Impeller natural frequency dynamic compensation method and system based on shape coefficient

Abstract

The application relates to the technical field of impeller equipment, and discloses an impeller natural frequency dynamic compensation method and system based on a shape coefficient. The mass ratio-frequency difference original curves of each half-open impeller sample are determined by combining finite element simulation and experimental testing; the original curves are fitted, and a shape coefficient is solved to make the original curves overlap, so that a frequency difference reference curve is obtained; a shape coefficient prediction model is trained based on the characteristic parameters of each half-open impeller sample and the corresponding shape coefficient; a target characteristic parameter of a target half-open impeller sample is input into the shape coefficient prediction model to obtain a predicted shape coefficient; the target mass ratio of the target half-open impeller sample is calculated, and the target test frequency of the target half-open impeller sample is determined by experimental testing; and the real natural frequency of the target half-open impeller sample is determined according to the target mass ratio, the target test frequency, the predicted shape coefficient and the frequency difference reference curve. The influence of the sensor mass on the test result of the impeller natural frequency is eliminated.

Description

Technical Field

[0001] This application relates to the field of impeller equipment technology, specifically to a dynamic compensation method and system for the natural frequency of an impeller based on its shape coefficient. Background Technology

[0002] Centrifugal compressors are key equipment in petrochemical, energy, and other fields, and their operational reliability directly affects the stable operation of the entire industrial system. The impeller, as the core rotating component of a centrifugal compressor, has natural frequency characteristics that are closely related to its fatigue life and operational reliability. Therefore, accurately determining the impeller's natural frequency is a crucial step in the design, manufacturing, and maintenance of centrifugal compressors.

[0003] Existing methods for determining the natural frequency of impellers mainly include experimental modal analysis and finite element numerical simulation. However, both methods have significant technical drawbacks in practical applications.

[0004] Experimental modal analysis is a common method for obtaining the natural frequency of an impeller through experimental testing. It uses a piezoelectric accelerometer and a force hammer for excitation, and obtains the frequency through frequency response function testing. However, the added mass of the sensor leads to a systematically low test frequency (for example, when the sensor mass is large and the impeller blade mass is small, the error can even reach more than 50%), which cannot meet the requirements of high-precision design.

[0005] The finite element method (FEM) numerical simulation, based on the impeller's geometric model and material properties, simulates the impeller's modal parameters through numerical calculations to obtain its natural frequencies. This method does not require additional sensor mass, but it necessitates building a separate parametric model for each impeller, resulting in low efficiency. Furthermore, the lack of quantitative analysis of geometric parameters and material properties leads to poor universality.

[0006] In addition to the shortcomings of the two mainstream methods mentioned above, the current determination of impeller natural frequency also faces the following key technical bottlenecks: First, there is no universal model between the mass of the additional sensor and the frequency offset, which cannot accurately describe the influence of the mass of the additional sensor on the experimental test results. Second, the lack of a systematic mechanism to eliminate the influence of material properties makes it impossible to meet the simulation requirements of the natural frequency of impellers with different material properties.

[0007] Therefore, developing a dynamic compensation method for the impeller's natural frequency that can eliminate the influence of the quality of additional sensors, achieve high accuracy and versatility, and has higher testing efficiency has become an urgent technical problem to be solved. Summary of the Invention

[0008] To address the aforementioned issues, this application provides a dynamic compensation method and system for impeller natural frequency based on shape coefficient. This method eliminates the influence of the mass of the additional sensor on the experimental test results of the impeller natural frequency, solves the problem of systematically low experimental natural frequency, establishes a universal mathematical model between mass ratio and frequency difference, achieves accurate calibration of the natural frequency of impellers with arbitrary geometric structures and material properties, and develops an adaptive compensation algorithm based on machine learning. This overcomes the limitations of traditional fixed coefficient correction and significantly improves the accuracy and efficiency of impeller natural frequency testing.

[0009] The embodiments of this application adopt the following technical solutions: In a first aspect, this application provides a dynamic compensation method for the natural frequency of an impeller based on a shape coefficient, including: Samples of each half-open impeller were obtained, and the original mass ratio-frequency difference curves of each half-open impeller sample were determined by combining finite element simulation and experimental testing. The mass ratio is the ratio of the sensor mass of each half-open impeller sample to the blade mass of the corresponding half-open impeller sample, and the frequency difference is the relative deviation between the theoretical true frequency of the finite element simulation and the test frequency of the corresponding experimental test. By fitting each original curve and solving for the shape factor that makes each original curve overlap, a reference curve of the frequency difference between the associated mass ratio and the corresponding shape factor is obtained. An initial shape coefficient prediction model is constructed, and the initial shape coefficient prediction model is trained based on the feature parameters and corresponding shape coefficients of each half-open impeller sample to obtain the shape coefficient prediction model. Obtain a target half-open impeller sample, and input the target feature parameters of the target half-open impeller sample into the shape coefficient prediction model to obtain the predicted shape coefficient of the target half-open impeller sample; Calculate the target mass ratio of the target half-open impeller sample, and determine the target test frequency of the target half-open impeller sample using experimental testing. The true natural frequency of the target semi-open impeller sample is determined based on the target mass ratio, target test frequency, predicted shape factor, and frequency difference baseline curve.

