Method and device for performing soft correction on precision of volumetric flow meter through physical neural information network model, readable storage medium and equipment

Abstract

The invention provides a method for soft correction of the precision of a volumetric flow meter through a physical neural information network model, and the method is used for a turbine flow meter in the volumetric flow meter and comprises the steps that 1, key physical data are collected, the key physical data comprises output pulse frequency, fluid viscosity, fluid density, fluid temperature and fluid pressure data of the turbine flowmeter; 2, preprocessing the collected key physical data, wherein the preprocessing comprises cleaning, normalization processing and abnormal data point removal; 3, constructing a physical neural information network model based on a physical principle and a neural network structure; 4, performing deep training on the physical neural information network model by using the preprocessed training data; and 5, applying the trained physical neural information network model to real-time data measurement of the turbine flowmeter, and carrying out soft correction on the measurement precision of the turbine flowmeter. By integrating the machine learning algorithm, the precision of the turbine flowmeter is improved.

Description

Technical Field

[0001] The present application relates to the technical field of flow meter design, and in particular to a method, apparatus, readable storage medium and device for soft correction of volume flow meter accuracy using a physical neural information network model. Background Art

[0002] Turbine flowmeter is a commonly used flow measurement instrument with the advantages of high accuracy, good repeatability and simple structure. It is widely used in many industrial fields such as petroleum, chemical industry, and electric power. However, in actual use, the measurement accuracy of turbine flowmeter will be affected by many factors, such as fluid viscosity, density, temperature, pressure changes, mechanical wear of turbine flowmeter itself, flow field disturbance, etc. These factors will cause the instrument coefficient of turbine flowmeter to drift, thereby reducing the measurement accuracy.

[0003] Traditional turbine flowmeter accuracy calibration methods mainly include hardware calibration and software calibration based on mathematical models. Hardware calibration methods usually require improvements or adjustments to the structure or sensor of the turbine flowmeter, which is costly and less flexible. Although the software calibration method based on mathematical models is relatively flexible, traditional mathematical models often have difficulty accurately describing the nonlinear characteristics of turbine flowmeters under complex working conditions, and the calibration effect is limited. Therefore, a more effective, accurate and adaptable turbine flowmeter accuracy calibration method is needed. Summary of the invention

[0004] In view of this, the present application provides a method for soft-correcting the accuracy of a turbine flowmeter using a physical neural information network model to improve the measurement accuracy of the turbine flowmeter under complex working conditions.

[0005] To achieve the above objectives, the technical solutions adopted in this application are as follows: A method for soft correction of volume flow meter accuracy using a physical neural information network model, which is used for a turbine flow meter in a volume flow meter, comprises: Step 1: Collect key physical data, including output pulse frequency of turbine flowmeter, fluid viscosity, fluid density, fluid temperature and fluid pressure data; Step 2: preprocessing the collected key physical data, including cleaning, normalization and removal of abnormal data points; Step 3: construct a physical neural information network model based on physical principles and neural network structure, wherein the physical neural information network model includes a physical constraint layer and a neural network layer, wherein the physical constraint layer integrates physical laws into the model architecture in the form of mathematical expressions; Step 4: Use the preprocessed training data to perform deep training on the physical neural information network model, and use advanced back-propagation algorithms to fine-tune model parameters, including weights and biases of the neural network layers; Step 5: Apply the trained physical neural information network model to the real-time data measurement of the turbine flowmeter, and perform soft correction on the measurement accuracy of the turbine flowmeter.

