A method for correcting the flow coefficient of large-diameter regulating valve

Abstract

The present application provides a method for correcting the flow coefficient of a large-diameter regulating valve, belonging to the field of aero-engine technology. The method includes filtering the original test data with a tracking differentiator to obtain filtered test data; constructing a flow characteristic model of the regulating valve, retrieving the theoretical flow coefficient of the original model by setting the same simulation conditions as the test conditions; constructing a flow coefficient calculation model, inputting the filtered test data into the model, and calculating the true flow coefficient; establishing a flow coefficient deviation prediction model based on a random forest algorithm, substituting the difference between the true flow coefficient and the theoretical flow coefficient into the model for training, obtaining the flow coefficient prediction deviation, superimposing the deviation with the theoretical flow coefficient to generate a new flow coefficient, and replacing the theoretical flow coefficient of the original model with the new flow coefficient; obtaining new test condition data, repeating the above steps, and continuously updating the flow coefficient. The present application scheme improves the accuracy of the flow coefficient and flow calculation of the transient state of the regulating valve.

Description

Technical Field

[0001] The present application relates to the technical field of aero-engines, and in particular to a method for correcting the flow coefficient of a large-diameter regulating valve. Background Art

[0002] The flight environment simulation system is the core system for establishing the in-flight operating environment of an engine. Its control quality directly impacts the authenticity of the test environment and the accuracy of the test performance simulation for the engine under test. Control algorithm development and engineering application verification for flight environment simulation systems are typically conducted through digital and semi-physical simulations. The accuracy of component models, particularly the DN3000 large-diameter control valve model (used to simulate aircraft engine flight conditions), is crucial to the confidence level of simulation results and the effectiveness of the simulation. The main existing methods for calculating valve flow coefficients, such as a flow calculation method for an air intake disc control valve for an altitude platform (CN201911236532.4), rely on steady-state operating data and require calculation of the pressure ratio before and after the disc control valve, and then checking the disc valve flow coefficient table to obtain the disc valve flow coefficient. This method cannot adapt to the dynamic flow changes under the transient operating conditions of the engine; a valve flow prediction method based on valve pressure difference (CN202110723676.3) proposes predicting the flow coefficient based on the valve pressure difference, and then matching the flow curve provided by the manual or numerical simulation to obtain the valve flow, which cannot improve the accuracy of the flow coefficient; a method, device, electronic equipment and medium for calculating the flow coefficient of a butterfly valve (CN202410205850.9) proposes using interpolation of the opening to the parameter table, which refines the number of flow coefficients and cannot solve the problem of accuracy of the dynamic flow coefficient. The flow coefficient determination method, device and electronic equipment (CN202110895654.5) simulates a preset valve process simulation model multiple times until the error composites the preset result to determine the flow coefficient. This method does not rely on a large amount of test data. It does not consider valve wear or performance degradation factors, cannot suppress the influence of external interference under transient conditions, and has limited room for improvement in the accuracy of the flow coefficient.

[0003] The test conditions faced by flight environment simulation systems are complex, with temperature and pressure varying from the ground to the air. Traditional flow coefficient lookup tables or universal flow coefficient curves often yield significant discrepancies between these and the actual flow coefficients encountered during test conditions. Furthermore, the dynamic flow of large-diameter control valves (DN3000 butterfly valves) is difficult to accurately calculate when engine flow undergoes rapid transient changes. Due to external interference, the test data contains significant noise, resulting in deviations exceeding 10%. These factors (such as external interference during rapid dynamic changes, butterfly valve wear, or performance degradation) all hinder the accurate determination of the flow coefficient for large-diameter control valves, making the modification of butterfly valve models more difficult. Summary of the Invention

[0004] In view of this, an embodiment of the present application provides a method for correcting the flow coefficient of a large-diameter control valve to solve technical problems that cannot be solved by traditional flow coefficient acquisition methods, such as the problem of accurately obtaining the dynamic flow coefficient under transition conditions, the impact of external interference on the accuracy of the flow coefficient, and the failure to consider mechanical wear or sealing performance degradation of the control valve.