[0010] Secondly, this application also provides a dynamic compensation system for the natural frequency of an impeller based on a shape coefficient, comprising: The original curve establishment unit is used to acquire samples of each half-open impeller and determine the original curve of mass ratio-frequency difference of each half-open impeller sample by combining finite element simulation and experimental testing. Among them, the mass ratio is the ratio of the sensor mass of each half-open impeller sample to the blade mass of the corresponding half-open impeller sample in the experimental test, and the frequency difference is the relative deviation between the theoretical true frequency of finite element simulation and the test frequency of the corresponding experimental test. The baseline curve establishment unit is used to fit each original curve, solve for the shape factor that makes each original curve overlap, and obtain the frequency difference baseline curve of the correlation mass ratio and the corresponding shape factor. The model training unit is used to build an initial shape coefficient prediction model. The initial shape coefficient prediction model is trained based on the feature parameters and corresponding shape coefficients of each half-open impeller sample to obtain the shape coefficient prediction model. The model prediction unit is used to acquire the target half-open impeller sample, input the target feature parameters of the target half-open impeller sample into the shape coefficient prediction model to obtain the predicted shape coefficient of the target half-open impeller sample; The experimental testing unit is used to calculate the target mass ratio of the target half-open impeller sample and to determine the target testing frequency of the target half-open impeller sample through experimental testing. The frequency compensation unit is used to determine the true natural frequency of the target semi-open impeller sample based on the target mass ratio, target test frequency, predicted shape factor, and frequency difference reference curve.

[0011] Thirdly, this application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described dynamic compensation method for impeller natural frequency based on shape coefficient.

[0012] Fourthly, this application also provides a computer-readable storage medium storing a computer program that, when instructed by a processor, implements the steps of the above-described dynamic compensation method for the impeller natural frequency based on the shape coefficient.

[0013] The above-described technical solutions adopted in the embodiments of this application can achieve the following beneficial effects: Eliminating the mass error of the additional sensor and achieving high measurement accuracy: By establishing a frequency difference benchmark curve between the mass ratio and the frequency difference, and combining it with dynamic correction of the shape factor, the influence of the mass of the additional sensor on the test frequency can be accurately quantified, and the relative test error can be controlled within 2%. This solves the core problem of the systematically low measured frequency in the traditional experimental modal analysis method.

[0014] High versatility and wide applicability: Through dimensionless design (mass ratio and frequency difference are both dimensionless ratios), the influence of material properties (elastic modulus and density) is systematically eliminated, and the frequency difference reference curve is applicable to semi-open impellers with any material properties; at the same time, the shape factor prediction model covers impellers with different geometric sizes and structures, eliminating the need for repeated modeling for different impellers, and greatly expanding the scope of application.

[0015] Adaptive dynamic compensation for complex working conditions: The shape coefficient prediction model based on machine learning can output the predicted shape coefficient in real time according to the target feature parameters of the target half-open impeller. Combined with the mass ratio to dynamically calculate the horizontal coordinate, it realizes adaptive compensation under dynamic working conditions, which breaks through the limitation of the traditional fixed coefficient compensation method that cannot adapt to parameter changes, and the compensation accuracy is higher.

[0016] High testing efficiency and reduced computational costs: Only one frequency difference baseline curve and shape coefficient prediction model need to be established and trained. Subsequent testing of the target semi-open impeller does not require finite element simulation. Only the target characteristic parameters and target test frequency need to be collected to complete the calculation of the true natural frequency, reducing the amount of finite element calculation by more than 90%. The testing time is shortened from several hours in the traditional method to tens of minutes, which greatly improves the testing efficiency in industrial production.

[0017] The system is highly integrated and easy to operate: It integrates functional modules such as sample modeling, model training, parameter acquisition, and prediction compensation. The process is standardized and highly automated. Operators do not need to have complex finite element analysis skills. They only need to follow the instructions to complete parameter measurement and input to quickly obtain accurate measurement results, which is convenient for industrial promotion and application. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A flowchart illustrating a dynamic compensation method for impeller natural frequency based on shape coefficient according to an embodiment of this application is shown. Figure 2 The original curve of mass ratio-frequency difference for each half-open impeller sample according to an embodiment of this application is shown. Figure 3 The original curve overlay diagram of a semi-open impeller sample B and a semi-open impeller sample A according to an embodiment of this application is shown. Figure 4 The original curve overlay diagram of a semi-open impeller sample C and a semi-open impeller sample A according to an embodiment of this application is shown; Figure 5 A schematic diagram of a dynamic compensation system for impeller natural frequency based on shape coefficient according to an embodiment of this application is shown; Figure 6 A schematic diagram of the resulting electronic device according to an embodiment of this application is shown. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] The purpose of this application is to provide a dynamic compensation method for the natural frequency of an impeller based on the shape factor, to eliminate the influence of the mass of the additional sensor on the test results of the impeller's natural frequency, and to solve the problem of the systematically low experimental test frequency; to establish a universal mathematical model between mass ratio and frequency difference, so as to achieve accurate testing of the natural frequency of a semi-open impeller with arbitrary geometric structure and arbitrary material properties; to develop an adaptive compensation algorithm based on machine learning, overcoming the limitations of traditional fixed coefficient correction, and achieving intelligent and accurate compensation under dynamic operating conditions; and to integrate testing, prediction, and correction functions, thereby significantly improving the testing efficiency of the impeller's natural frequency and reducing the amount of finite element calculation. To achieve the above objectives, Figure 1 A method for dynamic compensation of impeller natural frequencies based on shape coefficient, according to an embodiment of this application, is illustrated. From Figure 1 As can be seen, this embodiment includes steps S110 to S160: Step S110: Obtain each half-open impeller sample, and determine the original curve of mass ratio-frequency difference for each half-open impeller sample by combining finite element simulation and experimental testing; wherein, the mass ratio is the ratio of the sensor mass of each half-open impeller sample to the blade mass of the corresponding half-open impeller sample in the experimental test, and the frequency difference is the relative deviation between the theoretical true frequency of the finite element simulation and the test frequency of the corresponding experimental test.

