the pipeline loop is constructed according to the proportionally divided data.

[0063] This implementation method combines a small amount of measured sample data with simulation data to achieve an effective balance between experimental cost and efficiency while ensuring data quality.

[0064] This implementation uses the distance from the selected sample point to the burst valve as input and the pressure data measured by the pressure sensor as output, which serves as high-fidelity data for building a multi-fidelity proxy model. Simulation data is used as low-fidelity data to achieve accurate prediction of the input at any node position in the pipeline loop.

[0065] Furthermore, when the above ratio division is performed, the ratio of the measured group data to the simulation group data is designed to be between 1:5 and 1:10, which ensures the accuracy of the multi-fidelity proxy model of the pipeline loop and the control cost is within a certain range. If the ratio is too small, overfitting is likely to occur, resulting in poor fitting effect, and if the ratio is too large, the amount of calculation will increase, which is not conducive to cost saving.

[0066] Implementation method 3: This implementation method specifically describes step S2 in the bursting valve pipeline pressure detection method of fusion of simulation and measured data described in the above implementation method;

[0067] In this implementation, the ratio of the test group data to the simulation group data is set to 1:7.

[0068] Step S21: The node positions of 10 sets of experimental data are taken as input, which is high-fidelity data and is expressed as 10 sets of experimental data are output, represented by y j e =(j=1,...,m);

[0069] Step S22: The node positions of 70 sets of prediction data are taken as input, which is low-fidelity data and is expressed as 70 sets of prediction data are output, represented by y i c =(j=1,...,n);

[0070] Step S23: Construct kernel function:

[0071]

[0072] The kernel function is used to map high-fidelity data input and low-fidelity data input to high-dimensional space, and the linear equation of high-dimensional space is used to characterize the nonlinear regression relationship between input and output in low-dimensional space, thereby improving the fitting ability of the model and improving the accuracy of prediction.

[0073] Among them, K(x c ,x c ) is calculated as follows:

[0074]

[0075] K(x c ,x e ) is calculated as follows:

[0076]

[0077] K(x e ,x c ) is calculated as follows:

[0078]

[0079] K(x e ,x e ) is calculated as follows:

[0080]

[0081] In the above formula, ρ, σ c , σ s ,θ c and θ s They all represent hyperparameters.

[0082] Step S24: Optimize the hyperparameters ρ, σ in the kernel function by Bayesian optimization method c , σ s ,θ c and θ s , we can obtain an effective mapping of low-fidelity data input and high-fidelity data input to high-dimensional space.

[0083] Step S25: Construct loss function:

[0084]

[0085] Among them, ω and b represent the weight coefficient and bias respectively; γ represents the adjustable parameter whose value range is generally [0,5], but the specific value needs to be adjusted according to the actual situation.

[0086] Introducing the slack variable ζ ， ξ * , transform the loss function into the following optimization formula:

[0087]

[0088] Using the Lagrangian function to solve the above equation, we get:

[0089]

[0090] For the above variables ω, b, α, α * ,ζ,ξ * ,μ,μ * Take the partial derivative and substitute the partial derivative result into the above formula to get

[0091]

[0092] α i ,α * i ≥0,i=1,…,n+m

[0093] Using linear programming to solve the above formula, we can get αi ,α * i , the final regression model is as follows:

[0094]

[0095] Among them, y represents the target data, α i Represents the predicted data node position, α i * represents predicted data, x i represents the node position of the experimental data, x represents the experimental data, and b represents the bias.

[0096] Step S26: Use the root mean square error of the fidelity data as the training error of the model:

[0097]

[0098] The pressure curve of the entire pipeline loop can be calculated through the above regression model. By inputting the position point in the pipeline loop into the pressure curve, the pressure data of any point in the pipeline loop can be obtained.

[0099] Embodiment 4: The bursting valve pipeline pressure detection method by fusion of simulation and measured data described in the above embodiment can be fully implemented by computer software. Therefore, correspondingly, this embodiment provides a bursting valve pipeline pressure calculation system by fusion of simulation and measured data. The system includes a storage device, and the storage device performs the following steps:

[0100] Simulate the burst valve pipeline loop and process the measured data to obtain predicted experimental data;

[0101] The predicted experimental data are divided according to proportion, and a multi-fidelity proxy model of the pipeline loop is constructed based on the divided data;

[0102] The pressure curve of the pipeline loop is calculated based on the multi-fidelity proxy model;

[0103] By inputting the node locations into the pressure curve, pressure data within the piping loop can be obtained.

[0104] Embodiment 5: This embodiment provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, a bursting valve pipeline pressure detection method integrating simulation and measured data as described in any one of the above embodiments is executed.

[0105] Embodiment 6. This embodiment provides a computer device, which includes a memory and a processor, wherein the memory stores a computer program. When the processor runs the computer program stored in the memory, the processor executes a bursting valve pipeline pressure detection method that integrates simulation and measured data as described in any one of the above embodiments.