[0005] The present application provides a method for correcting the flow coefficient of a large-diameter regulating valve, the method comprising:

[0006] Step 1: Obtain original test data under test conditions, use a tracking differentiator to perform noise suppression and differential extraction on the original test data, and obtain filtered test data;

[0007] Step 2: construct a flow characteristic model of the large-diameter regulating valve, and retrieve the theoretical flow coefficient of the original flow characteristic model by setting the same simulation conditions as the test conditions;

[0008] Step 3: construct a flow coefficient calculation model for the large-diameter control valve, input the filtered test data into the flow coefficient calculation model, and calculate the actual flow coefficient of the large-diameter control valve under the test conditions;

[0009] Step 4: Establish a flow coefficient deviation prediction model based on the random forest algorithm, substitute the difference between the actual flow coefficient and the theoretical flow coefficient into the flow coefficient deviation prediction model for training, and obtain the flow coefficient prediction deviation;

[0010] Step 5: Calculate the determination coefficient of the flow coefficient prediction deviation. When the determination coefficient meets the preset conditions, stop training the flow coefficient deviation prediction model.

[0011] Step 6: Superimpose the predicted deviation of the flow coefficient and the theoretical flow coefficient to generate a new flow coefficient, and replace the theoretical flow coefficient of the original flow characteristic model with the new flow coefficient;

[0012] Step 7: Obtain new test condition data, repeat steps 1 to 6, and continuously update the flow coefficient of the large-diameter control valve.

[0013] According to a specific implementation of the embodiment of the present application, the tracking differentiator is a nonlinear tracking differentiator.

[0014] According to a specific implementation of the embodiment of the present application, the expression of the nonlinear tracking differentiator is:

[0015] ,

[0016] in, is the sampling step, fst is the optimal integrated function of fast control, x1 is the tracking value of the input signal, x2 is the differential value of the tracking signal, v(k) is the input signal, R1 is the fast factor, h is the filter factor, k is the current iteration number, and k+1 is the next iteration number.

[0017] According to a specific implementation of the embodiment of the present application, the expression of the flow characteristic model is:

[0018] ,

[0019] Among them, W a is the mass flow rate of the regulating valve, P1 is the pressure before the valve, C is the flow coefficient, A0 is the throttling area of ​​the regulating valve, R is the thermodynamic constant, and T1 is the temperature before the valve.

[0020] According to a specific implementation of the embodiment of the present application, the expression of the flow coefficient calculation model is:

[0021] ,

[0022] Where C1 is the true flow coefficient.

[0023] According to a specific implementation of an embodiment of the present application, establishing a flow coefficient deviation prediction model based on a random forest algorithm includes:

[0024] Establish feature engineering X including multiple characteristic parameters of large-caliber regulating valves;

[0025] Based on the actual flow coefficient and the theoretical flow coefficient, the actual flow coefficient deviation is obtained;

[0026] Build a flow coefficient deviation prediction model based on feature engineering X and the true flow coefficient deviation to obtain a single decision tree;

[0027] The flow coefficient prediction deviation of the flow coefficient deviation prediction model is obtained based on the total single decision tree.

[0028] According to a specific implementation of the embodiment of the present application, the establishment of a flow coefficient deviation prediction model based on a random forest algorithm further includes:

[0029] The feature parameters of feature engineering X are ranked by importance. If the change in the importance of a feature parameter exceeds the preset range, the wear condition of the large-diameter control valve and the status of the measurement sensor are checked, and the weight of the feature parameter of feature engineering X is adjusted.

[0030] According to a specific implementation of the embodiment of the present application, the calculation formula of the true flow coefficient deviation is:

[0031] ,

[0032] The expression of the single decision tree is:

[0033] ,

[0034] The expression of the flow coefficient prediction deviation is:

[0035] ,

[0036] Among them, e is the actual flow coefficient deviation, C0 is the theoretical flow coefficient, e i is the deviation of the true flow coefficient of the ith t (X) is the t-th decision tree function, l is the number of leaf nodes, S l is the set of training samples falling on leaf nodes, T is the total number of decision trees, e pred It is the flow coefficient prediction deviation.

