Gear pump valve element stress relaxation prediction method and performance degradation model data expansion method

Abstract

The invention relates to a gear pump valve core stress relaxation prediction method and a performance degradation model data expansion method. The method comprises the following steps: S1, selecting a test piece; s2, mounting the test piece on a loading tool, and starting a stress test; s3, collecting stress values of all time points; s4, calculating a load loss rate; s5, fitting a logarithmic regression model of the test piece by adopting a least square method; s6, determining parameter values in the regression model; s7, calculating a stress prediction value; s8, calculating an error between the stress prediction value and the actual stress value; s9, repeating the steps S2 to S8, and determining an optimal regression model of the test piece; and S10, through the optimal regression model, carrying out stress value prediction on similar test pieces at any time point. According to the gear pump key component performance degradation prediction method, stress relaxation data are fitted more accurately by introducing the degradation model and the regression model, subsequent numerical values are expanded by adopting error analysis and model verification strategies of the system, and the gear pump key component performance degradation prediction accuracy is remarkably improved.

Description

Technical Field

[0001] The invention relates to the technical field of performance degradation analysis of key components of a gear pump, and in particular to a gear pump valve core stress relaxation prediction method and a performance degradation model data expansion method. Background Art

[0002] Gear pumps are widely used in industrial automation, hydraulic transmission and lubrication systems. Their reliability and service life directly affect the performance and stability of the entire system. As the core component of the gear pump, the life of the valve core is crucial to the overall quality of the pump. However, due to the complexity of the working environment and the variability of operating conditions, it is challenging to accurately predict the service life of the valve core. In order to understand the degradation process of gear pumps, traditional methods mainly rely on large samples and long-term physical tests. This method is not only time-consuming, but also needs to be improved in accuracy. In practical applications, changes in environmental and operating conditions may lead to problems such as reduced repeatability and reliability of experimental results. By constructing a prediction model, simulating the performance degradation of gear pumps under different operating conditions has become a trend in solving such problems when the gear pump is physically tested for a certain period of time. By using simulation data, the key factors affecting the life of the gear pump are identified, providing an important reference for product design and optimization.

[0003] Therefore, in order to improve the design quality of the gear pump and extend its service life, the present invention proposes a gear pump valve core stress relaxation prediction method and a performance degradation model data expansion method to solve the above problems. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention provides a method for predicting stress relaxation of a gear pump valve core. By using a logarithmic regression model of load loss rate and relaxation time, stress relaxation test data can be fit more accurately, and the data can be used in data expansion of a performance degradation model of key components of a gear pump. Compared with traditional linear models, this method performs better in capturing nonlinear relaxation behavior and significantly improves fitting accuracy. At the same time, a systematic error analysis and model verification strategy are adopted, so that the model can maintain a high prediction accuracy under different experimental conditions.

[0005] The technical solution adopted by the present invention is a method for predicting stress relaxation of a gear pump valve core, which comprises the following steps:

[0006] S1. Select a plurality of identical gear pump valve core test pieces to be tested, and divide them into a first group of gear pump valve core test pieces, a second group of gear pump valve core test pieces, and an Nth group of gear pump valve core test pieces;

[0007] S2, installing the selected gear pump valve core specimens on the loading fixture in the stress relaxation testing machine, starting the loading fixture, and starting the stress test of the first group of gear pump valve core specimens;

[0008] S3, using a data acquisition and processing module in the stress relaxation testing machine, real-time acquisition of stress values ​​of the gear pump valve core specimen measured at each time point during the loading time, and dividing the data measured during the loading time into a training phase data set and a test phase data set;

[0009] S4. According to the training phase data set recorded in step S3, the load loss rate of the gear pump valve core specimen at each time point in the training phase is calculated, and the expression is:

[0010]

[0011] Where P0 represents the initial stress value of the gear pump valve core specimen, P t represents the stress value at time t;

[0012] S5. The least square method is used to fit the logarithmic regression model of the load loss rate and relaxation time of the gear pump valve core specimen, and the expression is:

[0013]

[0014] In the formula, represents the load loss rate, t R represents the relaxation time, A and B represent the parameters of the regression model;

[0015] S6, according to the regression model in step S5, determining the parameter values ​​of A and B in the regression model of the first group of gear pump valve core specimens;

[0016] S7. Calculate the stress prediction value of the first group of gear pump valve core specimens during the test phase according to the fitted regression model;

[0017] S8, calculating the error between the stress prediction value of the first group of gear pump valve core specimens during the test phase and the actual stress value of the test phase recorded in step S3, and obtaining the root mean square difference to verify the fitting quality of the regression model;

[0018] S9, repeating steps S2 to S8, carrying out stress tests on the second group of gear pump valve core specimens and the Nth group of gear pump valve core specimens in turn, and selecting a group of gear pump valve core specimen regression models with the smallest root mean square difference as the optimal regression model;

[0019] S10. Through the optimal regression model, the stress value of the same type of gear pump valve core specimen at any time point can be predicted, so that the stress relaxation amount of the gear pump valve core specimen can be predicted.