[0021] This step involves obtaining samples of each half-open impeller and determining the original mass ratio-frequency difference curve through a combination of finite element simulation and experimental testing.

[0022] First, sample selection is performed.

[0023] Select semi-open impellers with different geometric structures (e.g., different inlet blade lengths, different outlet blade lengths, different inlet blade root widths, different outlet blade root widths, different inlet blade tip widths, different outlet blade tip widths, different blade tip lengths, different blade root lengths, etc.) and different material properties (e.g., common metal materials such as aluminum alloys and titanium alloys) as samples to form a sample set, and make the number of samples as large as possible.

[0024] Next, finite element simulation is performed.

[0025] For a semi-open impeller sample, a parametric finite element model is established using 3D modeling software. The parametric finite element model can accurately reproduce the geometry of the semi-open impeller sample.

[0026] Import the parametric finite element model into the finite element analysis software, set constraints and material properties (such as elastic modulus, density, Poisson's ratio, etc.) consistent with the actual working conditions, perform finite element simulation, and calculate the theoretical true frequency of the semi-open impeller sample.

[0027] The theoretical true frequency of this partially open impeller sample is the frequency value without additional sensor mass interference. For this partially open impeller sample, its theoretical true frequency can be expressed as: ,in, This represents the index value of the partially open impeller sample.

[0028] Then, experimental tests were conducted.

[0029] An experimental modal analysis method using a piezoelectric accelerometer and a force hammer excitation was employed to test the semi-open impeller sample with different sensor masses, obtaining multiple test frequencies corresponding to the semi-open impeller sample. For example, testing the semi-open impeller sample with sensor a yielded one test frequency; testing with sensor b yielded another test frequency; sensor a and sensor b had different masses. The specific experimental testing method utilized existing mature methods.

[0030] The test frequencies for this semi-open impeller sample are the frequency values ​​after adding different sensor masses. For this semi-open impeller sample, each test frequency can be expressed as... ,in, This represents the sensor's index value.

[0031] Simultaneously, the mass of each sensor and the blade mass of the semi-open impeller sample need to be measured to calculate the mass ratios. Each mass ratio represents the ratio of the mass of a different sensor in the semi-open impeller sample to the mass of the blade in the semi-open impeller sample. For this semi-open impeller sample, each mass ratio can be expressed as: .

[0032] The frequency difference is defined as the difference between the theoretical true frequency and the measured frequency, where the numerator is the difference and the denominator is the theoretical true frequency. In other words, the frequency difference is expressed as... .

[0033] Finally, the original curve is established.

[0034] Based on the mass ratios of the semi-open impeller sample The x-axis represents the frequency differences of the semi-open impeller sample. Using the vertical axis as the ordinate, plot the original mass ratio-frequency difference curve for this semi-open impeller sample. This original curve can be represented as follows: The same process was used to plot the original mass ratio-frequency difference curves for each half-open impeller sample in the sample set.

[0035] Figure 2 The original curve of the mass ratio-frequency difference for each half-open impeller sample according to an embodiment of this application is shown. From Figure 2 It can be seen that the semi-open impeller samples A to V have mutually independent original mass ratio-frequency difference curves.

[0036] Therefore, in some optional implementations, step S110, acquiring each half-open impeller sample, and determining the original mass ratio-frequency difference curve of each half-open impeller sample using a combination of finite element simulation and experimental testing, includes: acquiring half-open impellers covering different geometric structures and different material properties as each half-open impeller sample; for a half-open impeller sample, establishing a parameterized finite element model of the half-open impeller sample, and using finite element simulation to calculate the theoretical true frequency of the half-open impeller sample without additional sensor mass interference; for the half-open impeller sample, using an accelerometer and... Using a hammer excitation method, the test frequencies corresponding to different sensor masses were measured for the semi-open impeller sample. The ratio of different sensor masses to blade masses was calculated to obtain the mass ratios for each semi-open impeller sample. For each semi-open impeller sample, the frequency difference was calculated with the difference between the theoretical true frequency and each measured frequency as the numerator and the theoretical true frequency as the denominator. The original mass ratio-frequency difference curve of the semi-open impeller sample was established with each mass ratio as the abscissa and each frequency difference as the ordinate. The corresponding original mass ratio-frequency difference curve was established for each semi-open impeller sample.