[0006] Furthermore, the specific method of constructing the physical neural information network model based on physical principles and neural network structure in step 3 is: 3.1) Construct a neural network basic model, with the output pulse frequency of the turbine flowmeter, fluid viscosity, fluid density, fluid temperature and fluid pressure data as the input layer, and the corrected flow rate as the output layer. The weight and bias of the model are proportioned according to the following formula; in is the weight, b is the bias, is the input, z is the output, and m is the number of input neurons; The number of neurons in the hidden layer is obtained according to the following formula; in, is the number of neurons in the hidden layer, m is the number of neurons in the input layer, n is the number of neurons in the output layer, and a is a constant between [1,10]; 3.2) Define the physical problem and the corresponding physical laws: The flow formula of the turbine flowmeter: in, represents the rotation frequency of the turbine, It is the instrument factor of the turbine flowmeter, which is related to the structure, size and fluid properties of the turbine; Indicates the volume flow rate of the fluid; Transform the PDE equation as a physical constraint; 3.3) Based on the determined physical equation, define the total loss function, which consists of the physical error term and the data error term; The data error term is usually measured using the mean square error MSE: in, It is the ideal fluid flow rate; is the output of the neural network, i.e., the corrected fluid flow rate; N represents the number of training samples; The physical error term is calculated by automatic differentiation in combination with specific physical laws; according to the physical principle of the turbine flowmeter, its PDE equation can be expressed as ; Physical error term: in is the volume flow rate, is the turbine rotation frequency, is the instrument factor of the turbine flowmeter; Among them, L is the total loss function, and is a hyperparameter that weighs the importance of the two error terms.

[0007] Furthermore, in step 5, the trained physical neural information network model is applied to the real-time data measurement of the turbine flowmeter, and the measurement accuracy of the turbine flowmeter is soft-corrected as follows: 5.1) Input the collected data of the output pulse frequency, fluid viscosity, fluid density, fluid temperature and fluid pressure of the turbine flowmeter into the model, and the model outputs the corrected flow value; 5.2) Compare and verify the corrected flow value output by the model with the standard flow measurement equipment to evaluate the improvement effect of the measurement accuracy of the turbine flowmeter after correction. 5.3) If the accuracy does not meet the expected requirements, return to step 3 to further optimize the model parameters or increase the amount of training data until the accuracy requirements are met.

[0008] A device for softly correcting the accuracy of a volume flow meter using a physical neural information network model, which is used for a turbine flow meter in a volume flow meter, comprises: An acquisition module, used for acquiring key physical data, wherein the key physical data includes output pulse frequency of a turbine flowmeter, fluid viscosity, fluid density, fluid temperature and fluid pressure data; A preprocessing module, used for preprocessing the collected key physical data, including cleaning, normalization and removal of abnormal data points; A network construction module, which is used to construct a physical neural information network model based on physical principles and a neural network structure. The physical neural information network model includes a physical constraint layer and a neural network layer. The physical constraint layer integrates physical laws into the model architecture in the form of mathematical expressions. A training module for deeply training the physical neural information network model using preprocessed training data and fine-tuning the model parameters using advanced back-propagation algorithms, including the weights and biases of the neural network layers; The soft correction module is used to apply the trained physical neural information network model to the real-time data measurement of the turbine flowmeter and to perform soft correction on the measurement accuracy of the turbine flowmeter.

[0009] Furthermore, the network construction module is specifically used to perform the following steps: 3.1) Construct a neural network basic model, with the output pulse frequency of the turbine flowmeter, fluid viscosity, fluid density, fluid temperature and fluid pressure data as the input layer, and the corrected flow rate as the output layer. The weight and bias of the model are proportioned according to the following formula; in is the weight, b is the bias, is the input, z is the output, and m represents the number of input neurons; The number of neurons in the hidden layer is obtained according to the following formula; in, is the number of neurons in the hidden layer, m is the number of neurons in the input layer, n is the number of neurons in the output layer, and a is a constant between [1,10]; 3.2) Define the physical problem and the corresponding physical laws: The flow formula of the turbine flowmeter: in, represents the rotation frequency of the turbine, It is the instrument factor of the turbine flowmeter, which is related to the structure, size and fluid properties of the turbine; Indicates the volume flow rate of the fluid; Transform the PDE equation as a physical constraint; 3.3) Based on the determined physical equation, define the total loss function, which consists of the physical error term and the data error term; The data error term is usually measured using the mean square error MSE: in, It is the ideal fluid flow rate; is the output of the neural network, that is, the corrected fluid flow rate; N represents the number of training samples.