[0037] According to a specific implementation of the embodiment of the present application, the expression of the feature engineering X is:

[0038] ,

[0039] Among them, K is the valve opening, P2 is the pressure after the valve, is the extracted differential characteristic of the pre-valve pressure, Pr is the pressure ratio, is the logarithmic transformation of temperature, is the thermodynamic factor, is the interaction characteristic between valve opening and pressure, is the interaction characteristic between valve opening and temperature.

[0040] According to a specific implementation of the embodiment of the present application, the expression of the determination coefficient is:

[0041] ,

[0042] Among them, R 2 is the coefficient of determination, is the predicted deviation of the i-th flow coefficient, is the average value of the discharge coefficient deviation in the test data, and n is the number of test data.

[0043] Beneficial effects:

[0044] The large-diameter regulating valve flow coefficient correction method in the embodiment of the present application is designed to design a large-diameter regulating valve flow coefficient correction method based on a tracking differentiator and a random forest, and the test data is filtered based on the tracking differentiator to suppress noise signals and interference, obtain a smooth filtered signal, and improve the accuracy of the flow coefficient calculation; at the same time, the deviation between the actual flow coefficient and the theoretical flow coefficient is trained and predicted based on the random forest algorithm to generate a new flow coefficient. The joint action of these two algorithms improves the model confidence and effectiveness, thereby effectively improving the transition state flow coefficient and flow calculation accuracy of the large-diameter butterfly valve. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0046] Figure 1 This is a diagram showing the effect of pressure data filtering based on a tracking differentiator according to an embodiment of the present invention;

[0047] Figure 2 Flow chart of a method for correcting a flow coefficient of a large-diameter regulating valve according to an embodiment of the present invention;

[0048] Figure 3 2. A diagram showing the evaluation result of the flow coefficient correction effect according to an embodiment of the present invention;

[0049] Figure 4 is a graph showing the importance ranking of features affecting the discharge coefficient according to an embodiment of the present invention;

[0050] Figure 5 4 is a comparison diagram of the regulating valve flow rate and the actual flow rate before and after correction under a transient working condition according to an embodiment of the present invention;

[0051] Figure 6 3 is a comparison diagram of the flow deviation of the regulating valve before and after correction under a transient working condition according to an embodiment of the present invention. DETAILED DESCRIPTION

[0052] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0053] The following describes the embodiments of the present application through specific examples, and those skilled in the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The present application can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that, in the absence of conflict, the features in the following embodiments and embodiments can be combined with each other. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without making creative work are within the scope of protection of this application.

[0054] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein can be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on this application, it should be understood by those skilled in the art that an aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects described herein can be used to implement an apparatus and / or practice a method. In addition, other structures and / or functionalities other than one or more of the aspects described herein can be used to implement this apparatus and / or practice this method.

[0055] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present application. The illustrations only show components related to the present application and are not drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component can be changed at will, and the component layout type may also be more complicated.

[0056] Additionally, in the following description, specific details are provided to provide a thorough understanding of the examples. However, one skilled in the art will appreciate that the aspects described can be practiced without these specific details.

[0057] An embodiment of the present application provides a method for correcting the flow coefficient of a large-diameter regulating valve, which is described in detail below with reference to the accompanying drawings.

[0058] Reference Figure 2 This embodiment provides a method for correcting the flow coefficient of a large-diameter regulating valve, the method comprising the following steps:

[0059] Step 1: Obtain original test data under test conditions, use a tracking differentiator to perform noise suppression and differential extraction on the original test data, and obtain filtered test data;

[0060] Step 2: construct a flow characteristic model of the large-diameter regulating valve, and retrieve the theoretical flow coefficient of the original flow characteristic model by setting the same simulation conditions as the test conditions;

[0061] Step 3: construct a flow coefficient calculation model for the large-diameter control valve, input the filtered test data into the flow coefficient calculation model, and calculate the actual flow coefficient of the large-diameter control valve under the test conditions;

[0062] Step 4: Establish a flow coefficient deviation prediction model based on the random forest algorithm, substitute the difference between the actual flow coefficient and the theoretical flow coefficient into the flow coefficient deviation prediction model for training, and obtain the flow coefficient prediction deviation;

[0063] Step 5: Calculate the determination coefficient of the flow coefficient prediction deviation. When the determination coefficient meets the preset conditions, stop training the flow coefficient deviation prediction model.