[0037] Step S120: Fit each original curve, solve for the shape coefficient that makes each original curve overlap, and obtain the frequency difference benchmark curve of the associated mass ratio and the corresponding shape coefficient.

[0038] This step fits the original curves and solves for the optimal shape coefficients that make the original curves overlap, thereby establishing a frequency difference reference curve.

[0039] First, perform a curve fitting analysis.

[0040] Reference Figure 2 As shown, the original mass ratio-frequency difference curves, which are independent of each other, have highly similar trends and shapes, and differ only in the numerical range of the horizontal axis (mass ratio). Therefore, the overlap of all the original curves can be achieved by scaling the horizontal axis.

[0041] Next, the shape factor is calculated.

[0042] The least squares method is used to fit all the original curves, and the optimal scaling factor that makes the original curves overlap is solved. The optimal scaling factor is defined as the shape coefficient. For a semi-open impeller sample, its shape coefficient can be expressed as: In other words, for any half-open impeller sample, after scaling the x-coordinate of its original curve by its corresponding shape factor, the original curves of all half-open impeller samples will overlap into a single unified curve.

[0043] Finally, a frequency difference baseline curve is established.

[0044] The unified curve resulting from the overlap is defined as the frequency difference reference curve, and its mathematical expression is as follows: This frequency difference reference curve is universally applicable to all semi-open impellers, eliminating the need to repeatedly establish reference curves for different semi-open impellers.

[0045] Figure 3 The original curve overlay diagram of a semi-open impeller sample B and a semi-open impeller sample A according to an embodiment of this application is shown. Figure 4 The original curve overlay diagram of a semi-open impeller sample C and a semi-open impeller sample A according to an embodiment of this application is shown. Figure 3 and Figure 4 In the diagram, the origin represents the data points of half-open impeller sample A, and the square points represent the data points of half-open impeller samples B and C. Figure 3 and Figure 4 It can be seen that the semi-open impeller sample B, after passing through the shape coefficient... After scaling, the curve overlaps with the original curve of the half-open impeller sample A, while the half-open impeller sample C is adjusted for shape factor. After scaling, it overlaps with the original curve of the half-open impeller sample A.

[0046] Therefore, in some optional implementations, step S120, fitting each original curve and solving for the shape coefficient that makes each original curve overlap, to obtain a frequency difference reference curve of the associated mass ratio and the corresponding shape coefficient, includes: fitting each original curve using the least squares method and solving for the shape coefficient that makes each original curve overlap; establishing a frequency difference reference curve with the product of the corresponding mass ratio and the shape coefficient as the abscissa and the frequency difference as the ordinate.

[0047] Step S130: Construct an initial shape coefficient prediction model. Train the initial shape coefficient prediction model based on the feature parameters and corresponding shape coefficients of each half-open impeller sample to obtain the shape coefficient prediction model.

[0048] This step involves building and training a shape coefficient prediction model.

[0049] First, an initial shape coefficient prediction model is constructed.

[0050] A Multilayer Perceptron (MLP) was chosen as the initial shape coefficient prediction model. The MLP possesses strong nonlinear mapping capabilities, effectively handling the complex relationship between multidimensional feature parameters and a single output. The input layer of the MLP consists of 9 neurons and may include 3 hidden layers. The ReLU activation function is used to avoid the vanishing gradient problem. The output layer consists of 1 neuron, the loss function is mean squared error, and the optimizer is the Adam optimizer.

[0051] Next, determine the characteristic parameters.

[0052] Parameters that can comprehensively reflect the blade characteristics of the semi-open impeller sample are selected as the model input features. The specific feature parameters include: inlet blade length, outlet blade length, inlet blade root width, outlet blade root width, inlet blade tip width, outlet blade tip width, blade tip length, blade root length, and blade mass.

[0053] Normalize each feature parameter. For example, use the min-max normalization method to map the feature parameters to the [0, 1] interval to avoid interference with model training caused by differences in the units and numerical ranges of the feature parameters.

[0054] Finally, the model is trained.

[0055] The normalized feature parameters of each half-open impeller sample are used as input, and the shape coefficients of the corresponding half-open impeller samples are used as output. Training and testing sets are created, and the initial shape coefficient prediction model is trained. An early stopping strategy can be used during training. After training, the final shape coefficient prediction model is obtained.

[0056] Therefore, in some optional implementations, step S130, constructing an initial shape coefficient prediction model, and training the initial shape coefficient prediction model based on the feature parameters and corresponding shape coefficients of each half-open impeller sample to obtain the shape coefficient prediction model, includes: establishing a multilayer perceptron model as the initial shape coefficient prediction model; collecting feature parameters of each half-open impeller sample; wherein, the feature parameters include: inlet blade length, outlet blade length, inlet blade root width, outlet blade root width, inlet blade tip width, outlet blade tip width, blade tip length, blade root length, and blade mass; using the normalized feature parameters of each half-open impeller sample as input, the shape coefficients of the corresponding half-open impeller samples as output, and the mean square error as the loss function to train the initial shape coefficient prediction model to obtain the shape coefficient prediction model.

