Test method for fatigue life of diaphragm compressor diaphragm

Abstract

The present application relates to the technical field of fatigue life testing, and specifically discloses a method for testing the fatigue life of a diaphragm compressor diaphragm, which measures the material tensile strength value of each diaphragm tensile specimen through a tensile fatigue test, and at the same time combines the geometric parameters of each diaphragm tensile specimen to construct an association representation between the geometric parameters of the diaphragm tensile specimen and the tensile test results, and then, by performing a global aggregation analysis on each group of constructed association representations, the σb value of the diaphragm compressor diaphragm material is intelligently estimated. In this way, the tensile performance of the diaphragm material can be more accurately evaluated, thereby providing a more reliable basis for the diaphragm fatigue life test.

Description

Technical Field

[0001] The present application relates to the technical field of fatigue life testing, and more specifically, to a method for testing the fatigue life of a diaphragm of a diaphragm compressor. Background Art

[0002] Diaphragm compressor is a kind of positive displacement compressor widely used in laboratory research, oil and gas industry, chemical industry, metallurgy industry, pharmaceutical industry and power industry. Its working principle is to drive the metal diaphragm through the hydraulic system to achieve the compression of process gas. As a key component in the diaphragm compressor, the metal diaphragm must be able to withstand repeated pressure changes and mechanical stress. Therefore, its fatigue life is one of the key indicators for evaluating the reliability and durability of the diaphragm compressor.

[0003] The invention patent with publication number CN117214014A discloses a test method for the fatigue life of a diaphragm compressor diaphragm. The method first groups the diaphragm tensile specimens for tensile fatigue tests and calculates the tensile strength σb value of the diaphragm material. Next, the fatigue test stress level is determined according to the actual typical working conditions, and standard fatigue test pieces are made to determine the test grouping. Then, the SN curve (stress-life curve) of the diaphragm material is fitted based on the fatigue test data as the boundary condition input for computer-aided engineering (CAE) analysis. At the same time, the actual three-dimensional model of the diaphragm is imported into CAE, the constraints, temperature and historical load conditions are adjusted, and the fatigue numerical analysis software is used to solve the life curve of the sample.

[0004] However, in actual applications, the geometric parameters of each diaphragm tensile specimen (such as thickness, width, length, etc.) significantly affect its stress distribution and deformation mode, thereby affecting its tensile strength test results. When calculating the tensile strength σb value of the diaphragm material, the above scheme simply removes invalid data from the tensile strength test results of each diaphragm tensile specimen and takes the average value, ignoring the potential impact of the diaphragm geometric parameters on the tensile strength test results, and cannot truly reflect the diaphragm performance under different geometric parameters. Therefore, it may lead to inaccurate evaluation of diaphragm material performance, which in turn affects the fatigue life prediction based on it.

[0005] Therefore, an optimized testing method for the fatigue life of diaphragm compressor diaphragms is expected. Summary of the invention

[0006] The present application provides a method for testing the fatigue life of a diaphragm of a diaphragm compressor, which can more accurately evaluate the tensile properties of the diaphragm material, thereby providing a more reliable basis for diaphragm fatigue life testing.

[0007] In a first aspect, a method for testing the fatigue life of a diaphragm of a diaphragm compressor is provided, comprising: grouping diaphragm stretching specimens, performing a stretching fatigue test to calculate a σb value; determining a fatigue test stress level according to actual typical working conditions; making fatigue test standard parts; determining fatigue test groups; fitting an SN curve of a diaphragm material according to fatigue test data; inputting the SN curve of a diaphragm material into CAE as a boundary condition; inputting an actual three-dimensional model of the diaphragm into CAE; changing diaphragm constraints, temperature, and historical load conditions, and solving through a solver of fatigue numerical analysis software to obtain a life curve of the specimen; and issuing a fatigue analysis report containing life prediction data, wherein the diaphragm stretching specimens are grouped, and a stretching fatigue test is performed to calculate a σb value, comprising:

[0008] Obtaining geometric parameters of each diaphragm stretching sample, wherein the geometric parameters include geometric shape, dimensional accuracy, thickness and parallelism;

[0009] Obtaining the material tensile strength value of each diaphragm tensile sample through a tensile fatigue test;

[0010] Based on the geometric parameters and material tensile strength values ​​of each diaphragm tensile specimen, a set of diaphragm tensile specimen geometric parameter-test result splicing coding vectors is constructed;

[0011] Performing feature aggregation based on feature dynamics analysis on the set of the diaphragm stretching sample geometric parameter-test result splicing coding vectors to obtain a diaphragm stretching sample geometric parameter-test result significant aggregation coding vector;

[0012] The σb value is determined based on the diaphragm stretching sample geometric parameters-test results significant aggregation coding vector.

[0013] The present application provides a method for testing the fatigue life of a diaphragm compressor diaphragm, which measures the material tensile strength value of each diaphragm tensile specimen through a tensile fatigue test, and at the same time, combines the geometric parameters of each diaphragm tensile specimen to construct an association representation between the geometric parameters of the diaphragm tensile specimen and the tensile test results, and then, by performing a global aggregation analysis on each group of constructed association representations, the σb value of the diaphragm compressor diaphragm material is intelligently estimated. In this way, the tensile performance of the diaphragm material can be more accurately evaluated, thereby providing a more reliable basis for the diaphragm fatigue life test. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments of the present application are briefly introduced below. Obviously, the drawings described below only relate to some embodiments of the present application, and are not intended to limit the present application.

[0015] Figure 1The present invention is a schematic flow chart of a method for testing the fatigue life of a diaphragm compressor diaphragm according to an embodiment of the present application.

[0016] Figure 2 This is a schematic flow chart of a method for testing the fatigue life of a diaphragm compressor diaphragm according to an embodiment of the present application, in which diaphragm tensile specimens are grouped and subjected to a tensile fatigue test to calculate a σb value.