[0106] A computer device provided in this embodiment, the hardware device of this part is a general model and is not shown in the form of a diagram. The system includes a processor and a memory, wherein the processor and the memory can be connected via a bus or other means. The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs, non-transient computer executable programs and modules, and corresponding program instructions / modules. The processor executes various functional applications and data processing of the processor by running the non-transient software programs, instructions and modules stored in the memory, so as to realize the bursting valve pipeline pressure detection method and steps of the fusion of simulation and measured data in the above method embodiment.

[0107] Embodiment 7: This embodiment compares and explains the bursting valve pipeline pressure detection method that integrates simulation and measured data described in the above embodiment with the existing experimental method;

[0108] The current experimental method for measuring the pressure of the bursting valve pipeline loop requires multiple test nodes, that is, the pressure at a certain point requires the installation of a pressure sensor at that point. If the pressure value of the entire pipeline is to be obtained, the pressure value of the pipeline must be obtained every approximately 10 cm. The required test points fluctuate between 200-250.

[0109] The bursting valve pipeline pressure detection method that integrates simulation and measured data proposed in the above-mentioned implementation method only needs to select 10-15 experimental nodes to complete the pressure prediction of the entire pipeline loop. Compared with the existing experimental methods, it not only saves costs but also effectively improves accuracy.

[0110] The above description is only the implementation mode of the present invention and is not limited to the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present invention shall be included in the scope of the claims of the present invention.

Claims

1. A bursting valve pipeline loop pressure detection method integrating simulation and measured data, characterized in that: The method is: S1: Simulate the burst valve pipeline loop and process the measured data to obtain predicted experimental data; S2: Divide the predicted experimental data according to the proportion, and build a multi-fidelity proxy model of the pipeline loop based on the divided data; S3: Calculate the pressure curve of the pipeline loop based on the multi-fidelity proxy model; S4: By inputting the node position into the pressure curve, the pressure data in the pipeline loop can be obtained.

2. The method for detecting the pressure of a bursting valve pipeline loop by fusing simulation and measured data according to claim 1 is characterized in that: S1 is specifically: S11: In the 20-meter burst valve pipeline loop, equidistant sampling was adopted, and a node was selected every 25 cm, with a total of 80 nodes as nodes for extracting data; S12: From the 80 nodes, a sampling strategy is adopted to select one node every 8 nodes, and 10 nodes are selected as sample points, and experimental tests are carried out to obtain the pressure data of each node as the experimental group data; S13: 80 nodes are used as simulation points, and pressure data of each node is obtained through simulation as simulation data; S14: 80 groups of predicted experimental data are obtained through the data relationship of the same nodes in the experimental group data and the simulation group data.

3. The method for detecting the pressure of a bursting valve pipeline loop by fusion of simulation and measured data according to claim 2 is characterized in that: The ratio in S2 is between 1:5 and 1:

10.

4. A bursting valve pipeline loop pressure detection method based on the fusion of simulation and measured data according to claim 3, characterized in that: The ratio is 1:

7.

5. A bursting valve pipeline loop pressure detection method based on the fusion of simulation and measured data according to claim 4, characterized in that: S2 is specifically: S21: The node positions of 10 sets of experimental data are used as input, which is high-fidelity data and is expressed as 10 sets of experimental data are output, represented by y j e =(j=1,...,m); S22: The node positions of 70 sets of prediction data are used as input, which is low-fidelity data and is expressed as 70 sets of predicted data are output, represented by y i c =(j=1,...,n); S23: construct a kernel function, and use the kernel function to map the high-fidelity data input and the low-fidelity data input to a high-dimensional space; S24: Optimize the hyperparameters in the kernel function to obtain an effective mapping of high-fidelity data input and low-fidelity data input to the high-dimensional space; S25: construct a loss function and solve the loss function to obtain a regression model; S26: The root mean square error of the high-fidelity data is used as the training error of the model to obtain the pressure curve of the pipeline loop.

6. A bursting valve pipeline loop pressure detection method based on the fusion of simulation and measured data according to claim 5, characterized in that: The kernel function is expressed as:

7. The method for detecting the pressure of a bursting valve pipeline loop by fusing simulation and measured data according to claim 5 is characterized in that: The regression model is: Among them, y represents the target data, α i Represents the predicted data node position, α i * represents predicted data, x i represents the node position of the experimental data, x represents the experimental data, and b represents the bias.

8. A bursting valve pipeline pressure detection system integrating simulation and measured data, characterized in that: The system comprises a storage device, and the storage device executes the steps described in claim 1.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the bursting valve pipeline pressure detection method of integrating simulation and measured data as described in any one of claims 1 to 7 is executed.

10. A computer device, characterized in that: The device includes a memory and a processor, wherein a computer program is stored in the memory. When the processor runs the computer program stored in the memory, the processor executes a bursting valve pipeline pressure detection method integrating simulation and measured data as described in any one of claims 1 to 7.

Images