[0064] Step 6: Superimpose the predicted deviation of the flow coefficient and the theoretical flow coefficient to generate a new flow coefficient, and replace the theoretical flow coefficient of the original flow characteristic model with the new flow coefficient;

[0065] Step 7: Obtain new test condition data, repeat steps 1 to 6, and continuously update the flow coefficient of the large-diameter control valve.

[0066] In this embodiment, a flow coefficient correction method for large-diameter regulating valves based on a tracking differentiator and a random forest is designed, and the test data is filtered based on the tracking differentiator to suppress noise signals and interference, obtain a smooth filtered signal, and improve the accuracy of flow coefficient calculation; at the same time, the deviation between the actual flow coefficient and the theoretical flow coefficient is trained and predicted based on the random forest algorithm to generate a new flow coefficient. The joint action of these two algorithms improves the confidence and effectiveness of the model, thereby effectively improving the transition state flow coefficient and flow calculation accuracy of the large-diameter butterfly valve.

[0067] In practice, raw equipment data often contains noise (such as sensor noise and external interference), resulting in low data quality. Using on-site sensors and a PLC control device to collect raw test data under different test conditions, a nonlinear tracking differentiator was used to suppress noise and perform differential extraction on raw test data such as the inlet pressure P1, mass flow Wa, and valve opening k of a large-diameter control valve (DN3000 butterfly valve), obtaining filtered test data.

[0068] Furthermore, the expression of the nonlinear tracking differentiator is:

[0069] ,

[0070] in, is the sampling step, fst is the optimal integrated function of fast control, x1 is the tracking value of the input signal, x2 is the differential value of the tracking signal, v(k) is the input signal, R1 is the fast factor, h is the filter factor, k is the current iteration number, and k+1 is the next iteration number.

[0071] The specific steps include:

[0072] Step 1.1, obtain the valve inlet pressure P1 and the mass flow rate W of the large-diameter regulating valve (DN3000 butterfly valve) a After preliminary screening and classification of the original test data such as valve opening k, the high-frequency signal is subjected to frequency matching processing;

[0073] Step 1.2: The preliminarily processed test data is filtered through a tracking differentiator to obtain cleaner data, effectively suppress high-frequency noise or remove outliers, retain the effective components in the signal, and improve data quality. The nonlinear tracking differentiator used in this method is:

[0074] (1).

[0075] In one embodiment, the flow characteristic model is expressed as:

[0076] ,

[0077] Among them, W a is the mass flow rate of the regulating valve, P1 is the pressure before the valve, C is the flow coefficient, A0 is the throttling area of ​​the regulating valve, R is the thermodynamic constant, and T1 is the temperature before the valve.

[0078] In specific implementation, the acquisition of theoretical flow coefficient includes:

[0079] Step 2.1: Establish a flow characteristic model for a large-diameter control valve. Assuming that the energy conversion process is adiabatic and ignoring the effects of fluid friction and eddy currents, the model is as follows:

[0080] (2),

[0081] Where C is the flow coefficient. The flow coefficient of the original model is obtained by simulating the valve structure parameters, and a flow coefficient table related to the valve opening and pressure ratio is generated.

[0082] Step 2.2: By setting the same simulation conditions as the test data conditions, the theoretical flow coefficient C0 of the original model is retrieved through the large-diameter control valve model (2).

[0083] Furthermore, the experimental data processed by tracking the differentiator filtering in step 1 is substituted into the flow coefficient calculation model of the large-diameter butterfly valve. The expression of the flow coefficient calculation model is:

[0084] (3),

[0085] Among them, the true flow coefficient C1 of the large-diameter control valve under given test conditions can be calculated.