[0020] Furthermore, in step S2, the selected gear pump valve core specimen is installed on a loading fixture in the stress relaxation testing machine, the loading fixture is started, and the first group of gear pump valve core specimen stress tests are started. The specific steps include:

[0021] S21, installing the first set of gear pump valve core test pieces on the loading fixture, and setting the initial test parameters;

[0022] S22, raising the temperature of the incubator to a predetermined test temperature and performing heat preservation;

[0023] S23, starting the loading push rod to apply a loading force to the first group of gear pump valve core test pieces so that the gear pump valve core test pieces maintain a certain elastic displacement;

[0024] S24, setting the loading temperature and loading time of the gear pump valve core test piece;

[0025] S25, recording in real time the elastic force generated by the gear pump valve core test piece at each time point through a force sensor;

[0026] S26. When the elastic force of the gear pump valve core specimen decays to the loading time, the stress test of the first group of gear pump valve core specimens is stopped.

[0027] Preferably, the displacement control accuracy of the loading tool in step S23 is greater than or equal to ±0.1%.

[0028] Furthermore, the step S3 divides the data measured during the loading time into a training phase data set and a test phase data set, and specifically includes the following steps:

[0029] S31, performing equidistant sampling on the stress value data of the gear pump valve core test piece measured at each time point during the loading time collected in real time;

[0030] S32, dividing the data after equidistant sampling into data sets;

[0031] S33, using 60% of the sampled data as a training phase data set, which can be used to fit the regression model of the gear pump valve core test piece;

[0032] S34. 40% of the sampled data are used as a test phase data set to test the error between the stress value predicted by the regression model and the actual stress value in the test phase.

[0033] Preferably, the sampling frequency of the data acquisition processing module in step S3 is greater than or equal to 1 Hz.

[0034] A second aspect of the present invention provides a method for expanding data of a gear pump valve core performance degradation model, which comprises the following steps:

[0035] S1. Select the gear pump valve core specimen to be tested;

[0036] S2. Install the selected gear pump valve core specimen on the loading fixture in the stress relaxation testing machine, start the loading fixture, and start the gear pump valve core specimen stress test;

[0037] S3, collecting the initial stress value of the gear pump valve core specimen and the stress values ​​measured at each time point in real time through the data acquisition and processing module in the stress relaxation tester, and storing the collected stress values ​​in the data acquisition and processing module;

[0038] S4, preprocessing the collected stress value data of the gear pump valve core test piece;

[0039] S5, using a variety of fitting methods to fit the stress value data preprocessed in step S4;

[0040] S6, calculating the error between the predicted value in each fitting method and the actual stress value in step S3, and verifying the fitting quality of each fitting method;

[0041] S7. According to the preset fitting model, the optimal data evaluation value expression of the gear pump valve core specimen stress value is determined as follows:

[0042]

[0043] In the formula, w1 represents the residual weight coefficient, w2 represents the root mean square error weight coefficient, and w3 represents the determination coefficient weight coefficient. Represents the average value of the absolute value of the residual, RMSE represents the root mean square error, R 2 represents the coefficient of determination;

[0044] S8. By determining the optimal fitting method, the stress value of the same type of gear pump valve core specimen at any time point is expanded, so as to predict the stress relaxation amount of the gear pump valve core specimen degradation model in advance.

[0045] Furthermore, in step S4, the collected stress value data of the gear pump valve core specimen is preprocessed, which specifically includes the following steps:

[0046] S41, the stress values ​​of the gear pump valve core specimen collected during the heating stage and the cooling stage of the insulation box during the removal test;

[0047] S42, using a wavelet transform method to perform denoising on the collected stress values ​​of the gear pump valve core specimen;

[0048] S43, normalizing the denoised stress value data to ensure consistency of subsequent processing, and the expression is:

[0049] X= X i -X min

[0050] X max -X min

[0051] Where, X i represents the stress value data after denoising, X min Represents the minimum value of the data set, X max Indicates the maximum value of the data set.

[0052] Preferably, the multiple fitting methods in step S5 include a linear regression fitting method, a polynomial regression fitting method and a nonlinear regression fitting method.

[0053] The characteristics and beneficial effects of the present invention are:

[0054] 1. The gear pump valve core stress relaxation prediction method provided by the present invention can more accurately fit the stress relaxation test data by using the logarithmic regression model of load loss rate and relaxation time, and adopts a systematic error analysis and model verification strategy, so that the model can maintain a high prediction accuracy under different experimental conditions. Compared with the traditional linear model, this method performs better in capturing nonlinear relaxation behavior and significantly improves the fitting accuracy.