[0017] Figure 3 This is a data flow diagram of a method for testing the fatigue life of a diaphragm compressor diaphragm in an embodiment of the present application, in which diaphragm tensile specimens are grouped and subjected to tensile fatigue tests to calculate σb values.

[0018] Figure 4 In the testing method for the fatigue life of the diaphragm of the diaphragm compressor of the embodiment of the present application, a schematic flowchart of a set of diaphragm tensile specimen geometric parameter-test result splicing coding vectors is constructed based on the geometric parameters and material tensile strength values ​​of each diaphragm tensile specimen.

[0019] Figure 5 In the testing method for the fatigue life of the diaphragm of a diaphragm compressor in an embodiment of the present application, a schematic flow chart of a diaphragm tensile specimen geometric parameter-test result splicing coding vector set is subjected to feature aggregation based on feature dynamics analysis to obtain a diaphragm tensile specimen geometric parameter-test result significant aggregation coding vector. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0021] As mentioned in the background technology above, patent CN117214014A proposes a test method for the fatigue life of a diaphragm compressor diaphragm, including: S101, grouping diaphragm tensile specimens, and performing tensile fatigue tests to calculate σb values; S102, determining fatigue test stress levels according to actual typical working conditions; S103, making fatigue test standard parts; S104, determining fatigue test groups; S105, fitting the SN curve of the diaphragm material according to fatigue test data; S106, inputting the SN curve of the diaphragm material into CAE as boundary conditions; S107, inputting the actual three-dimensional model of the diaphragm into CAE; S108, changing the diaphragm constraints, temperature, and historical load conditions, and solving through the solver of the fatigue numerical analysis software to obtain the life curve of the sample; S109, issuing a fatigue analysis report containing life prediction data.

[0022] The tensile strength σb value is the maximum stress that a material can withstand before breaking, and is a key indicator for evaluating the mechanical properties of a material. However, in actual tensile strength tests, the geometric parameters of each diaphragm tensile specimen (such as thickness, width, length, etc.) significantly affect its stress distribution and deformation mode, thereby affecting its tensile strength test results. When calculating the tensile strength σb value of the diaphragm material, the above scheme simply removes invalid data from the tensile strength test results of each diaphragm tensile specimen and takes the average value, ignoring the potential impact of the diaphragm geometric parameters on the tensile strength test results, and cannot truly reflect the diaphragm performance under different geometric parameters. Therefore, it may lead to an inaccurate evaluation of the diaphragm material performance, which in turn affects the fatigue life prediction based on this.

[0023] In view of the above technical problems, the technical concept of this application is: to measure the material tensile strength value of each diaphragm tensile specimen through a tensile fatigue test, and at the same time, to construct the correlation representation between the geometric parameters of the diaphragm tensile specimen and the tensile test results in combination with the geometric parameters of each diaphragm tensile specimen, and then, to perform a global aggregation analysis on each group of constructed correlation representations, so as to intelligently estimate the σb value of the diaphragm material of the diaphragm compressor. In this way, the tensile performance of the diaphragm material can be more accurately evaluated, thereby providing a more reliable basis for the diaphragm fatigue life test.

[0024] like Figure 2 and Figure 3 As shown, the diaphragm stretching specimens are grouped and subjected to a tensile fatigue test to calculate the σb value, including: S1, obtaining the geometric parameters of each diaphragm stretching specimen, the geometric parameters including geometric shape, dimensional accuracy, thickness and parallelism; S2, obtaining the material tensile strength value of each diaphragm stretching specimen through a tensile fatigue test; S3, constructing a set of diaphragm stretching specimen geometric parameter-test result splicing coding vectors based on the geometric parameters and material tensile strength values ​​of each diaphragm stretching specimen; S4, performing feature aggregation based on feature dynamics analysis on the set of diaphragm stretching specimen geometric parameter-test result splicing coding vectors to obtain a diaphragm stretching specimen geometric parameter-test result significantly aggregated coding vector; S5, determining the σb value based on the diaphragm stretching specimen geometric parameter-test result significantly aggregated coding vector.

[0025] Exemplarily, in steps S1 and S2, the geometric parameters of each diaphragm stretching specimen are obtained, and the geometric parameters include geometric shape, dimensional accuracy, thickness and parallelism, and the material tensile strength value of each diaphragm stretching specimen is obtained through a tensile fatigue test. It should be understood that the geometric parameters of the diaphragm stretching specimen directly affect the stress distribution and deformation mode of the diaphragm when it is subjected to force. For example, different geometric shapes will lead to different stress concentration phenomena, while dimensional accuracy, thickness and parallelism will affect the effective area and uniformity of the actual load bearing. Therefore, changes in the geometric parameters of the diaphragm stretching specimen will significantly affect the performance of the material under tensile conditions. Based on this, in order to improve the accuracy and reliability of the test results, the present application collects the detailed geometric parameters and corresponding tensile strength data of each diaphragm stretching specimen to construct an association between the geometric parameters of the diaphragm stretching specimen and the tensile test results, thereby achieving a more accurate evaluation of the tensile strength performance of the diaphragm material.

[0026] Exemplarily, in step S3, based on the geometric parameters and material tensile strength values ​​of each diaphragm stretching specimen, a set of diaphragm stretching specimen geometric parameters-test result splicing coding vectors is constructed. Specifically, based on the geometric parameters and material tensile strength values ​​of each diaphragm stretching specimen, a set of diaphragm stretching specimen geometric parameters-test result splicing coding vectors can be constructed to comprehensively consider the influence of the diaphragm stretching specimen geometric parameters (such as shape, dimensional accuracy, thickness and parallelism) on its stress distribution and deformation mode, thereby affecting its tensile strength test results. Different geometric shapes will lead to different stress concentration phenomena, while dimensional accuracy, thickness and parallelism will affect the effective area and uniformity of the actual load bearing. By constructing a set of geometric parameter-test result splicing coding vectors, complex multi-dimensional geometric parameters can be combined with tensile strength data to form a unified, fixed-length vector space, simplifying the subsequent data processing process, so that machine learning algorithms or statistical models can more effectively extract useful information from large amounts of data. At the same time, the geometric parameters are encoded using embedded coding technology and spliced ​​together with the results of one-hot encoding of tensile strength, which can reveal the material performance laws under different combinations of geometric parameters. It can not only capture the influence of a single parameter, but also discover the synergistic effect between multiple parameters, and better predict the performance of the diaphragm under actual working conditions.