[0086] In one embodiment, the establishment of a flow coefficient deviation prediction model based on a random forest algorithm includes:

[0087] Step 4.1, establish a feature engineering X that includes multiple characteristic parameters of large-caliber regulating valves;

[0088] Step 4.2: Based on the actual flow coefficient and the theoretical flow coefficient, obtain the actual flow coefficient deviation;

[0089] Step 4.3: Build a flow coefficient deviation prediction model based on feature engineering X and the true flow coefficient deviation to obtain a single decision tree;

[0090] Step 4.4: Obtain the flow coefficient prediction deviation of the flow coefficient deviation prediction model based on the total single decision tree.

[0091] Furthermore, the expression of the feature engineering X is:

[0092] (4),

[0093] Among them, K is the valve opening, P2 is the pressure after the valve, is the extracted differential characteristic of the pre-valve pressure, Pr is the pressure ratio, is the logarithmic transformation of temperature, is the thermodynamic factor, is the interaction characteristic between valve opening and pressure, is the interaction characteristic between valve opening and temperature.

[0094] Furthermore, the calculation formula of the true flow coefficient deviation is:

[0095] (5),

[0096] The expression of the single decision tree is:

[0097] (6),

[0098] The expression of the flow coefficient prediction deviation is:

[0099] (7),

[0100] Among them, e is the actual flow coefficient deviation, C0 is the theoretical flow coefficient, e i is the deviation of the true flow coefficient of the ith t (X) is the t-th decision tree function, l is the number of leaf nodes, S l is the set of training samples falling on leaf nodes, T is the total number of decision trees, e pred It is the flow coefficient prediction deviation.

[0101] In specific implementation, it is the number of leaf nodes, which is set to 5; S l The training sample set falling on the leaf node takes 1000 test data samples as an example, and sets the ratio of the training set to the test set of the discharge coefficient deviation prediction model to 7:3.

[0102] Furthermore, the expression of the determination coefficient is:

[0103] (8),

[0104] Among them, R 2 is the coefficient of determination, is the predicted deviation of the i-th flow coefficient, is the average value of the discharge coefficient deviation in the test data, and n is the number of test data.

[0105] In one embodiment, the establishing of a flow coefficient deviation prediction model based on a random forest algorithm further includes:

[0106] The feature parameters of feature engineering X are ranked by importance. If the change in the importance of a feature parameter exceeds the preset range, the wear condition of the large-diameter control valve and the status of the measurement sensor are checked, and the weight of the feature parameter of feature engineering X is adjusted.

[0107] In specific implementation, the prediction result e pred Perform the evaluation and calculate the coefficient of determination, including the following steps:

[0108] Step 5.1: Run the random forest algorithm to sort the characteristic parameters that affect the flow coefficient by importance. When the importance of a characteristic parameter suddenly increases, it is necessary to check the valve wear and whether the measurement sensor status is normal. Then, adjust the weight of the characteristic parameter in the feature engineering formula (4) and retrain the model to improve the training accuracy.

[0109] Step 5.2: Calculate the prediction result e pred The coefficient of determination (R 2 );

[0110] Step 5.3: Use formula (8) to calculate the coefficient of determination. When the coefficient of determination is greater than 0.9, stop model training.

[0111] Furthermore, the flow coefficient prediction deviation e predicted by the random forest algorithm is pred The new flow coefficient C2 is superimposed on the theoretical flow coefficient C0, and the new flow coefficient C2 is used to replace the flow coefficient C0 of the original model. Specifically, the flow coefficient prediction deviation e obtained in step 4 is used. pred Superimposed with the original theoretical flow coefficient C0, a new flow coefficient is generated:

[0112] (9),

[0113] The new flow coefficient C2 replaces the flow coefficient of the original control valve flow characteristic model;

[0114] Obtain new operating data, repeat the above steps, and continuously update the flow coefficient of the large-diameter control valve to ensure the accuracy of the control valve flow calculation.