[0055] 2. The gear pump valve core stress relaxation prediction method provided by the present invention has an integrated data processing and analysis process to reduce human errors and improve overall work efficiency. The automated data collection and calculation optimize the laboratory workflow, allowing researchers to obtain effective results more quickly, and solving the problem that traditional methods mainly rely on large samples and long-term physical experiments, resulting in long testing time and low accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 is a flow chart of a first data fitting method of the present invention;

[0057] Figure 2 This is a flow chart of the first gear pump valve core test piece stress test of the present invention;

[0058] Figure 3 It is a flow chart for verifying the fitting quality of the regression model in the first data fitting method of the present invention;

[0059] Figure 4 It is a structural schematic diagram of the loading tooling of the present invention;

[0060] Figure 5 is the original data collected in Example 1 of the present invention;

[0061] Figure 6is a logarithmic regression fitting curve of load loss rate and time in Example 1 of the present invention;

[0062] Figure 7 is the residual graph of embodiment 1 of the present invention;

[0063] Figure 8 is a schematic diagram of the training process of Embodiment 2 of the present invention;

[0064] Fig. 9 2 is a schematic diagram of the comparison between the training set prediction results and the real data of Example 2 of the present invention;

[0065] Fig.10 2 is a schematic diagram of the comparison between the prediction results of the prediction set and the real data of Example 2 of the present invention;

[0066] Fig.11 is a flow chart of a second data fitting method of the present invention;

[0067] Fig.12 is a schematic diagram of a second data fitting method of the present invention;

[0068] Fig.13 is the fitting curve when the linear regression method is fitted using half the sample size in Example 3 of the present invention;

[0069] Fig.14 is a residual graph when the linear regression method is fitted using half the sample size in Example 3 of the present invention;

[0070] Fig.15 is the fitting curve when the linear regression method is used to fit the whole sample in Example 3 of the present invention;

[0071] Fig.16 This is a residual graph when the linear regression method is fitted using the entire sample in Example 3 of the present invention;

[0072] Fig.17 is the fitting curve when the quadratic polynomial fitting method is used to fit the sample size using half the sample size in Example 3 of the present invention;

[0073] Fig.18 is a residual graph when the quadratic polynomial fitting method is used to fit the sample size using half the sample size in Example 3 of the present invention;

[0074] Fig.19 is the fitting curve when the quadratic polynomial fitting method is used to fit the whole sample in Example 3 of the present invention;

[0075] Fig. 20 is a residual graph when the quadratic polynomial fitting method is used to fit the entire sample in Example 3 of the present invention;

[0076] Fig.21is the fitting curve when the cubic polynomial fitting method is used to fit half the sample size in Example 3 of the present invention;

[0077] Fig. 22 is a residual graph when a cubic polynomial fitting method is used to fit the half sample size in Example 3 of the present invention;

[0078] Fig.23 is the fitting curve when the cubic polynomial fitting method is used to fit the whole sample in Example 3 of the present invention;

[0079] Fig.24 is a residual graph when the cubic polynomial fitting method is used to fit the entire sample in Example 3 of the present invention;

[0080] Fig.25 is the fitting curve when the sine function model fitting method is used to fit the half sample size in Example 3 of the present invention;

[0081] Fig.26 It is the fitting curve when the custom model fitting method is used to fit the half sample size in Example 3 of the present invention.

[0082] Main reference numerals:

[0083] Fixed bracket 1; fixed end 2; loading end 3; loading push rod 4; protective cover 5; force sensor 6; connecting adapter 7; gear pump valve core test piece 8; data acquisition and processing module 9; loading push rod control module 10. DETAILED DESCRIPTION

[0084] In order to fully describe the technical content, structural features, objectives and effects of the present invention, the following will be described in detail with reference to the accompanying drawings.

[0085] A method for predicting stress relaxation of a gear pump valve core of the present invention is as follows: Figure 1 As shown, it includes the following steps:

[0086] S1, select a plurality of identical gear pump valve core test pieces 8 to be tested, and divide them into a first group of gear pump valve core test pieces, a second group of gear pump valve core test pieces and an Nth group of gear pump valve core test pieces. In this embodiment, the Nth group of gear pump valve core test pieces is a third group of gear pump valve core test pieces;

[0087] S2, install the selected gear pump valve core specimen 8 on the loading fixture in the stress relaxation testing machine, start the loading fixture, and start the first group of gear pump valve core specimen 8 stress test, such as Figure 2 As shown, the specific steps are as follows:

[0088] S21, installing the first set of gear pump valve core test pieces on the loading fixture, and setting the initial test parameters;

[0089] S22, raising the temperature of the incubator to a predetermined test temperature and performing heat preservation;

[0090] S23, starting the loading push rod to apply a loading force to the first group of gear pump valve core test pieces so that the gear pump valve core test pieces maintain a certain elastic displacement;

[0091] S24, setting the loading temperature and loading time of the gear pump valve core test piece;

[0092] S25, recording in real time the elastic force generated by the gear pump valve core test piece at each time point through a force sensor;

[0093] S26. When the elastic force of the gear pump valve core specimen decays to the loading time, the stress test of the first group of gear pump valve core specimens is stopped.