[0027] In one embodiment, Figure 4As shown, based on the geometric parameters and material tensile strength values ​​of each diaphragm stretching specimen, a set of diaphragm stretching specimen geometric parameter-test result splicing coding vectors is constructed, including: S31, embedding coding the geometric parameters of each diaphragm stretching specimen to obtain a set of diaphragm stretching specimen geometric parameter embedded coding vectors; S32, one-hot coding the material tensile strength values ​​of each diaphragm stretching specimen to obtain a set of diaphragm stretching specimen tensile strength one-hot coding vectors; S33, splicing each group of corresponding diaphragm stretching specimen geometric parameter embedded coding vectors and diaphragm stretching specimen tensile strength one-hot coding vectors in the set of diaphragm stretching specimen geometric parameter embedded coding vectors and the set of diaphragm stretching specimen tensile strength one-hot coding vectors to obtain the set of diaphragm stretching specimen geometric parameter-test result splicing coding vectors.

[0028] Exemplarily, in step S31, the geometric parameters of each diaphragm stretching specimen are embedded and encoded to obtain a set of embedded coding vectors of the geometric parameters of the diaphragm stretching specimens. It should be understood that considering that the geometric parameters of the diaphragm stretching specimens, such as shape, dimensional accuracy, thickness and parallelism, may be classified data or values ​​with different dimensions, it is difficult to use them directly for computer processing and analysis. Therefore, the present application further adopts embedded coding technology to embed the geometric parameters of each diaphragm stretching specimen respectively, so as to map the multi-dimensional geometric parameters of the diaphragm stretching specimens to a unified, fixed-length vector space, so as to learn the potential correlation between the multi-dimensional geometric parameters, generate the geometric parameter embedded coding vectors corresponding to each diaphragm stretching specimen, thereby providing a data basis for the subsequent analysis of the synergistic effect of specific geometric parameter combinations on the tensile strength of the material.

[0029] In one embodiment, the geometric parameters of each diaphragm stretching sample are embedded and encoded to obtain a set of embedded encoding vectors of the geometric parameters of the diaphragm stretching samples, including: using an embedded encoder based on a fully connected layer to embed the geometric parameters of each diaphragm stretching sample to obtain a set of embedded encoding vectors of the geometric parameters of the diaphragm stretching samples.

[0030] In a specific embodiment, the geometric parameters of the diaphragm stretching sample are embedded and encoded using an embedding encoder based on a fully connected layer to obtain a set of embedded encoding vectors of the geometric parameters of the diaphragm stretching sample, including: first, the geometric parameters of each diaphragm stretching sample need to be collected, and these parameters include shape, dimensional accuracy, thickness and parallelism. For numerical data, such as size and thickness, standardization or normalization is performed to ensure that data of different dimensions are on the same scale; for categorical data, it is converted into a numerical representation by means of unique hot encoding or the like. Then, all geometric parameters of each diaphragm stretching sample are combined into a feature vector. If some parameters are multidimensional, such as the shape can be described by multiple coordinate points, they are expanded as part of a single long vector. Next, a multi-layer perceptron (MLP) is constructed as the core part of the embedding encoder. This network consists of a series of fully connected layers, each of which applies linear transformations and nonlinear activation functions, such as ReLU. The last layer outputs a representation in a low-dimensional embedding space. Determining the appropriate number of hidden layers, the number of neurons in each layer, and the final embedding dimension size are key steps. The selection of these hyperparameters depends on the specific problem requirements and experimental results. If there is labeled data, such as known tensile strength values, supervised learning can be used to train the embedding encoder, with the goal of minimizing the error between the predicted value and the true value, while encouraging samples with similar geometric characteristics to be close to each other in the embedding space. In the absence of direct annotation, unsupervised or self-supervised methods can also be considered, using contrastive learning or other techniques to let the model learn to capture the intrinsic relationship between geometric parameters. After sufficient training, the original geometric parameters are input into the trained embedding encoder to obtain a low-dimensional embedding encoding vector corresponding to each diaphragm stretching sample. These vectors should be able to express the differences and commonalities between each sample more compactly while maintaining the original information. Subsequently, the embedding encoding vectors corresponding to all diaphragm stretching samples are collected to form a set containing all sample embedding representations in the entire data set, that is, the set of embedding encoding vectors of the geometric parameters of the diaphragm stretching samples.

[0031] Exemplarily, in step S32, the material tensile strength values ​​of the respective diaphragm stretching specimens are one-hot encoded to obtain a set of one-hot encoded vectors of the tensile strength of the diaphragm stretching specimens. It should be understood that, considering that in the process of data encoding the material tensile strength values ​​of the respective diaphragm stretching specimens, if the original tensile strength values ​​are directly used as the input of the model, it may mislead the model into thinking that there is a certain sequential association or distance measurement meaning between these values, while in fact, each material tensile strength value is only an identifier of a different category or level. Therefore, in order to avoid the potential misleading of the model by the size of the value, the present application uses the one-hot encoding technology to process the material tensile strength values ​​of the respective diaphragm stretching specimens separately, so as to convert each tensile strength value into an independent binary vector, in which only one element is 1 and the rest are 0, thereby obtaining a set of one-hot encoded vectors of the tensile strength of the diaphragm stretching specimens. In this way, each tensile strength value can be uniquely identified in a way that is independent of its numerical size, thereby ensuring that the model is not misleadingly weighted by numerical size when processing data, ensuring equal status between various strength levels, and ensuring that the model focuses on learning the correlation characteristics between geometric parameters and tensile strength.