[0057] Step S140: Obtain the target half-open impeller sample, and input the target feature parameters of the target half-open impeller sample into the shape coefficient prediction model to obtain the predicted shape coefficient of the target half-open impeller sample.

[0058] This step predicts the shape coefficient of the target semi-open impeller sample based on the target feature parameters of the target semi-open impeller sample.

[0059] First, obtain a sample of the target half-open impeller.

[0060] The target half-open impeller sample is the half-open impeller sample whose natural frequency needs to be tested.

[0061] Next, determine the target feature parameters.

[0062] Consistent with the characteristic parameters of the semi-open impeller samples in the sample set, the target characteristic parameters of the target semi-open impeller samples also include: inlet blade length, outlet blade length, inlet blade root width, outlet blade root width, inlet blade tip width, outlet blade tip width, blade tip length, blade root length, and blade mass.

[0063] The target feature parameters are mapped to the [0, 1] interval using the same normalization method as the sample set.

[0064] Finally, the predicted shape coefficient is output.

[0065] The normalized feature parameters of the target semi-open impeller sample are input into the trained shape coefficient prediction model, which then outputs the predicted shape coefficient of the target semi-open impeller. The predicted shape coefficient can be expressed as... .

[0066] Therefore, in some optional implementations, step S140, obtaining a target semi-open impeller sample and inputting the target feature parameters of the target semi-open impeller sample into a shape coefficient prediction model to obtain the predicted shape coefficient of the target semi-open impeller sample, includes: obtaining the target semi-open impeller sample and collecting the target feature parameters of the target semi-open impeller sample; wherein, the target feature parameters include: inlet blade length, outlet blade length, inlet blade root width, outlet blade root width, inlet blade tip width, outlet blade tip width, blade tip length, blade root length, and blade mass; inputting the normalized target feature parameters into the shape coefficient prediction model and using the shape coefficient prediction model to output the predicted shape coefficient.

[0067] Step S150: Calculate the target mass ratio of the target half-open impeller sample and determine the target test frequency of the target half-open impeller sample using experimental testing.

[0068] This step calculates the mass ratio of the target semi-open impeller and determines the target test frequency.

[0069] First, calculate the mass ratio of the target half-open impeller sample.

[0070] The mass of the piezoelectric accelerometer of the semi-open impeller sample used in the experimental test is obtained, and the mass of the sensor is expressed as: Obtain the blade mass of the target semi-open impeller sample, which is denoted as... Therefore, the mass ratio of the target semi-open impeller sample is expressed as: .

[0071] Using the same experimental modal analysis method as the sample set, the target semi-open impeller sample was tested using the piezoelectric accelerometer and force hammer excitation method to determine the target test frequency of the target semi-open impeller sample. The target test frequency is the frequency value after adding the mass of the modified piezoelectric accelerometer. The target test frequency can be expressed as: .

[0072] Therefore, in some optional implementations, step S150, calculating the target mass ratio of the target half-open impeller sample and determining the target test frequency of the target half-open impeller sample using experimental testing, includes: obtaining the sensor mass and blade mass of the target half-open impeller sample for experimental testing, calculating the ratio of sensor mass to blade mass as the target mass ratio; and testing the target test frequency of the target half-open impeller sample after adding sensor mass using an accelerometer and a force hammer excitation method.

[0073] Step S160: Determine the true natural frequency of the target semi-open impeller sample based on the target mass ratio, target test frequency, predicted shape factor, and frequency difference baseline curve.

[0074] This step calculates the true natural frequency of the target semi-open impeller sample based on the baseline curve and predicted shape coefficient, according to the target mass ratio and target test frequency.

[0075] First, calculate the target's x-coordinate.

[0076] The target abscissa is calculated based on the target mass ratio and the predicted shape factor. The target abscissa is represented as follows: .

[0077] Next, query the target frequency differences.

[0078] Substituting the target x-axis into the frequency difference baseline curve, the corresponding target frequency difference can be obtained. The target frequency difference can be expressed as... .

[0079] Finally, we can deduce the true inherent frequency.

[0080] According to the definition of frequency difference By reverse calculation, the true natural frequency can be obtained. .

[0081] Therefore, in some optional implementations, step S160, determining the true natural frequency of the target semi-open impeller sample based on the target mass ratio, target test frequency, predicted shape factor, and frequency difference reference curve, includes: calculating the target abscissa based on the target mass ratio and predicted shape factor, querying the target abscissa in the frequency difference reference curve to obtain the corresponding target frequency difference; and calculating the true natural frequency of the target semi-open impeller sample based on the target frequency difference and the target test frequency.

[0082] The beneficial effects of this application will now be explained in detail.