[0010] The physical error term is calculated by automatic differentiation in combination with specific physical laws; according to the physical principle of the turbine flowmeter, its PDE equation can be expressed as ; Physical error term: in is the volume flow rate, is the turbine rotation frequency, is the instrument factor of the turbine flowmeter; Where L is the total loss function; and is a hyperparameter that weighs the importance of the two error terms.

[0011] Furthermore, the soft correction module is specifically used to perform the following steps: 5.1) Input the collected data of the output pulse frequency, fluid viscosity, fluid density, fluid temperature and fluid pressure of the turbine flowmeter into the model, and the model outputs the corrected flow value; 5.2) Compare and verify the corrected flow value output by the model with the standard flow measurement equipment to evaluate the improvement effect of the measurement accuracy of the turbine flowmeter after correction. 5.3) If the accuracy does not meet the expected requirements, return to step 3 to further optimize the model parameters or increase the amount of training data until the accuracy requirements are met.

[0012] According to another aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored, and when the program is executed by a processor, the steps of the method of softly correcting the accuracy of a volumetric flow meter using a physical neural information network model of the present application are implemented.

[0013] According to another aspect of the present application, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, steps in a method for softly correcting the accuracy of a volumetric flow meter using a physical neural information network model of the present application are implemented.

[0014] Compared with the prior art, the beneficial effects of this application are: 1. This application improves the accuracy of turbine flowmeter by integrating machine learning algorithms; 2. The physical neural information network model of the present application includes a physical constraint layer, which imposes physical principle-based restrictions on the training and correction process of the model, ensuring the physical rationality and accuracy of the model output results; 3. During the model training process, this application uses the back-propagation algorithm to fine-tune the model parameters and uses cross-validation technology to effectively prevent model overfitting. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0016] Figure 1 It is a flow chart of a method for soft correction of volume flow meter accuracy using a physical neural information network model; Figure 2 This is a schematic diagram of the structure of the physical neural information network model of this application; Figure 3 The structure block diagram of a device for soft correction of volume flow meter accuracy using a physical neural information network model. DETAILED DESCRIPTION

[0017] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments.

[0018] like Figure 1 and 2 As shown, a method for soft correction of volume flow meter accuracy using a physical neural information network model is used for a turbine flow meter in a volume flow meter, comprising: Step 1: Collect key physical data, including output pulse frequency of turbine flowmeter, fluid viscosity, fluid density, fluid temperature and fluid pressure data; Under different fluid conditions, the pulse frequency signal of the turbine flowmeter is collected using a high-precision sensor , fluid viscosity ,temperature ,pressure At the same time, the corresponding standard flow value is obtained through the strictly calibrated high-precision flow measurement equipment ; These data should cover the various operating conditions in which the turbine flowmeter may operate to ensure the comprehensiveness and representativeness of the training data.

[0019] Step 2: preprocessing the collected key physical data, including cleaning, normalization and removal of abnormal data points; Preprocess the collected data. First, check the outliers in the data, such as data points that are obviously deviated from the normal measurement range or inconsistent with physical laws, and treat them as abnormal data and remove them. Then, use the normalization method to map the data to a specific interval for subsequent model training, such as normalizing all data to the [0, 1] interval. The specific normalization formula can be selected according to the data characteristics, such as for data , we can use the formula (in and are the minimum and maximum values ​​of the data in the collected samples respectively).