[0032] Exemplarily, in step S33, each set of corresponding diaphragm stretching sample geometric parameter embedding encoding vectors and diaphragm stretching sample tensile strength one-hot encoding vectors in the set of the diaphragm stretching sample geometric parameter embedding encoding vectors and the diaphragm stretching sample tensile strength one-hot encoding vectors are spliced ​​to obtain the set of diaphragm stretching sample geometric parameter-test result splicing encoding vectors. It should be understood that by splicing each set of corresponding diaphragm stretching sample geometric parameter embedding encoding vectors and diaphragm stretching sample tensile strength one-hot encoding vectors, the geometric characteristics of the diaphragm and the material performance characteristics are combined, and the association representation between the geometric parameters and tensile strength of the diaphragm stretching sample is constructed to form a set of diaphragm stretching sample geometric parameter-test result splicing encoding vectors, thereby allowing the model to learn the influence pattern of the diaphragm geometric structure on the tensile performance.

[0033] Exemplarily, in step S4, the set of the diaphragm stretching sample geometric parameter-test result splicing coding vectors is subjected to feature aggregation based on feature dynamics analysis to obtain the diaphragm stretching sample geometric parameter-test result significant aggregation coding vector. It should be understood that in order to consider the information of all diaphragm stretching samples in a global scope to learn the general law between diaphragm geometric parameters and tensile strength, the present application further performs feature aggregation processing based on feature dynamics analysis on the set of the diaphragm stretching sample geometric parameter-test result splicing coding vectors to extract the representative correlation pattern between diaphragm geometric parameters and tensile strength, revealing the material performance law under a wider range of conditions.

[0034] In one embodiment, Figure 5 As shown, the set of diaphragm stretching sample geometric parameter-test result splicing coding vectors is subjected to feature aggregation based on feature dynamics analysis to obtain diaphragm stretching sample geometric parameter-test result significant aggregation coding vectors, including: S41, each diaphragm stretching sample geometric parameter-test result splicing coding vector in the set of diaphragm stretching sample geometric parameter-test result splicing coding vectors is input into the characteristic static potential energy measurement network respectively to obtain a set of diaphragm stretching sample geometric parameter-test result static potential energy measurement coefficients; S42, based on the set of diaphragm stretching sample geometric parameter-test result static potential energy measurement coefficients, calculate the diaphragm The diaphragm stretching sample geometric parameters - test results pseudo-anchor aggregation center representation vector; S43, based on the aggregation movement direction of each diaphragm stretching sample geometric parameters - test results splicing coding vector in the set of diaphragm stretching sample geometric parameters - test results splicing coding vectors relative to the diaphragm stretching sample geometric parameters - test results pseudo-anchor aggregation center representation vector, the set of diaphragm stretching sample geometric parameters - test results splicing coding vectors is dynamically aggregated toward the diaphragm stretching sample geometric parameters - test results pseudo-anchor aggregation center representation vector to obtain the diaphragm stretching sample geometric parameters - test results significantly aggregated coding vector.

[0035] In one embodiment, in step S41, each diaphragm stretching sample geometric parameter-test result splicing coding vector in the set of the diaphragm stretching sample geometric parameter-test result splicing coding vector is respectively input into the characteristic static potential energy measurement network to obtain a set of diaphragm stretching sample geometric parameter-test result static potential energy measurement coefficients, including: performing Z-score normalization on the diaphragm stretching sample geometric parameter-test result splicing coding vector to obtain a standardized diaphragm stretching sample geometric parameter-test result splicing coding vector; calculating the sum of the cube of each eigenvalue in the standardized diaphragm stretching sample geometric parameter-test result splicing coding vector divided by the characteristic scale value of the standardized diaphragm stretching sample geometric parameter-test result splicing coding vector to obtain the diaphragm stretching sample geometric parameter-test result static potential energy measurement coefficient. Specifically, the process can be expressed by the formula:

[0036]

[0037]

[0038] in, represents the set of geometric parameters-test result concatenated coding vectors of the diaphragm tensile specimen, , , and They represent the first, second, and third vectors in the set of geometric parameters-test results of the diaphragm tensile specimen. and The geometric parameters of the membrane tensile specimens and the test results splicing coding vector, is the number of characteristic vectors in the set of the diaphragm stretching sample geometric parameter-test result splicing coding vector, Indicates the The geometric parameters of the membrane tensile specimen-test results splicing coding vector The eigenvalues ​​at the positions, and Respectively represent the The characteristic mean and characteristic variance of the concatenated coding vector of the geometric parameters and test results of the membrane tensile specimens, is the characteristic scale value of the geometric parameter-test result splicing coding vector of the diaphragm stretching sample, For the said The static potential energy metric coefficient of the geometric parameters of the membrane tensile specimen-test results splicing encoding vector.

[0039] It should be understood that by implementing Z-score normalization on the concatenated encoding vector composed of the original geometric parameters and material tensile strength values, all features are converted to a common standard range, that is, a distribution with a mean of 0 and a standard deviation of 1. This ensures that the weights of features of different dimensions and scales in subsequent analysis are relatively equal, and avoids some features with a large numerical range dominating the analysis results. For the membrane tensile specimen, its geometric parameters (such as thickness, dimensional accuracy, etc.) and material tensile strength values ​​may be in different orders of magnitude. Direct use of the original data may cause the model to favor those features with larger values. Therefore, Z-score normalization enables these features to be evaluated on a fair basis, improving the accuracy of subsequent analysis. Next, the sum of the cubes of each eigenvalue in each standardized concatenated encoding vector is calculated and divided by the characteristic scale value of the vector (usually the standard deviation or other indicators to measure the degree of data dispersion). This operation actually constructs a special metric to quantify the static potential energy level of each sample point in its feature space. The cube is chosen instead of a simple square or other power because it not only retains the relative size relationship between the original features, but also enhances the sensitivity to extreme values. In the fatigue life prediction scenario, this means that those geometric parameter combinations or tensile strength performances that show significant differences will be given higher weights, reflecting that they may have a more important impact in practical applications. At the same time, normalization by dividing by the characteristic scale value ensures that different samples can be directly compared, and a relatively stable measurement coefficient is obtained, which is not affected by the absolute value of the original data.