[0094] S3, through the data acquisition and processing module 9 in the stress relaxation tester, the stress values ​​measured by the gear pump valve core test piece 8 at each time point during the loading time are collected in real time, and the collected stress values ​​are stored in the data acquisition and processing module 9, and the data measured during the loading time are divided into a training phase data set and a test phase data set, which specifically includes the following steps:

[0095] S31, performing equidistant sampling on the stress value data of the gear pump valve core test piece measured at each time point during the loading time collected in real time;

[0096] S32, dividing the data after equidistant sampling into data sets;

[0097] S33, using 60% of the sampled data as a training phase data set, which can be used to fit the regression model of the gear pump valve core test piece;

[0098] S34. 40% of the sampled data are used as a test phase data set to test the error between the stress value predicted by the regression model and the actual stress value in the test phase.

[0099] S4. According to the training phase data set recorded in step S3, the load loss rate of the gear pump valve core test piece 8 at each time point in the training phase is calculated, and the expression is:

[0100]

[0101] Where P0 represents the initial stress value of the gear pump valve core specimen, P t represents the stress value at time t;

[0102] S5. The least square method is used to fit the logarithmic regression model of the load loss rate and relaxation time of the gear pump valve core specimen 8, and its expression is:

[0103]

[0104] In the formula, represents the load loss rate, t R represents the relaxation time, A and B represent the parameters of the regression model;

[0105] S6, according to the regression model in step S5, determining the parameter values ​​of A and B in the regression model of the first group of gear pump valve core test pieces 8;

[0106] S7, calculating the stress prediction value of the first group of gear pump valve core specimens 8 during the test phase according to the fitted regression model;

[0107] S8, calculate the error between the stress prediction value of the first group of gear pump valve core specimens 8 during the test phase and the actual stress value recorded in step S3 during the test phase, and obtain the root mean square error to verify the fitting quality of the regression model, such as Figure 3 As shown, the verification is carried out through the following steps:

[0108] S81. Calculate the root mean square error between the predicted value of the regression model and the actual stress value, and its expression is:

[0109]

[0110] Where n is the total number of samples. It represents the number of samples in the dataset, that is, the number of prediction-true value pairs we consider when calculating the error. i represents the actual value. In regression problems, this is the true observed value or label of the (i)th sample. f(x i ) represents the predicted value. This is the prediction result given by the model for the (i)th sample. i -f(x i ) represents the square of the error. It represents the square of the prediction error of the (i)th sample, reflecting the degree of deviation between the predicted value and the true value. The square operation amplifies the larger error and ensures that the error is non-negative.

[0111] The smaller the RMSE, the closer the model's predicted value is to the actual value, and the better the model's prediction performance is. RMSE is particularly suitable for performance comparison between different models of the same or similar data sets.

[0112] S82. Calculate the coefficient of determination between the predicted value of the regression model and the actual stress value, and its expression is:

[0113]

[0114] In the formula, y i represents the actual stress value, represents the predicted value, represents the average value of the true observation, and n represents the number of samples;

[0115] S83. Calculate the residual between the predicted value of the regression model and the actual stress value. The expression is:

[0116]

[0117] In the formula, y i represents the actual stress value, represents the predicted value;

[0118] S84. Evaluate the data fitting quality of the gear pump valve core specimen based on the regression model through residual analysis;

[0119] S9, repeating steps S2 to S8, carrying out stress tests on the remaining gear pump valve core specimens, and selecting a set of gear pump valve core specimen regression models with the smallest root mean square difference as the optimal regression model;

[0120] S10. Through the optimal regression model, the stress value of the same type of gear pump valve core specimen at any time point can be predicted, so that the stress relaxation amount of the gear pump valve core specimen can be predicted in advance.

[0121] In a preferred embodiment, the gear pump valve core test piece 8 includes a tensile gear pump valve core and a compression gear pump valve core.

[0122] In a preferred embodiment, the predetermined test temperature in step S22 is -35°C to 175°C.

[0123] In a preferred embodiment, the displacement control accuracy of the loading tool in step S23 is greater than or equal to ±0.1%.

[0124] In a preferred embodiment, the sampling frequency of the data acquisition processing module in step S3 is greater than or equal to 1 Hz.

[0125] like Figure 4As shown, the loading tool is mainly used to load the compression force on the gear pump valve core test piece 8, so that the gear pump valve core test piece 8 is fixed for a long time under a certain displacement, and the longest fixing time can be up to one month. The loading tool includes a fixed bracket 1, a fixed end 2, a loading end 3 and a loading push rod 4. The loading tool is arranged in a protective cover 5, and an insulation box is arranged in the protective cover 5. A guide rail is arranged on the fixed bracket 1, and a fixed end 2 is arranged at the first end of the fixed bracket 1, and a force sensor 6 is arranged on the fixed end 2. The force sensor 6 is used to measure the tension value of the gear pump valve core, convert the tension into an electrical signal, and transmit it to the data acquisition and processing module 9, and finally obtain the tension value on the software for reliability measurement and evaluation. A connection adapter 7 is arranged on the force sensor 6, and the connection adapter 7 can be replaced according to the type of the product to be tested. And the loading push rod 4 is arranged at the second end of the fixed bracket 1, and the loading push rod control module 10 is controlled by the main control computer, and the control signal is passed through the driving circuit to control the loading push rod 4 to perform linear motion of loading and unloading. The change in the spacing between the loading end 3 and the fixed end 2 produces a stretching effect on the product to be tested, and finally achieves the effect of test loading. A loading end 3 is provided at the second end of the fixed bracket 1, and the loading end 3 is slidably arranged on the guide rail, and a first end of a loading push rod 4 passes through a side plate of a protective cover 5 and is connected to the loading end 3, a gear pump valve core specimen 8 is placed between the fixed end 2 and the loading end 3 and is located in an insulation box, and the loading push rod 4 is started, and the loading push rod 4 drives the loading end 3 to move along the guide rail, so as to apply a loading force to the gear pump valve core specimen 8.