[0083] This application constructs a material-independent mechanism through dimensionless design, making the frequency difference reference curve universally applicable to semi-open impellers with any material properties, without requiring additional corrections for specific materials. Its core principle lies in the fact that the first-order natural frequency of the impeller blades is inherently directly related to material properties (elastic modulus, density, etc.), and traditional testing methods suffer from limited universality due to the failure to isolate the influence of material properties. The mass ratio defined in this application... and frequency differences All are dimensionless ratios, achieving systematic cancellation of material properties during curve generation and benchmark establishment. Specifically, mass ratio The calculation only involves the ratio relationship of the mass dimension; the material density factor included in the blade mass does not have a specific effect on the ratio calculation; frequency differences The calculation is based on the theoretical true frequency. Both the theoretical true frequency and the test frequency contain the same scaling factor determined by the material's elastic modulus and density. These factors cancel each other out in the difference and ratio calculations between the numerator and denominator, ultimately resulting in... Completely decoupled from material properties. This design enables a reference curve for frequency differences based on shape factor normalization. By eliminating the dependence on specific material parameters, the frequency correction of semi-open impellers made of different materials such as aluminum alloy, titanium alloy or alloy steel can be directly performed using this reference curve, which greatly expands the applicability of the technical solution and solves the technical pain point of traditional methods that require parameter adjustment for different materials.

[0084] The method proposed in this application will be described below through specific embodiments.

[0085] The mass of the blades of a certain aluminum alloy impeller is The experiment tested the mass of the additional sensor on the aluminum alloy impeller. Therefore, the mass ratio is calculated to be... .

[0086] The test frequency is .

[0087] Based on the normalized target characteristic parameters of the aluminum alloy impeller, the predicted shape coefficient is obtained. Therefore, the target's x-coordinate is calculated as follows: .

[0088] Query the frequency difference benchmark curve to obtain the corresponding target frequency difference. .

[0089] The true natural frequency of the aluminum alloy impeller is calculated as follows: .

[0090] Results of finite element simulation .

[0091] After dynamic compensation, the actual natural frequency of the aluminum alloy impeller deviates from the result of finite element simulation by less than 2%.

[0092] Figure 5 An embodiment of the impeller natural frequency dynamic compensation system based on the shape coefficient according to this application is shown, from Figure 5 It can be seen that the impeller natural frequency dynamic compensation system 500 based on the shape factor includes: The original curve establishment unit 510 is used to acquire samples of each half-open impeller and determine the original curve of the mass ratio-frequency difference of each half-open impeller sample by combining finite element simulation and experimental testing. Among them, the mass ratio is the ratio of the sensor mass of each half-open impeller sample to the blade mass of the corresponding half-open impeller sample in the experimental test, and the frequency difference is the relative deviation between the theoretical true frequency of the finite element simulation and the test frequency of the corresponding experimental test. The reference curve establishment unit 520 is used to fit each original curve, solve for the shape coefficient that makes each original curve overlap, and obtain the frequency difference reference curve of the correlation mass ratio and the corresponding shape coefficient. The model training unit 530 is used to construct an initial shape coefficient prediction model. The initial shape coefficient prediction model is trained based on the feature parameters and corresponding shape coefficients of each half-open impeller sample to obtain the shape coefficient prediction model. The model prediction unit 540 is used to acquire the target half-open impeller sample and input the target feature parameters of the target half-open impeller sample into the shape coefficient prediction model to obtain the predicted shape coefficient of the target half-open impeller sample. Experimental test unit 550 is used to calculate the target mass ratio of the target half-open impeller sample and to determine the target test frequency of the target half-open impeller sample using experimental testing. The frequency compensation unit 560 is used to determine the true natural frequency of the target semi-open impeller sample based on the target mass ratio, target test frequency, predicted shape factor, and frequency difference reference curve.

[0093] In some optional embodiments, in the above system, the original curve establishment unit 510 is used to: acquire semi-open impellers covering different geometries and different material properties as samples of each semi-open impeller; for a sample of a semi-open impeller, establish a parameterized finite element model of the sample of the semi-open impeller, and use finite element simulation to calculate the theoretical true frequency of the sample of the semi-open impeller without the interference of the mass of the additional sensor; for the sample of the semi-open impeller, use an accelerometer and a force hammer to test the test frequencies corresponding to the addition of different sensor masses to the sample of the semi-open impeller, and calculate the ratio of the mass of the different sensors to the mass of the blade to obtain each mass ratio; for the sample of the semi-open impeller, calculate each frequency difference with the difference between the theoretical true frequency and each measured frequency as the numerator and the theoretical true frequency as the denominator, and establish the original mass ratio-frequency difference curve of the sample of the semi-open impeller with each mass ratio as the abscissa and each frequency difference as the ordinate; establish a corresponding original mass ratio-frequency difference curve for each sample of the semi-open impeller.

[0094] In some alternative implementations, in the above system, the reference curve establishment unit 520 is used to: fit each original curve using the least squares method, solve for the shape factor that makes each original curve overlap; and establish a frequency difference reference curve with the product of the corresponding mass ratio and the shape factor as the abscissa and the frequency difference as the ordinate.