[0040] That is, by performing characteristic static potential energy measurement on each diaphragm stretching specimen geometric parameter-test result splicing coding vector respectively, its potential energy level under static conditions and its characteristic importance and influence under the global characteristic distribution are revealed, thereby obtaining a set of static potential energy measurement coefficients of diaphragm stretching specimen geometric parameter-test result.

[0041] In one embodiment, in step S42, based on the set of static potential energy measurement coefficients of the diaphragm stretching sample geometric parameters-test results, the pseudo-anchored aggregation center representation vector of the diaphragm stretching sample geometric parameters-test results is calculated, including: performing characteristic energy level gating screening on the set of static potential energy measurement coefficients of the diaphragm stretching sample geometric parameters-test results to obtain a set of static potential energy weight factors of the diaphragm stretching sample geometric parameters-test results; using the set of static potential energy weight factors of the diaphragm stretching sample geometric parameters-test results as the weight distribution, calculating the weighted sum of the set of spliced ​​coding vectors of the diaphragm stretching sample geometric parameters-test results to obtain the pseudo-anchored aggregation center representation vector of the diaphragm stretching sample geometric parameters-test results. Specifically, the process can be expressed by the formula:

[0042]

[0043]

[0044]

[0045] in, Indicates Normalized diaphragm tensile specimen geometric parameters - test results static potential energy measurement coefficient, represents the gated mask function, represents the gating threshold, Indicates Geometric parameters of the membrane tensile specimen - static potential energy weighting factor of the test result, The vector representing the geometric parameters of the membrane tensile specimen and the pseudo-anchor aggregation center of the test results.

[0046] It should be understood that, firstly, through the feature level gating screening, those feature representations with significant influence in the global feature space can be selected from a large number of static potential energy metric coefficients. This process is similar to a filter, which determines which features should be retained and given higher weights according to the set threshold value or rule, and which features are considered unimportant or noise, and are therefore suppressed or ignored. For the diaphragm tensile specimens, this means that the key geometric parameter combinations that truly affect the tensile strength and fatigue properties of the material can be highlighted, rather than just relying on the surface trends in the data. This screening mechanism ensures that the subsequent analysis is more focused on meaningful information and improves the accuracy and reliability of the prediction model. Then, the screened static potential energy weight factor set is used as a weight distribution to calculate the weighted sum of the diaphragm tensile specimen geometric parameter-test result splicing encoding vector set. The core of this step is to generate a "pseudo-anchored aggregate center representation vector" that can represent the overall sample characteristics. This vector does not simply take the average value of all sample features, but comprehensively considers the importance of each sample and its position in the global feature space. Specifically, samples with higher weight factors will occupy a larger proportion in the final aggregation center representation vector, making this vector closer to the typical situation in actual engineering applications. This method can not only capture the tensile strength performance mode under different geometric parameter combinations, but also reflect the response characteristics of the material under different working conditions to a certain extent.

[0047] That is, based on the gating mechanism, the obtained set of static potential energy metric coefficients of the diaphragm stretching sample geometric parameters-test results is gated and masked to screen out the feature representations with significant influence in the global feature space and assign them corresponding weight factors, while suppressing the feature representations with weaker influence to reduce their interference with the model prediction results. Then, based on the obtained weight factors, the set of the diaphragm stretching sample geometric parameters-test results splicing encoding vectors is weighted aggregated to reveal the characteristic distribution trend center of the correlation pattern between the diaphragm stretching sample geometric parameters and tensile strength, and generate the diaphragm stretching sample geometric parameters-test results pseudo-anchor aggregation center representation vector, so as to reflect the overall influence pattern of different geometric parameters on the diaphragm tensile strength performance, thereby providing effective guidance for the subsequent feature aggregation process.

[0048] In one embodiment, in step S43, based on the aggregation movement direction of each diaphragm stretching sample geometric parameter-test result splicing coding vector in the set of the diaphragm stretching sample geometric parameter-test result splicing coding vector relative to the diaphragm stretching sample geometric parameter-test result pseudo-anchor aggregation center representation vector, the set of the diaphragm stretching sample geometric parameter-test result splicing coding vectors is dynamically aggregated toward the diaphragm stretching sample geometric parameter-test result pseudo-anchor aggregation center representation vector to obtain the diaphragm stretching sample geometric parameter-test result significant aggregation coding vector, including: calculating the diaphragm stretching sample geometric parameter-test result splicing coding vector The arccosine function value between each diaphragm stretching sample geometric parameter-test result splicing coding vector in the set of code vectors and the diaphragm stretching sample geometric parameter-test result pseudo-anchor aggregation center representation vector is used as the aggregation movement direction to obtain the set of diaphragm stretching sample geometric parameter-test result aggregation movement directions; based on the set of diaphragm stretching sample geometric parameter-test result aggregation movement directions, the set of diaphragm stretching sample geometric parameter-test result splicing coding vectors is dynamically aggregated toward the diaphragm stretching sample geometric parameter-test result pseudo-anchor aggregation center representation vector to obtain the diaphragm stretching sample geometric parameter-test result significant aggregation coding vector. Specifically, the process can be expressed by the formula:

[0049]

[0050]

[0051] in represents the magnitude of a vector, is the arccosine function, Indicates the The geometric parameters of the membrane stretching sample-test results splicing encoding vector is relative to the aggregation movement direction of the pseudo-anchor aggregation center representation vector of the geometric parameters of the membrane stretching sample-test results, represents the sine function, and denote the weight matrix and bias term respectively, The coding vector representing the significant aggregation of the geometric parameters and test results of the diaphragm tensile specimen.