[0126] Example 1

[0127] like Figure 5 to Figure 7 As shown, the present invention provides a method for predicting stress relaxation of a gear pump valve core, and the specific steps are as follows:

[0128] S1. Select multiple identical gear pump valve core specimens 8 to be tested. The initial length of the selected gear pump valve core specimens to be tested is 10mm to 300mm; the gear pump valve core specimen can withstand a tensile / compression distance of ±100mm; the gear pump valve core specimen can withstand a maximum tension / pressure of 200N; the outer diameter of the gear pump valve core is 2mm to 10mm.

[0129] S2, installing the selected gear pump valve core specimen 8 on the loading fixture in the stress relaxation testing machine, starting the loading fixture, and starting the stress test of the first group of gear pump valve core specimens 8;

[0130] S3. The stress values ​​measured at each time point of the gear pump valve core specimen 8 during the loading time are collected in real time through the data acquisition and processing module 9 in the stress relaxation testing machine, and the collected stress values ​​are stored in the data acquisition and processing module 9, and the data measured during the loading time are divided into a training phase data set and a test phase data set, specifically: 3 groups of gear pump valve core pressure data of the same type are obtained and integrated into one data set, each group of data contains 360,000 time points; data sampling processing, equidistant sampling of 3 groups of 360,000 data each, a total of 3 groups of 10,000 data each are extracted; training set and test set are selected: the first 60% of the data are selected as training data, i.e., the training phase data set, and the last 40% are selected as test data, i.e., the test phase data set.

[0131] S4. According to the training phase data set recorded in step S3, the load loss rate of the gear pump valve core test piece 8 at each time point in the training phase is calculated, and the expression is:

[0132]

[0133] Where P0 represents the initial stress value of the gear pump valve core specimen, P t represents the stress value at time t;

[0134] S5. Use statistical software (MATLAB) to fit the data, and use the least squares method to fit the logarithmic regression model of the load loss rate and relaxation time of the gear pump valve core specimen. The expression is:

[0135]

[0136] In the formula, represents the load loss rate, t R represents the relaxation time, A and B represent the parameters of the regression model;

[0137] S6. According to the regression model in step S5, the parameter values ​​of A and B in the regression model of the first group of gear pump valve core test pieces 8 are determined, and A=-2.3675, B=0.1094 are calculated;

[0138] S7. Calculate the stress prediction value of the first group of gear pump valve core specimens during the test phase according to the fitted regression model;

[0139] S8, calculate the error between the stress prediction value of the first group of gear pump valve core specimens during the test phase and the actual stress value recorded in step S3 during the test phase, and obtain the root mean square error, mainly using the mean square error (RMSE) and the coefficient of determination (R 2 ) as an evaluation index to verify the fitting quality of the regression model;

[0140] S9, repeating steps S2 to S8, and performing stress tests on the remaining gear pump valve core specimens, that is, performing stress tests on the second group of gear pump valve core specimens and the third group of gear pump valve core specimens in turn, and selecting a group of gear pump valve core specimen regression models with the smallest root mean square difference as the optimal regression model;

[0141] Table 1 Parameter values ​​of multiple groups of gear pump valve core test pieces

[0142] Serial number Parameter value A Parameter value B Root mean square error (RMSE) <![CDATA[Coefficient of determination (R 2 )]]> result 1 -2.3675 0.1094 0.8939 0.0101 reserve 2 -1.6154 0.1480 0.8010 0.0227 reserve 3 -4.8597 0.4868 12.8559 0.1506 give up

[0143] In model evaluation, it is usually desirable to have a smaller mean square error, as this indicates that the model has better predictive performance. Therefore, the fitting parameters for the gear pump valve core specimen are most appropriate.

[0144] S10. Through the optimal regression model, the stress value of the same type of gear pump valve core specimen at any time point can be predicted, so that the stress relaxation amount of the gear pump valve core specimen can be predicted in advance.

[0145] Furthermore, the present invention can also perform predictive analysis by constructing an LSTM model, specifically:

[0146] Step 1: Obtain 5 sets of gear pump valve core pressure data of the same type and integrate them into one data set, each set of data contains 360,000 time points.