[0115] The method of this embodiment filters the test data using a tracking differentiator to obtain smooth data for transitional test conditions and suppress random noise interference signals. Simultaneously, a feature engineering approach is constructed that takes into account factors such as valve structural wear and sealing performance degradation. A random forest algorithm is then established to train and predict the deviation between the actual flow coefficient and the theoretical flow coefficient, obtaining a predicted flow coefficient deviation. This prediction deviation is then superimposed on the theoretical flow coefficient to generate a new flow coefficient. This method can suppress random noise in the measurement signal and obtain more accurate test data. The flow coefficient and flow characteristic model, intelligently optimized by the random forest algorithm, achieves significantly improved accuracy under steady-state and transitional test conditions, demonstrating high accuracy and confidence, capable of meeting the requirements for dynamic simulation of altitude conditions in flight environment simulation systems.

[0116] In order to verify the correction effect of the large-diameter control valve flow coefficient correction method based on tracking differentiator and random forest proposed in the present invention on the control valve flow coefficient and flow characteristic model, including the filtering effect of the tracking differentiator and the prediction effect of random forest on the flow coefficient, the flow coefficient correction example under the following working conditions is used to illustrate.

[0117] The test data is filtered. Taking the valve front pressure data as an example, the tracking differentiator is used for filtering. The specific effect is as follows: Figure 1 As shown, Figure 1 The raw pressure data shown refers to the pressure before the large-diameter control valve without filtering, and the filtered pressure data refers to the pressure before the large-diameter control valve after filtering by the tracking differentiator. A feature engineering model and a random forest algorithm model were established, with the number of decision trees set to 200 and the minimum number of leaf nodes set to 5. The flow coefficient deviation was predicted. Figure 2The diagram below shows the principle diagram of flow coefficient deviation prediction of control valve based on random forest algorithm. The prediction results of flow coefficient are evaluated by random forest algorithm. Figure 3 As shown in the figure, the coefficient of determination calculated according to formula (8) is 0.999, indicating that the flow coefficient training results under the current working conditions can better meet the requirements of engineering applications. At the same time, the algorithm is used to sort the importance of the characteristic parameters that affect the accuracy of the flow coefficient, and the results are as follows: Figure 4 shown.

[0118] The comparison results of the control valve flow rate before and after correction by the random forest algorithm and the actual flow rate under the transition state are as follows: Figure 5 As shown, Figure 5 The “actual flow” here refers to the actual flow of the control valve obtained through test measurement, the “control valve flow before correction” refers to the control valve flow calculated using the uncorrected flow coefficient, and the “control valve flow after correction” refers to the control valve flow calculated using the corrected flow coefficient.

[0119] Comparison of the flow deviation of the regulating valve before and after correction by the random forest algorithm under the transient state Figure 6 As shown, the "control valve flow deviation before correction" refers to the deviation between the flow calculated using the uncorrected flow coefficient and the actual flow, and the "control valve flow deviation after correction" refers to the deviation between the flow calculated using the corrected flow coefficient and the actual flow.

[0120] Depend on Figure 5 、 Figure 6 It can be seen that the large-diameter control valve flow coefficient correction method based on tracking differentiator and random forest significantly improves the accuracy of the flow coefficient of the large-diameter control valve under transient conditions. The flow calculated by the corrected control valve model is much more accurate than that before correction, and the relative error is reduced from 11% to 2.7%. This method has a good correction effect on the flow model of large-diameter control valves.

[0121] The embodiment provided by the present invention designs a flow coefficient correction method for large-diameter regulating valves based on a tracking differentiator and a random forest, and filters the test data based on the tracking differentiator to suppress noise signals and interference, obtain a smooth filtered signal, and improve the accuracy of flow coefficient calculation; at the same time, the deviation between the actual flow coefficient and the theoretical flow coefficient is trained and predicted based on the random forest algorithm to generate a new flow coefficient. The joint action of these two algorithms improves the confidence and effectiveness of the model, thereby effectively improving the transition state flow coefficient and flow calculation accuracy of the large-diameter butterfly valve.