[0052] That is, by calculating the aggregate movement direction of each diaphragm stretching sample geometric parameter-test result splicing coding vector relative to the diaphragm stretching sample geometric parameter-test result pseudo-anchor aggregation center representation vector, the change trend of the individual characteristics of each diaphragm stretching sample relative to the center point is quantified to reveal the relative relationship and group behavior pattern between the individual characteristics of each diaphragm stretching sample. Finally, based on the aggregate movement direction and characteristic static potential energy measurement coefficient of each diaphragm stretching sample geometric parameter-test result splicing coding vector, each diaphragm stretching sample geometric parameter-test result splicing coding vector is dynamically aggregated toward the diaphragm stretching sample geometric parameter-test result pseudo-anchor aggregation center representation vector to obtain the diaphragm stretching sample geometric parameter-test result significant aggregation coding vector, thereby achieving an in-depth understanding of the correlation pattern between the diaphragm stretching sample geometric parameter and tensile strength. Through this dynamic polymerization process, it is possible to capture the subtle differences in the tensile strength performance of different diaphragm tensile specimens under changes in the geometric parameters, and extract representative feature representations from a large amount of sample data to summarize the effects of different geometric parameter combinations on the tensile strength of the material, thereby providing a more accurate basis for subsequent fatigue life predictions.

[0053] Exemplarily, in step S5, the σb value is determined based on the diaphragm stretching specimen geometric parameter-test result significant aggregation coding vector. In one embodiment, the σb value is determined based on the diaphragm stretching specimen geometric parameter-test result significant aggregation coding vector, including: inputting the diaphragm stretching specimen geometric parameter-test result significant aggregation coding vector into a decoder-based σb value estimation module to obtain the σb value. It should be understood that the diaphragm stretching specimen geometric parameter-test result significant aggregation coding vector integrates the geometric parameters and tensile strength test results of each diaphragm stretching specimen, and after characteristic dynamics analysis and global aggregation processing, it can more accurately reflect the material performance laws under different geometric configurations. Through further processing by the decoder, these complex multi-dimensional information can be mapped back to a specific physical quantity-σb value, thereby providing directly available data support for engineering applications.

[0054] The role of the decoder here is to convert the high-dimensional, abstract diaphragm tensile specimen geometric parameters-test results significant aggregate coding vector into a specific value that can be directly used for engineering decision-making. This is because although the original aggregate coding vectors contain rich information, they exist in a high-dimensional form and are not easy to directly interpret or apply to actual engineering scenarios. The decoder can simplify these complex representations into a numerical value that is easy to understand and use, while improving its generalization ability by learning patterns in the training data. Even when faced with new samples that have never been seen, it can make reasonable estimates based on the knowledge it has learned. In addition, the decoder can adjust its internal structure and parameters according to the needs of a specific task to achieve the best prediction effect, thereby optimizing the prediction accuracy.

[0055] In an embodiment of the present application, the decoder architecture includes an input layer that receives data from a diaphragm stretching sample geometric parameter-test result significant aggregation coding vector, and a hidden layer that includes several fully connected layers, each of which applies a linear transformation and a nonlinear activation function (such as ReLU). The output layer has only one neuron, which is used to output the predicted σb value, and a linear activation function is usually used because the σb value is a continuous variable. The loss function selects the mean square error (MSE) as a criterion for measuring the difference between the predicted σb value and the true value. Minimizing this error during training allows the decoder to restore the σb value as accurately as possible. After initializing the weight parameters of each layer of the decoder, the training data is input into the decoder to calculate the predicted σb value. The gradient is calculated based on the difference between the predicted value and the true value, and the network weights are updated using an optimization algorithm (such as Adam) to gradually reduce the error. Once the decoder training is completed, it can be used on a new diaphragm stretching sample geometric parameter-test result significant aggregation coding vector to quickly and accurately predict the corresponding σb value.

[0056] Preferably, the diaphragm stretching sample geometric parameter-test result significant aggregation coding vector is input into a decoder-based σb value estimation module to obtain the σb value, including

[0057] Arrange the eigenvalues ​​of the diaphragm stretching sample geometric parameter-test result significant aggregation coding vector in ascending order to form a diaphragm stretching sample geometric parameter-test result significant aggregation sequence coding vector;

[0058] In response to the diaphragm stretch sample geometric parameters - test results significantly aggregate the order encoding vector The eigenvalue and The absolute value of the difference between the eigenvalues ​​is less than or equal to the distance difference hyperparameter , calculate the The eigenvalues ​​are similar to the The weighted sum between the eigenvalues ​​is the optimized Eigenvalue;

[0059] Calculate the square root of the sum of squares of all eigenvalues ​​of the diaphragm tensile specimen geometric parameter-test result significant aggregation encoding vector:

[0060]

[0061] in, The first one represents the significant aggregation sequence encoding vector of the geometric parameters and test results of the diaphragm tensile specimen. Eigenvalues, represents the length of the diaphragm tensile specimen geometric parameter-test result significant aggregation encoding vector, The square root of the sum of squares of all eigenvalues ​​of the diaphragm tensile specimen geometric parameter-test result significant aggregation coding vector;

[0062] The square root of the sum of the squares of all eigenvalues ​​of the diaphragm stretching sample geometric parameter-test result significant aggregation coding vector is multiplied by 2 and then divided by the square of the length of the diaphragm stretching sample geometric parameter-test result significant aggregation coding vector to obtain the diaphragm stretching sample geometric parameter-test result significant aggregation space primitive value ;

[0063] In response to the diaphragm stretch sample geometric parameters - test results significantly aggregate the order encoding vector The eigenvalue and The absolute value of the difference between the eigenvalues ​​is greater than the distance difference hyperparameter , multiply the geometric parameter of the membrane tensile specimen-the significant aggregation space primitive value of the test result by the first After the eigenvalue, calculate the product with the first The weighted reduction between the eigenvalues ​​is the optimized Eigenvalue;

[0064] On the basis of keeping the first eigenvalue of the significant aggregation sequence encoding vector of the diaphragm tensile specimen geometric parameters-test results unchanged, the combined optimization of the Eigenvalues ​​are used to obtain the optimized geometric parameters of the membrane tensile specimen - the test results are significantly aggregated encoding vectors;

[0065] The optimized diaphragm stretching specimen geometric parameters-test results significant aggregation coding vector is input into the decoder-based σb value estimation module to obtain the σb value.