[0147] Step 2: Data sampling: 5 groups of 360,000 data are sampled at equal distances, and a total of 5 groups of 10,000 data are selected in each group; the training set and test set are selected: the first 60% of the data are selected as training data, and the last 40% are selected as test data;

[0148] Step 3: Data normalization: Both training data and test data are normalized. The original data is converted into data within a certain range by using the maximum and minimum values ​​of the variables, thereby eliminating the impact of dimension and order of magnitude.

[0149] Step 4: Build LSTM: Long Short-Term Memory Network LSTM is a variant of RNN, which is specially designed to process sequence data. It is able to handle sequence inputs of different lengths and influence the current output by utilizing the hidden state of the previous time step. Traditional RNNs usually encounter the problem of gradient disappearance or gradient explosion when learning long-term dependencies, making it difficult to effectively capture information over long time spans. LSTM solves this problem by introducing three key gating mechanisms in its hidden layer: forget gate, input gate, and output gate. These gating units regulate the flow of information and state updates, allowing LSTM to better retain and utilize long-term memory, thereby excelling in tasks involving long-term dependencies.

[0150] Step 5: Build LSTM: Long Short-Term Memory Network LSTM is a variant of RNN, which is specially designed to process sequence data. It is able to handle sequence inputs of different lengths and influence the current output by utilizing the hidden state of the previous time step. Traditional RNNs usually encounter the problem of gradient disappearance or gradient explosion when learning long-term dependencies, making it difficult to effectively capture information over long time spans. LSTM solves this problem by introducing three key gating mechanisms in its hidden layer: forget gate, input gate, and output gate. These gating units regulate the flow of information and state updates, allowing LSTM to better retain and utilize long-term memory, thereby excelling in tasks involving long-term dependencies.

[0151] f t =σ(W f [h t-1 ,x t ]+b f )

[0152] In the formula, f t Represents the output vector of the forget gate, and the value range of the elements is between 0 and 1. Each element represents the degree to which the corresponding component in the cell state is retained. A value of 0 indicates complete forgetting, and a value of 1 indicates complete retention; σ represents the Sigmoid activation function. It compresses the result of the linear combination to between 0 and 1; W f Represents the weight matrix of the forget gate. It contains the weights used to calculate the forget gate, which are connected to the current input and the hidden state of the previous moment; [h t-1 ,x t ] indicates that the expression represents the connection of two vectors; h t-1 represents the hidden state vector at the previous moment. It carries the information passed down from the previous time step; x t Represents the input vector of the current time step. It contains the external input data of the current time step; b f Represents the bias vector of the forget gate. It is used to adjust the output of each calculation and provide greater flexibility for the network.

[0153] Step 6: Construct an input gate to determine the update in the state cell: First, a Sigmoid activation function is used to generate an "input gate" vector to indicate which information can be updated; at the same time, a Tanh activation function is used to generate a "candidate state" vector to indicate the new information content. After multiplying these two, the result is added to the current cell state to determine the extent to which new information is written.

[0154] I t =σ(W1[h t-1 ,x t ]+b1)

[0155]

[0156] In the formula, I t Represents the input gate vector, and the value of the element is between 0 and 1. It determines which elements' new information will be added to the cell state; σ represents the Sigmoid activation function. It compresses the result of the linear combination to between 0 and 1; W1 represents the weight matrix of the input gate. It is used to calculate the input gate vector, connecting the current input and the hidden state of the previous moment; [h t-1 ,x t ] represents the connection between the hidden state of the previous moment and the input vector of the current time step; h t-1 represents the hidden state at the previous moment; x t Represents the input vector of the current time step; b1 represents the bias vector of the input gate. It is used to adjust the output of the input gate vector; Represents the candidate cell state vector, and compresses the result of the linear combination to between -1 and 1 through the tanh activation function. Represents the new information that may be added to the cell state at the current time step; tanh represents the hyperbolic tangent activation function. It is used to limit the value of the candidate state to between -1 and 1; W c represents the weight matrix used to calculate the candidate cell state; b c The bias vector representing the candidate cell state; C t Represents the updated cell state. It combines the output of the forget gate to represent the retained old information and the output of the input gate to represent the newly added information.

[0157] Step 7: Construct the output gate to determine the output at time t: Combine the current input and the hidden state at the previous moment to generate an output gate vector. Then, this vector is multiplied by the cell state processed by the Tanh activation function to obtain the hidden state output at the current moment.

[0158] O t =σ(W0[h t-1 ,x t ]+b0)

[0159] h t =O t tanhC t

[0160] In the formula, O t Represents the output gate vector, and the value of the element is between 0 and 1. It determines which information in the cell state will be output to the current hidden state; σ represents the Sigmoid activation function. It is used to compress the result of the linear combination to between 0 and 1; W0 represents the weight matrix of the output gate. It is used to calculate the output gate vector, connecting the current input and the hidden state of the previous moment; [h t-1 ,xt ] represents the connection between the hidden state of the previous moment and the input vector of the current time step; h t-1 represents the hidden state at the previous moment; x t represents the input vector of the current time step; b0 represents the bias vector of the output gate. It is used to adjust the calculation result of the output gate vector; h t Represents the hidden state of the current time step. It is obtained by multiplying the output gate vector with the nonlinear transformation of the cell state, that is, through the tanh function. It contains the output information of the current time step and will be passed to the next time step; C t Represents the cell state at the current time step. Contains the information updated by the forget gate and input gate.