[0122] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A method for correcting the flow coefficient of a large-diameter regulating valve, characterized in that: The method comprises: Step 1: Obtain original test data under test conditions, use a tracking differentiator to perform noise suppression and differential extraction on the original test data, and obtain filtered test data; Step 2: Construct a flow characteristic model of a large-diameter regulating valve. By setting the same simulation conditions as the test conditions, retrieve the theoretical flow coefficient of the original flow characteristic model. The expression of the flow characteristic model is: ， Among them, W a is the mass flow rate of the regulating valve, P1 is the pressure before the valve, C is the flow coefficient, A0 is the throttling area of ​​the regulating valve, R is the thermodynamic constant, and T1 is the temperature before the valve; Step 3: Construct a flow coefficient calculation model for a large-diameter regulating valve, input the filtered test data into the flow coefficient calculation model, and calculate the true flow coefficient of the large-diameter regulating valve under the test conditions. The expression of the flow coefficient calculation model is: ， Wherein, C1 is the true flow coefficient; Step 4: Establish a flow coefficient deviation prediction model based on the random forest algorithm, substitute the difference between the actual flow coefficient and the theoretical flow coefficient into the flow coefficient deviation prediction model for training, and obtain the flow coefficient prediction deviation; The method of establishing a flow coefficient deviation prediction model based on a random forest algorithm includes: A feature engineering X including multiple characteristic parameters of a large-caliber regulating valve is established. The expression of the feature engineering X is: ， Among them, K is the valve opening, P2 is the pressure after the valve, is the extracted differential characteristic of the pre-valve pressure, Pr is the pressure ratio, is the logarithmic transformation of temperature, is the thermodynamic factor, is the interaction characteristic between valve opening and pressure, is the interaction characteristic between valve opening and temperature; Based on the actual flow coefficient and the theoretical flow coefficient, the actual flow coefficient deviation is obtained; Build a flow coefficient deviation prediction model based on feature engineering X and the true flow coefficient deviation to obtain a single decision tree; Obtaining the flow coefficient prediction deviation of the flow coefficient deviation prediction model based on the total single decision tree; The method of establishing a flow coefficient deviation prediction model based on a random forest algorithm further includes: Sort the importance of the feature parameters of feature engineering X. If the change in the importance of a feature parameter exceeds the preset range, check the wear of the large-diameter control valve and the status of the measurement sensor, and adjust the weight of the feature parameter of feature engineering X; Step 5: Calculate the determination coefficient of the flow coefficient prediction deviation. When the determination coefficient meets the preset conditions, stop training the flow coefficient deviation prediction model. Step 6: Superimpose the predicted deviation of the flow coefficient and the theoretical flow coefficient to generate a new flow coefficient, and replace the theoretical flow coefficient of the original flow characteristic model with the new flow coefficient; Step 7: Obtain new test condition data, repeat steps 1 to 6, and continuously update the flow coefficient of the large-diameter control valve.

2. The method for correcting the flow coefficient of a large-diameter regulating valve according to claim 1, characterized in that: The tracking differentiator is a nonlinear tracking differentiator.

3. The method for correcting the flow coefficient of a large-diameter regulating valve according to claim 2, characterized in that: The expression of the nonlinear tracking differentiator is: ， in, is the sampling step, fst is the optimal integrated function of fast control, x1 is the tracking value of the input signal, x2 is the differential value of the tracking signal, v(k) is the input signal, R1 is the fast factor, h is the filter factor, k is the current iteration number, and k+1 is the next iteration number.

4. The method for correcting the flow coefficient of a large-diameter regulating valve according to claim 1, characterized in that: The calculation formula of the true flow coefficient deviation is: ， The expression of the single decision tree is: ， The expression of the flow coefficient prediction deviation is: ， Among them, e is the actual flow coefficient deviation, C0 is the theoretical flow coefficient, e i is the deviation of the true flow coefficient of the ith t (X) is the t-th decision tree function, l is the number of leaf nodes, S l is the set of training samples falling on leaf nodes, T is the total number of decision trees, e pred It is the flow coefficient prediction deviation.

5. The method for correcting the flow coefficient of a large-diameter regulating valve according to claim 4, characterized in that: The expression of the coefficient of determination is: ， Among them, R 2 is the coefficient of determination, is the predicted deviation of the i-th flow coefficient, is the average value of the discharge coefficient deviation in the test data, and n is the number of test data.

Images