[0066] Taking into account that each diaphragm stretching sample geometric parameter-test result splicing coding vector in the set of the diaphragm stretching sample geometric parameter-test result splicing coding vectors respectively represents the low-dimensional embedded coding splicing features of the diaphragm stretching sample geometric parameter and the test result, when performing feature aggregation based on feature dynamics analysis, the insufficient feature dynamic correspondence of the low-dimensional semantic splicing features of each local diaphragm stretching sample geometric parameter and the test result will also lead to insufficient long-distance feature aggregation representation of the diaphragm stretching sample geometric parameter-test result significant aggregation coding vector, thereby affecting the accuracy of the σb value obtained by its input based on the decoder's σb value estimation module.

[0067] Therefore, in order to solve the problem that the feature set of the diaphragm stretching sample geometric parameter-test result significant aggregation coding vector is insufficient in expressing the global interactive response due to the long distance exceeding the predetermined local distribution interval threshold under the predetermined eigenvalue sequence distribution, the high-dimensional feature space primitive representation of the diaphragm stretching sample geometric parameter-test result significant aggregation coding vector based on self-inner product fusion is used to capture the complex structure of the global network interaction of its eigenvalues, so as to reconstruct the interactive response relationship between the eigenvalues ​​of the diaphragm stretching sample geometric parameter-test result significant aggregation coding vector by simulating the scale-based high-dimensional feature space potential primitive, so as to realize the coding reconstruction of the real sequence distribution behavior of the diaphragm stretching sample geometric parameter-test result significant aggregation coding vector under the long distance, and improve the interactive response expression effect of the diaphragm stretching sample geometric parameter-test result significant aggregation coding vector. In this way, the accuracy of the σb value obtained by the σb value estimation module based on the decoder of the diaphragm stretching sample geometric parameter-test result significant aggregation coding vector input is improved.

[0068] In summary, the test method for the fatigue life of the diaphragm of the diaphragm compressor according to the embodiment of the present application is explained, which measures the material tensile strength value of each diaphragm tensile specimen through a tensile fatigue test, and at the same time combines the geometric parameters of each diaphragm tensile specimen to construct an association representation between the geometric parameters of the diaphragm tensile specimen and the tensile test results, and then, by performing a global aggregation analysis on each group of constructed association representations, the σb value of the diaphragm compressor diaphragm material is intelligently estimated. In this way, the tensile performance of the diaphragm material can be more accurately evaluated, thereby providing a more reliable basis for the diaphragm fatigue life test.

[0069] The prefixes such as "first" and "second" used in the embodiments of the present application are only used to distinguish different description objects, and have no limiting effect on the position, order, priority, quantity or content of the described objects. The use of prefixes such as ordinal numbers used to distinguish description objects in the embodiments of the present application does not constitute a limitation on the described objects. For the statement of the described objects, please refer to the description in the context of the claims or embodiments, and the use of such prefixes should not constitute an unnecessary limitation.

[0070] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0071] In the various embodiments of the present application, unless otherwise specified or logically conflicting, the terms and / or descriptions between the various embodiments are consistent and can be referenced to each other, and the technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationships.

[0072] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0073] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0074] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for testing the fatigue life of a diaphragm compressor diaphragm, comprising: The diaphragm tensile specimens were grouped and subjected to tensile fatigue tests to calculate the σb value; Determine the fatigue test stress level based on actual typical working conditions; Make standard parts for fatigue testing; Determine the fatigue test grouping; Fitting the SN curve of the diaphragm material according to the fatigue test data; inputting the SN curve of the diaphragm material into CAE as the boundary condition; inputting the actual three-dimensional model of the diaphragm into CAE; changing the diaphragm constraint, temperature, and historical load conditions, solving through the solver of the fatigue numerical analysis software, and obtaining the life curve of the sample; issuing a fatigue analysis report containing life prediction data, characterized in that the diaphragm tensile samples are grouped, and the tensile fatigue test is performed to calculate the σb value, including: Obtaining geometric parameters of each diaphragm stretching sample, wherein the geometric parameters include geometric shape, dimensional accuracy, thickness and parallelism; Obtaining the material tensile strength value of each diaphragm tensile sample through a tensile fatigue test; Based on the geometric parameters and material tensile strength values ​​of each diaphragm tensile specimen, a set of diaphragm tensile specimen geometric parameter-test result splicing coding vectors is constructed; Performing feature aggregation based on feature dynamics analysis on the set of the diaphragm stretching sample geometric parameter-test result splicing coding vectors to obtain a diaphragm stretching sample geometric parameter-test result significant aggregation coding vector; Determining the σb value based on the diaphragm stretching sample geometric parameter-test result significant aggregation coding vector; Based on the geometric parameters and material tensile strength values ​​of each diaphragm tensile specimen, a set of diaphragm tensile specimen geometric parameter-test result splicing coding vectors is constructed, including: Embedding and coding the geometric parameters of each diaphragm stretching sample to obtain a set of embedded coding vectors of the geometric parameters of the diaphragm stretching sample; One-hot encoding is performed on the material tensile strength values ​​of each of the diaphragm stretching specimens to obtain a set of one-hot encoding vectors of the tensile strength of the diaphragm stretching specimens; Each corresponding group of diaphragm stretching sample geometric parameter embedding coding vectors and diaphragm stretching sample tensile strength one-hot coding vectors in the set of diaphragm stretching sample geometric parameter embedding coding vectors and the set of diaphragm stretching sample tensile strength one-hot coding vectors are spliced ​​to obtain the set of diaphragm stretching sample geometric parameter-test result splicing coding vectors.