[0161] Step 8: Train and evaluate the LSTM model to obtain the optimized model: Use the first dataset to train the LSTM model, and then test it on the second dataset. The root mean square error (RMSE) is used as the loss function to guide the optimization of the model during the training process.

[0162]

[0163] Where n is the total number of samples. It represents the number of samples in the dataset, that is, the number of prediction-true value pairs we consider when calculating the error. i represents the actual value. In regression problems, this is the true observed value or label of the (i)th sample. f(x i ) represents the predicted value. This is the prediction result given by the model for the (i)th sample. i -f(x i ) represents the square of the error. It represents the square of the prediction error of the (i)th sample, reflecting the degree of deviation between the predicted value and the true value. The square operation amplifies the larger error and ensures that the error is non-negative.

[0164] The smaller the RMSE, the closer the model's predicted value is to the actual value, and the better the model's prediction performance is. RMSE is particularly suitable for performance comparison between different models of the same or similar data sets.

[0165] Example 2

[0166] In a preferred embodiment 2 of the present invention, a life test is conducted on the gear pump valve core, and the life prediction behavior of the gear pump valve core based on the LSTM model is simulated. The test piece that has been subjected to the durability test is used to continue the test. The motor of the test bench is energized, and the gear pump valve core is compressed to a certain displacement. The pressure value in a high temperature environment of 180°C is recorded in real time by the pressure sensor. 6 groups of pressure values ​​are collected, and these groups of data are preprocessed; the six groups of test data are aligned, and the data are normalized so that the values ​​of each group are close to 1; using the operating status data, 6 groups of 10,000 length samples are obtained by equidistant value taking to obtain readable data of the complete time. The operating status data is divided into a first data set and a second data set, 60% of the operating status data is used as the first data set for model training, and the remaining 40% of the operating status data is used as the second data set for life prediction.

[0167] A long short-term memory network (LSTM) model is constructed and trained using the operating status data. After the model training is completed, the second data set is input into the model for multiple iterative calculations to predict the service life of the gear pump valve core.

[0168] In this example, the learning rate of the LSTM model was set to 0.0001 and trained for 100 iterations. By selecting the RMSE loss function, the model's prediction results were compared with the actual data to evaluate the accuracy of the prediction. The visualization results of the entire training process are shown in Figure 2. Figure 8 As shown, the trend of loss is shown. In addition, Fig. 9 It shows the comparison between the predicted results and the true values ​​on the training set. Fig.10 The figure below shows the comparison between the prediction results on the test set and the real data. These images help us intuitively understand the performance and prediction accuracy of the model, thus providing a basis for further adjustment and optimization of the model.

[0169] The gear pump valve core stress relaxation prediction method provided by the present invention can fit stress relaxation test data more accurately by using a logarithmic regression model of load loss rate and relaxation time, and adopts a systematic error analysis and model verification strategy, so that the model can maintain a high prediction accuracy under different experimental conditions. Compared with the traditional linear model, this method performs better in capturing nonlinear relaxation behavior and significantly improves the fitting accuracy. In addition, the integrated data processing and analysis process reduces human errors and improves overall work efficiency. The automated data acquisition and calculation optimizes the laboratory workflow, allowing researchers to obtain effective results more quickly, solving the problem that traditional methods mainly rely on large samples, long-term physical experiments, and take a long time and are costly.

[0170] In a second aspect, the present invention provides a method for predicting stress relaxation of a gear pump valve core. Fig.11 and Fig.12 As shown, it includes the following steps:

[0171] S1. Select the gear pump valve core specimen to be tested;

[0172] S2. Install the selected gear pump valve core specimen on the loading fixture in the stress relaxation testing machine, start the loading fixture, and start the gear pump valve core specimen stress test;

[0173] S3, using the data acquisition and processing module in the stress relaxation tester, collecting the initial stress value of the gear pump valve core specimen and the stress values ​​measured at each time point in real time, and storing the collected stress values ​​in the data acquisition and processing module;

[0174] S4, preprocessing the collected stress value data of the gear pump valve core specimen, specifically including the following steps:

[0175] S41, the stress values ​​of the gear pump valve core specimen collected during the heating stage and the cooling stage of the insulation box during the removal test;

[0176] S42, using a wavelet transform method to perform denoising on the collected stress values ​​of the gear pump valve core specimen;

[0177] S43, normalizing the denoised stress value data to ensure consistency of subsequent processing, and the expression is:

[0178]

[0179] Where, X i represents the stress value data after denoising, X min Represents the minimum value of the data set, X max Indicates the maximum value of the data set.