[0020] Step 3: construct a physical neural information network model based on physical principles and neural network structure, wherein the physical neural information network model includes a physical constraint layer and a neural network layer, wherein the physical constraint layer integrates physical laws into the model architecture in the form of mathematical expressions; Specifically, the method for constructing a physical neural information network model based on physical principles and neural network structure in step 3 is: 3.1) Construct a neural network basic model, with the output pulse frequency of the turbine flowmeter, fluid viscosity, fluid density, fluid temperature and fluid pressure data as the input layer, and the corrected flow rate as the output layer. The weight and bias of the model are proportioned according to the following formula; in is the weight, b is the bias, is the input, z is the output, and m is the number of input neurons; The number of neurons in the hidden layer is obtained according to the following formula; in, is the number of neurons in the hidden layer, m is the number of neurons in the input layer, n is the number of neurons in the output layer, and a is a constant between [1,10]; 3.2) Define the physical problem and the corresponding physical laws. This paper mainly studies the output compensation problem of turbine flowmeter, so the flow formula of turbine flowmeter is considered; in, Indicates the rotation frequency of the turbine; It is the instrument factor of the turbine flowmeter, which is related to the structure, size and fluid properties of the turbine; Represents the volume flow rate of the fluid; the above formula shows that the rotation frequency of the turbine is proportional to the volume flow rate of the fluid. By measuring the rotation frequency of the turbine, the flow rate of the fluid can be calculated; Transform the PDE equation as a physical constraint; 3.3) Based on the determined physical equation, define the total loss function, which consists of the physical error term and the data error term; The data error term is usually measured using the mean square error MSE: in, It is the ideal fluid flow rate; is the output of the neural network, i.e. the corrected fluid flow rate; N is the number of training samples. The purpose of the data fitting term is to make the output of the neural network as close to the ideal data as possible.

[0021] The physical error term is calculated by automatic differentiation in combination with specific physical laws; according to the physical principle of the turbine flowmeter, its PDE equation can be expressed as ; Physical error term: in is the volume flow rate, is the turbine rotation frequency, is the instrument factor of the turbine flowmeter; Where L is the total loss function; and is a hyperparameter that weighs the importance of the two error terms; and The choice of value should minimize the loss function value as much as possible. You can start by trying , and then adjust according to the model performance, such as adjusting to , gradually find the optimal combination, and select appropriate hyperparameters through debugging. While fitting the data, the model constrains the predicted core data within the range allowed by the laws of physics.

[0022] Step 4: Use the preprocessed training data to perform deep training on the physical neural information network model, and use advanced back-propagation algorithms to fine-tune model parameters, including weights and biases of the neural network layers; Use the preprocessed training data to train the constructed model. For example, set the learning rate to 0.01, the number of training times to 1000, use the back propagation algorithm, calculate the gradient according to the loss function, and update the weights and biases of the neural network layer.

[0023] During the training process, cross-validation is performed every 100 training runs. The training data is divided into a training set and a validation set, for example, in a ratio of 80:20. The loss function value and evaluation indicators (mean square error, mean absolute error) are calculated on the validation set. If the performance of the model on the validation set begins to decline (the loss function value increases or the evaluation indicators become worse), overfitting may occur. At this time, corresponding measures can be taken, such as increasing the regularization strength (adjusting the L2 regularization parameter), stopping training early, etc.

[0024] In addition, a target loss value threshold can be set, e.g. During the training process, the loss function is calculated after each round of training If , then stop training, at which point the model has achieved the expected optimization effect; if , the next round of training will continue, and the model parameters will be continuously optimized until the loss function meets the requirements. At the same time, cross-validation is used to effectively prevent overfitting during training, ensuring that the model has excellent generalization ability.

[0025] Step 5: Apply the trained physical neural information network model to the real-time data measurement of the turbine flowmeter, and perform soft correction on the measurement accuracy of the turbine flowmeter.

[0026] Specifically, further, in step 5, the trained physical neural information network model is applied to the real-time data measurement of the turbine flowmeter, and the measurement accuracy of the turbine flowmeter is soft-corrected as follows: 5.1) Input the collected data of the output pulse frequency, fluid viscosity, fluid density, fluid temperature and fluid pressure of the turbine flowmeter into the model, and the model outputs the corrected flow value; 5.2) Compare and verify the corrected flow value output by the model with the standard flow measurement equipment to evaluate the improvement effect of the measurement accuracy of the turbine flowmeter after correction. 5.3) If the accuracy does not meet the expected requirements, return to step 3 to further optimize the model parameters or increase the amount of training data until the accuracy requirements are met.