[0095] In some optional implementations, in the above system, the model training unit 530 is used to: establish a multilayer perceptron model as an initial shape coefficient prediction model; collect feature parameters of each half-open impeller sample; wherein, the feature parameters include: inlet blade length, outlet blade length, inlet blade root width, outlet blade root width, inlet blade tip width, outlet blade tip width, blade tip length, blade root length, and blade mass; use the normalized feature parameters of each half-open impeller sample as input, use the shape coefficient of the corresponding half-open impeller sample as output, and use the mean square error as the loss function to train the initial shape coefficient prediction model to obtain the shape coefficient prediction model.

[0096] In some optional embodiments, in the above system, the model prediction unit 540 is used to: acquire a target half-open impeller sample, and collect target feature parameters of the target half-open impeller sample; wherein, the target feature parameters include: inlet blade length, outlet blade length, inlet blade root width, outlet blade root width, inlet blade tip width, outlet blade tip width, blade tip length, blade root length, and blade mass; input the normalized target feature parameters into the shape coefficient prediction model, and output the predicted shape coefficient using the shape coefficient prediction model.

[0097] In some optional embodiments, in the above system, the experimental testing unit 550 is used to: acquire the sensor mass and blade mass of the target half-open impeller sample, calculate the ratio of sensor mass to blade mass as the target mass ratio; and test the target test frequency of the target half-open impeller sample after adding sensor mass using an accelerometer and a force hammer excitation method.

[0098] In some optional implementations, in the above system, the frequency compensation unit 560 is used to: calculate the target abscissa based on the target mass ratio and the predicted shape factor, look up the target abscissa in the frequency difference reference curve to obtain the corresponding target frequency difference; and calculate the true natural frequency of the target half-open impeller sample based on the target frequency difference and the target test frequency.

[0099] It should be noted that the aforementioned impeller natural frequency dynamic compensation system 500 based on the shape factor can implement the aforementioned impeller natural frequency dynamic compensation method based on the shape factor, which will not be elaborated further.

[0100] Figure 6 This invention illustrates a schematic diagram of the structure of an electronic device according to an embodiment of the present application. Figure 6 As shown, the electronic device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used for communication with external devices via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a dynamic compensation method for the impeller natural frequency based on the shape factor.

[0101] In one embodiment, the electronic device provided in this application includes a memory and a processor. The memory stores a database and a computer program that can run on the processor. When the processor executes the computer program, it implements the steps of the aforementioned dynamic compensation method for impeller natural frequency based on shape coefficient.

[0102] The above is as stated in this application. Figure 5The method for dynamic compensation of impeller natural frequency based on shape coefficient disclosed in the illustrated embodiment can be applied to a processor or implemented by a processor. During implementation, each step of the above method can be completed by integrated logic circuits in the processor's hardware or by instructions in software form. The processor can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. The steps of the method disclosed in the embodiments of this application can be directly implemented by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method.

[0103] In one embodiment, a computer-readable storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements the steps of the aforementioned dynamic compensation method for impeller natural frequency based on shape coefficient.

[0104] It should be noted that the functions or steps that the above-mentioned electronic devices or computer-readable storage media can achieve can be referred to the relevant descriptions in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0105] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0106] Those skilled in the art will understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above.

[0107] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A dynamic compensation method for the natural frequency of an impeller based on its shape factor, characterized in that, include: Samples of each half-open impeller were obtained, and the original mass ratio-frequency difference curves of each half-open impeller sample were determined by combining finite element simulation and experimental testing. The mass ratio is the ratio of the sensor mass of each half-open impeller sample to the blade mass of the corresponding half-open impeller sample, and the frequency difference is the relative deviation between the theoretical true frequency of the finite element simulation and the test frequency of the corresponding experimental test. By fitting each original curve and solving for the shape factor that makes each original curve overlap, a reference curve of the frequency difference between the associated mass ratio and the corresponding shape factor is obtained. An initial shape coefficient prediction model is constructed, and the initial shape coefficient prediction model is trained based on the feature parameters and corresponding shape coefficients of each half-open impeller sample to obtain the shape coefficient prediction model. Obtain a target half-open impeller sample, and input the target feature parameters of the target half-open impeller sample into the shape coefficient prediction model to obtain the predicted shape coefficient of the target half-open impeller sample; Calculate the target mass ratio of the target half-open impeller sample, and determine the target test frequency of the target half-open impeller sample using experimental testing. The true natural frequency of the target semi-open impeller sample is determined based on the target mass ratio, target test frequency, predicted shape factor, and frequency difference baseline curve.

2. The method according to claim 1, characterized in that, The process of obtaining samples from each half-open impeller and determining the original mass ratio-frequency difference curve for each half-open impeller sample using a combination of finite element simulation and experimental testing includes: Obtain semi-open impellers with different geometries and material properties as samples of each semi-open impeller; For a semi-open impeller sample, a parameterized finite element model of the semi-open impeller sample is established, and the theoretical true frequency of the semi-open impeller sample without the mass interference of additional sensors is calculated using finite element simulation. For the semi-open impeller sample, the test frequencies corresponding to the addition of different sensor masses were tested by using an accelerometer and a force hammer excitation method. The ratio of different sensor masses to blade masses of the semi-open impeller sample was calculated to obtain each mass ratio. For the semi-open impeller sample, the frequency difference is calculated with the difference between the theoretical true frequency and each measured frequency as the numerator and the theoretical true frequency as the denominator. The original mass ratio-frequency difference curve of the semi-open impeller sample is established with each mass ratio as the abscissa and each frequency difference as the ordinate. For each half-open impeller sample, a corresponding original curve of mass ratio-frequency difference was established.