2. The method for testing the fatigue life of a diaphragm compressor according to claim 1, characterized in that: Embedding and encoding the geometric parameters of each diaphragm stretching sample to obtain a set of diaphragm stretching sample geometric parameter embedded encoding vectors, including: The geometric parameters of each diaphragm stretching sample are embedded and encoded using an embedding encoder based on a fully connected layer to obtain a set of embedded encoding vectors of the geometric parameters of the diaphragm stretching sample.

3. The method for testing fatigue life of a diaphragm compressor according to claim 2, characterized in that: The set of the diaphragm stretching sample geometric parameter-test result splicing coding vectors is subjected to feature aggregation based on feature dynamics analysis to obtain a diaphragm stretching sample geometric parameter-test result significant aggregation coding vector, including: Inputting each diaphragm stretching sample geometric parameter-test result splicing encoding vector in the set of diaphragm stretching sample geometric parameter-test result splicing encoding vectors into the characteristic static potential energy measurement network to obtain a set of diaphragm stretching sample geometric parameter-test result static potential energy measurement coefficients; Based on the set of static potential energy metric coefficients of the diaphragm stretching sample geometric parameters-test results, a pseudo-anchored aggregation center representation vector of the diaphragm stretching sample geometric parameters-test results is calculated; Based on the aggregation movement direction of each diaphragm stretching specimen geometric parameter-test result splicing coding vector in the set of diaphragm stretching specimen geometric parameter-test result splicing coding vectors relative to the diaphragm stretching specimen geometric parameter-test result pseudo-anchor aggregation center representation vector, the set of diaphragm stretching specimen geometric parameter-test result splicing coding vectors is dynamically aggregated toward the diaphragm stretching specimen geometric parameter-test result pseudo-anchor aggregation center representation vector to obtain the diaphragm stretching specimen geometric parameter-test result significantly aggregated coding vector.

4. The method for testing fatigue life of a diaphragm compressor according to claim 3, characterized in that: Inputting each diaphragm stretching sample geometric parameter-test result splicing encoding vector in the set of diaphragm stretching sample geometric parameter-test result splicing encoding vectors into the characteristic static potential energy measurement network to obtain a set of diaphragm stretching sample geometric parameter-test result static potential energy measurement coefficients, including: Performing Z-score normalization processing on the diaphragm stretching sample geometric parameter-test result splicing coding vector to obtain a standardized diaphragm stretching sample geometric parameter-test result splicing coding vector; The sum of the cube of each eigenvalue in the standardized diaphragm stretching sample geometric parameter-test result splicing coding vector is calculated and divided by the characteristic scale value of the standardized diaphragm stretching sample geometric parameter-test result splicing coding vector to obtain the diaphragm stretching sample geometric parameter-test result static potential energy measurement coefficient.

5. The method for testing the fatigue life of a diaphragm compressor according to claim 4, characterized in that: Based on the set of static potential energy metric coefficients of the diaphragm stretching sample geometric parameters-test results, the pseudo-anchored aggregation center representation vector of the diaphragm stretching sample geometric parameters-test results is calculated, including: Performing characteristic energy level gating screening on the set of diaphragm stretching sample geometric parameters-test result static potential energy measurement coefficients to obtain a set of diaphragm stretching sample geometric parameters-test result static potential energy weight factors; Taking the set of static potential energy weight factors of the diaphragm stretching specimen geometric parameters-test results as the weight distribution, the weighted sum of the set of spliced ​​coding vectors of the diaphragm stretching specimen geometric parameters-test results is calculated to obtain the pseudo-anchored aggregation center representation vector of the diaphragm stretching specimen geometric parameters-test results.

6. The method for testing the fatigue life of a diaphragm compressor according to claim 5, characterized in that: Based on the aggregation movement direction of each diaphragm stretching sample geometric parameter-test result splicing coding vector in the set of the diaphragm stretching sample geometric parameter-test result splicing coding vector relative to the diaphragm stretching sample geometric parameter-test result pseudo-anchor aggregation center representation vector, the set of the diaphragm stretching sample geometric parameter-test result splicing coding vectors is dynamically aggregated toward the diaphragm stretching sample geometric parameter-test result pseudo-anchor aggregation center representation vector to obtain the diaphragm stretching sample geometric parameter-test result significant aggregation coding vector, including: Calculate the arccosine function value between each diaphragm stretching sample geometric parameter-test result splicing coding vector in the set of the diaphragm stretching sample geometric parameter-test result splicing coding vector and the diaphragm stretching sample geometric parameter-test result pseudo-anchor aggregation center representation vector as the aggregation movement direction to obtain the set of diaphragm stretching sample geometric parameter-test result aggregation movement directions; Based on the set of aggregation movement directions of the diaphragm stretching specimen geometric parameters-test results, the set of the diaphragm stretching specimen geometric parameters-test results splicing coding vectors is dynamically aggregated toward the diaphragm stretching specimen geometric parameters-test results pseudo-anchor aggregation center representation vector to obtain the diaphragm stretching specimen geometric parameters-test results significant aggregation coding vector.

7. The method for testing fatigue life of a diaphragm compressor according to claim 6, characterized in that: Based on the diaphragm tensile specimen geometric parameter-test result significant aggregation coding vector, the σb value is determined, including: The diaphragm stretching sample geometric parameter-test result significant aggregation coding vector is input into the decoder-based σb value estimation module to obtain the σb value.

Images