[0027] The trained model is applied to the real-time measurement data of the turbine flowmeter. , viscosity of the fluid ,temperature ,pressure Data is input into the model, and the model outputs the corrected flow value .

[0028] Specifically, the flow rate value measured by the high-precision calibration device is compared and verified. Under the same real-time working conditions, the flow rate value measured by the high-precision calibration device is . Calculate the relative error between the corrected flow value and the reference flow value For example, if the relative error is less than 0.01, it means that the accuracy of the calibrated turbine flowmeter meets the expectations; if not, return to the model training and optimization steps, further adjust the model parameters (such as reselecting hyperparameters, increasing the number of hidden layer neurons, etc.) or increase the amount of training data (collecting more data under different working conditions), and then train, verify and calibrate again until the accuracy requirements are met.

[0029] like Figure 3 As shown, a device for softly correcting the accuracy of a volume flow meter using a physical neural information network model is used for a turbine flow meter in a volume flow meter, comprising: The acquisition module 310 is used to acquire key physical data, wherein the key physical data includes the output pulse frequency of the turbine flowmeter, fluid viscosity, fluid density, fluid temperature and fluid pressure data; A preprocessing module 320 is used to preprocess the collected key physical data, including cleaning, normalization and removal of abnormal data points; A network construction module 330 is used to construct a physical neural information network model based on physical principles and a neural network structure. The physical neural information network model includes a physical constraint layer and a neural network layer. The physical constraint layer integrates physical laws into the model architecture in the form of mathematical expressions. A training module 340 is used to perform deep training on the physical neural information network model using the preprocessed training data, and fine-tune the model parameters using an advanced back-propagation algorithm, wherein the model parameters include weights and biases of the neural network layers; The soft correction module 350 is used to apply the trained physical neural information network model to the real-time data measurement of the turbine flowmeter and perform soft correction on the measurement accuracy of the turbine flowmeter.

[0030] Furthermore, the network construction module 330 is specifically configured to perform the following steps: 3.1) Construct a neural network basic model, with the output pulse frequency of the turbine flowmeter, fluid viscosity, fluid density, fluid temperature and fluid pressure data as the input layer, and the corrected flow rate as the output layer. The weight and bias of the model are proportioned according to the following formula; in is the weight, b is the bias, is the input, z is the output, and m is the number of input neurons; The number of neurons in the hidden layer is obtained according to the following formula; in, is the number of neurons in the hidden layer, m is the number of neurons in the input layer, n is the number of neurons in the output layer, and a is a constant between [1,10]; 3.2) Define the physical problem and the corresponding physical laws: The flow formula of the turbine flowmeter: in, represents the rotation frequency of the turbine, It is the instrument factor of the turbine flowmeter, which is related to the structure, size and fluid properties of the turbine; Indicates the volume flow rate of the fluid; Transform the PDE equation as a physical constraint; 3.3) Based on the determined physical equation, define the total loss function, which consists of the physical error term and the data error term; The data error term is usually measured using the mean square error MSE: in, It is the ideal fluid flow rate; is the output of the neural network, that is, the corrected fluid flow rate; N represents the number of training samples.

[0031] The physical error term is calculated by automatic differentiation in combination with specific physical laws; according to the physical principle of the turbine flowmeter, its PDE equation can be expressed as ; Physical error term: in is the volume flow rate, is the turbine rotation frequency, is the instrument factor of the turbine flowmeter; Where L is the total loss function; and is a hyperparameter that weighs the importance of the two error terms.

[0032] Furthermore, the soft correction module 350 is specifically configured to perform the following steps: 5.1) Input the collected data of the output pulse frequency, fluid viscosity, fluid density, fluid temperature and fluid pressure of the turbine flowmeter into the model, and the model outputs the corrected flow value; 5.2) Compare and verify the corrected flow value output by the model with the standard flow measurement equipment to evaluate the improvement effect of the measurement accuracy of the turbine flowmeter after correction. 5.3) If the accuracy does not meet the expected requirements, return to step 3 to further optimize the model parameters or increase the amount of training data until the accuracy requirements are met.