3. The method according to claim 1, characterized in that, The process of fitting each original curve, solving for the shape coefficient that causes the original curves to overlap, and obtaining the frequency difference benchmark curve of the associated mass ratio and the corresponding shape coefficient includes: The least squares method is used to fit each original curve, and the shape coefficient that makes each original curve overlap is solved. Establish a frequency difference baseline curve with the product of the corresponding mass ratio and shape factor as the x-axis and the frequency difference as the y-axis.

4. The method according to claim 1, characterized in that, The initial shape coefficient prediction model is constructed by training the initial shape coefficient prediction model based on the feature parameters and corresponding shape coefficients of each half-open impeller sample, resulting in the shape coefficient prediction model, including: A multilayer perceptron model was established as the initial shape coefficient prediction model; Collect characteristic parameters of each half-open impeller sample; among which, the characteristic parameters include: inlet blade length, outlet blade length, inlet blade root width, outlet blade root width, inlet blade tip width, outlet blade tip width, blade tip length, blade root length, and blade mass; The initial shape coefficient prediction model is trained by using the normalized feature parameters of each half-open impeller sample as input and the shape coefficient of the corresponding half-open impeller sample as output, with the mean square error as the loss function, and the shape coefficient prediction model is obtained.

5. The method according to claim 1, characterized in that, The process of obtaining a target semi-open impeller sample and inputting the target feature parameters of the target semi-open impeller sample into a shape coefficient prediction model to obtain the predicted shape coefficient of the target semi-open impeller sample includes: Obtain a target semi-open impeller sample and collect the target feature parameters of the target semi-open impeller sample; among which, the target feature parameters include: inlet blade length, outlet blade length, inlet blade root width, outlet blade root width, inlet blade tip width, outlet blade tip width, blade tip length, blade root length, and blade mass; The normalized target feature parameters are input into the shape coefficient prediction model, and the predicted shape coefficient is output using the shape coefficient prediction model.

6. The method according to claim 1, characterized in that, The calculation of the target mass ratio of the target semi-open impeller sample and the determination of the target testing frequency of the target semi-open impeller sample using experimental testing include: Obtain the sensor mass and blade mass of the target half-open impeller sample for experimental testing, and calculate the ratio of sensor mass to blade mass as the target mass ratio. The target test frequency of a target half-open impeller sample after adding sensor mass was tested using an accelerometer and force hammer excitation method.

7. The method according to claim 1, characterized in that, The determination of the true natural frequency of the target semi-open impeller sample based on the target mass ratio, target test frequency, predicted shape factor, and frequency difference baseline curve includes: Calculate the target abscissa based on the target mass ratio and the predicted shape factor, and then look up the target frequency difference in the frequency difference baseline curve. Based on the target frequency difference and the target test frequency, the true natural frequency of the target half-open impeller sample is calculated.

8. A dynamic compensation system for the natural frequency of an impeller based on a shape factor, characterized in that, include: The original curve establishment unit is used to acquire samples of each half-open impeller and determine the original curve of mass ratio-frequency difference of each half-open impeller sample by combining finite element simulation and experimental testing. Among them, the mass ratio is the ratio of the sensor mass of each half-open impeller sample to the blade mass of the corresponding half-open impeller sample in the experimental test, and the frequency difference is the relative deviation between the theoretical true frequency of finite element simulation and the test frequency of the corresponding experimental test. The baseline curve establishment unit is used to fit each original curve, solve for the shape factor that makes each original curve overlap, and obtain the frequency difference baseline curve of the correlation mass ratio and the corresponding shape factor. The model training unit is used to build an initial shape coefficient prediction model. The initial shape coefficient prediction model is trained based on the feature parameters and corresponding shape coefficients of each half-open impeller sample to obtain the shape coefficient prediction model. The model prediction unit is used to acquire the target half-open impeller sample, input the target feature parameters of the target half-open impeller sample into the shape coefficient prediction model to obtain the predicted shape coefficient of the target half-open impeller sample; The experimental testing unit is used to calculate the target mass ratio of the target half-open impeller sample and to determine the target testing frequency of the target half-open impeller sample through experimental testing. The frequency compensation unit is used to determine the true natural frequency of the target semi-open impeller sample based on the target mass ratio, target test frequency, predicted shape factor, and frequency difference reference curve.

9. An electronic 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 computer program, it implements the steps of the dynamic compensation method for impeller natural frequency based on shape coefficient as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is instructed by the processor, it implements the steps of the dynamic compensation method for impeller natural frequency based on shape coefficient as described in any one of claims 1 to 7.

Images