[0033] As for the device embodiment, since it basically corresponds to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The device embodiment described above is only illustrative, and the units described as separate components may or may not be physically separated.

[0034] The components may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units, such as distributed on a server and a client. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art may understand and implement it without creative work.

[0035] According to another aspect of the present application, a computer-readable storage medium is provided, on which a computer program is stored, and when the program is executed by a processor, the steps of the method of softly correcting the accuracy of a volumetric flow meter using a physical neural information network model of the present application are implemented.

[0036] According to another aspect of the present application, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, steps in a method for softly correcting the accuracy of a volumetric flow meter using a physical neural information network model of the present application are implemented.

[0037] The present application improves the accuracy of the turbine flowmeter by integrating a machine learning algorithm; and the physical neural information network model of the present application includes a physical constraint layer, which imposes physical principle-based restrictions on the training and correction process of the model, thereby ensuring the physical rationality and accuracy of the model output results.

[0038] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for soft correction of volume flow meter accuracy using a physical neural information network model, which is used for a turbine flow meter in a volume flow meter, and is characterized in that: include: Step 1: Collect key physical data, including output pulse frequency of turbine flowmeter, fluid viscosity, fluid density, fluid temperature and fluid pressure data; Step 2: preprocessing the collected key physical data, including cleaning, normalization and removal of abnormal data points; Step 3: construct a physical neural information network model based on physical principles and neural network structure, wherein the physical neural information network model includes a physical constraint layer and a neural network layer, wherein the physical constraint layer integrates physical laws into the model architecture in the form of mathematical expressions; Step 4: Use the preprocessed training data to perform deep training on the physical neural information network model, and use advanced back-propagation algorithms to fine-tune model parameters, including weights and biases of the neural network layers; Step 5: Apply the trained physical neural information network model to the real-time data measurement of the turbine flowmeter, and perform soft correction on the measurement accuracy of the turbine flowmeter.

2. The method for soft correction of volume flow meter accuracy using a physical neural information network model as claimed in claim 1, characterized in that: The specific method of constructing the physical neural information network model based on physical principles and neural network structure in step 3 is: 1) Construct a neural network basic model, with the output pulse frequency of the turbine flowmeter, fluid viscosity, fluid density, fluid temperature and fluid pressure data as the input layer, and the corrected flow rate as the output layer. The weight and bias of the model are proportioned according to the following formula; in is the weight, b is the bias, is the input, z is the output, and m is the number of input neurons; The number of neurons in the hidden layer is obtained according to the following formula; in, is the number of neurons in the hidden layer, m is the number of neurons in the input layer, n is the number of neurons in the output layer, and a is a constant between [1,10]; 2) Define the physical problem and the corresponding physical laws: The flow formula of the turbine flowmeter: in, represents the rotation frequency of the turbine, It is the instrument factor of the turbine flowmeter, which is related to the structure, size and fluid properties of the turbine; Indicates the volume flow rate of the fluid; Transform the PDE equation as a physical constraint; 3) Based on the determined physical equation, define the total loss function, which consists of the physical error term and the data error term; The data error term is usually measured using the mean square error MSE: in, It is the ideal fluid flow rate; is the output of the neural network, which is the corrected fluid flow rate; Indicates the number of training samples; The physical error term is calculated by automatic differentiation in combination with specific physical laws; according to the physical principle of the turbine flowmeter, its PDE equation can be expressed as ; Physical error term: in is the volume flow rate, is the turbine rotation frequency, is the instrument factor of the turbine flowmeter; Indicates the number of training samples; Where L is the total loss function; and is a hyperparameter that weighs the importance of the two error terms.

3. The method for soft correction of volume flow meter accuracy using a physical neural information network model as claimed in claim 2, characterized in that: In step 5, the trained physical neural information network model is applied to the real-time data measurement of the turbine flowmeter, and the measurement accuracy of the turbine flowmeter is soft-corrected as follows: 1) Input the collected data of the output pulse frequency, fluid viscosity, fluid density, fluid temperature and fluid pressure of the turbine flowmeter into the model, and the model outputs the corrected flow value; 2) Compare and verify the corrected flow value output by the model with the standard flow measurement equipment to evaluate the improvement effect of the measurement accuracy of the turbine flowmeter after correction. 3) If the accuracy does not meet the expected requirements, return to step 3 to further optimize the model parameters or increase the amount of training data until the accuracy requirements are met.

4. A device for soft correction of volume flow meter accuracy using a physical neural information network model, used for a turbine flow meter in a volume flow meter, characterized in that: include: An acquisition module, used for acquiring key physical data, wherein the key physical data includes output pulse frequency of a turbine flowmeter, fluid viscosity, fluid density, fluid temperature and fluid pressure data; A preprocessing module, used for preprocessing the collected key physical data, including cleaning, normalization and removal of abnormal data points; A network construction module, which is used to construct a physical neural information network model based on physical principles and a neural network structure. The physical neural information network model includes a physical constraint layer and a neural network layer. The physical constraint layer integrates physical laws into the model architecture in the form of mathematical expressions. A training module for deeply training the physical neural information network model using preprocessed training data and fine-tuning the model parameters using advanced back-propagation algorithms, including the weights and biases of the neural network layers; The soft correction module is used to apply the trained physical neural information network model to the real-time data measurement of the turbine flowmeter and to perform soft correction on the measurement accuracy of the turbine flowmeter.

5. The device for soft correction of volume flow meter accuracy using a physical neural information network model as claimed in claim 4, characterized in that: The network construction module is specifically used to perform the following steps: 1) Construct a neural network basic model, with the output pulse frequency of the turbine flowmeter, fluid viscosity, fluid density, fluid temperature and fluid pressure data as the input layer, and the corrected flow rate as the output layer. The weight and bias of the model are proportioned according to the following formula; in is the weight, b is the bias, is the input, z is the output, and m is the number of input neurons; The number of neurons in the hidden layer is obtained according to the following formula; in, is the number of neurons in the hidden layer, m is the number of neurons in the input layer, n is the number of neurons in the output layer, and a is a constant between [1,10]; 2) Define the physical problem and the corresponding physical laws: The flow formula of the turbine flowmeter: in, represents the rotation frequency of the turbine, It is the instrument factor of the turbine flowmeter, which is related to the structure, size and fluid properties of the turbine; Indicates the volume flow rate of the fluid; Transform the PDE equation as a physical constraint; 3) Based on the determined physical equation, define the total loss function, which consists of the physical error term and the data error term; The data error term is usually measured using the mean square error MSE: in, It is the ideal fluid flow rate; is the output of the neural network, which is the corrected fluid flow rate; Indicates the number of training samples; The physical error term is calculated by automatic differentiation in combination with specific physical laws; according to the physical principle of the turbine flowmeter, its PDE equation can be expressed as ; Physical error term: in is the volume flow rate, is the turbine rotation frequency, is the instrument factor of the turbine flowmeter; Where L is the total loss function; and is a hyperparameter that weighs the importance of the two error terms.

6. A device for soft correction of volume flow meter accuracy using a physical neural information network model as claimed in claim 5, characterized in that: The soft correction module is specifically used to perform the following steps: 1) Input the collected data of the output pulse frequency, fluid viscosity, fluid density, fluid temperature and fluid pressure of the turbine flowmeter into the model, and the model outputs the corrected flow value; 2) Compare and verify the corrected flow value output by the model with the standard flow measurement equipment to evaluate the improvement effect of the measurement accuracy of the turbine flowmeter after correction. 3) If the accuracy does not meet the expected requirements, return to step 3 to further optimize the model parameters or increase the amount of training data until the accuracy requirements are met.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps in the method for soft correction of volume flow meter accuracy using a physical neural information network model as described in any one of claims 1 to 3 are implemented.

8. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, it implements the steps in the method of soft-correcting the accuracy of a volume flow meter using a physical neural information network model as described in any one of claims 1 to 3.

Images