[0180] S5, using a variety of fitting methods to fit the stress value data preprocessed in step S4;

[0181] S6, calculating the error between the predicted value in each fitting method and the actual stress value in step S3, and verifying the fitting quality of each fitting method;

[0182] S7. Determine the optimal data fitting method for the stress value of the gear pump valve core specimen according to the preset fitting model, and the fitting model expression is:

[0183]

[0184] In the formula, w1 represents the residual weight coefficient, w2 represents the root mean square error weight coefficient, and w3 represents the determination coefficient weight coefficient. Represents the average value of the absolute value of the residual, RMSE represents the root mean square error, R 2 represents the coefficient of determination.

[0185] S8. Predict the stress value of the same type of gear pump valve core specimen at any time point through a determined fitting method, so as to predict the stress relaxation amount of the gear pump valve core specimen in advance.

[0186] Furthermore, the multiple fitting methods of step S5 include a linear regression fitting method, a polynomial regression fitting method, an exponential regression fitting method and a nonlinear regression fitting method.

[0187] Example 3

[0188] like Figures 11 to 26 As shown, the gear pump valve core performance degradation model data expansion method provided by the present invention comprises the following specific steps:

[0189] S1. Select the gear pump valve core specimen to be tested;

[0190] S2. Install the selected gear pump valve core specimen onto the loading fixture in the stress relaxation testing machine, start the loading fixture, and start the gear pump valve core specimen stress test, as follows:

[0191] S21. Fix the gear pump valve core test piece on a special loading fixture, and ensure that the compression amount of the gear pump valve core test piece is always maintained at 1.7 cm;

[0192] S22. Use a force sensor to record the stress relaxation data of the gear pump valve core specimen under this compression amount.

[0193] S23. Set the sampling frequency to 1 Hz and the test time to 100 hours, i.e. 360,000 seconds, to ensure the continuity and integrity of the data during the entire gear pump valve core specimen relaxation process.

[0194] S3, using the data acquisition and processing module in the stress relaxation tester, collecting the initial stress value of the gear pump valve core specimen and the stress values ​​measured at each time point in real time, and storing the collected stress values ​​in the data acquisition and processing module;

[0195] S4, preprocessing the collected stress value data of the gear pump valve core specimen, specifically including the following steps:

[0196] S41. After obtaining data for a total of 100 hours (360,000 seconds), the loading fixture will have unstable data at both ends from the beginning to the end of loading. According to the time regulations for heating and cooling the test chamber, the collected data is preliminarily checked to remove obviously wrong or abnormal measurements to ensure the rationality of the data set. In this method, from the overall sample data, the first 1 / 10 and the last 1 / 100 of the data are removed, that is, the data from 3600 seconds to 356,400 seconds are selected. The following two methods can be used for sample selection and estimation:

[0197] Method 1: Half sample size: Half sample size estimation is done by randomly sampling half of the original data set. This method can be used to quickly obtain data characteristics and is used for preliminary analysis and model validation. 50% of the data points are randomly selected from the complete data set to ensure the randomness of the sampling and the representativeness of the sample. It is fast, has low computational complexity, and is used for preliminary exploration.

[0198] Method 2: Full sample estimation: Full sample estimation uses all collected data for comprehensive analysis to provide more accurate and reliable estimation results. This method uses data from 3600-356400s as full sample estimation.

[0199] S42, the wavelet transform method is used to denoise the stress values ​​of the gear pump valve core specimens collected, mainly:

[0200] S421, selecting a suitable mother wavelet (Daubechies wavelet) to perform multi-scale decomposition;

[0201] S422, performing threshold processing on the decomposed high-frequency coefficients to remove noise;

[0202] S423, reconstructing the signal to obtain denoised pressure data;

[0203] S43, normalizing the denoised stress value data to ensure consistency of subsequent processing, and the expression is:

[0204]

[0205] Where, X i represents the stress value data after denoising, X min Represents the minimum value of the data set, X max Indicates the maximum value of the data set.

[0206] S5. Using a variety of fitting methods to fit the stress value data preprocessed in step S4, the various fitting methods include linear regression fitting method, polynomial regression fitting method, exponential regression fitting method and nonlinear regression fitting method.

[0207] Method 1: Linear regression fitting method

[0208] Linear fitting is a basic statistical method used to describe the relationship between two variables. Its goal is to find a straight line that can describe the linear relationship between the variables as accurately as possible. Linear fitting is usually achieved through the least squares method. The goal of the least squares method is to find the line that minimizes the sum of the squares of the vertical distances from all data points to the fitted line.

[0209] y=ax+b

[0210] In the formula, y represents the dependent variable as the response variable); x represents the independent variable as the predictor variable; a represents the slope, which represents the change in the dependent variable for each unit change in the independent variable; b represents the intercept, which represents the value of the dependent variable when the independent variable is zero.

[0211] Specifically, when half sample size data is used, the total sample size is S = 359959, the sampling number is s = 176,380, and the fitting result is y = 0.0000036x-48.69. The fitted curve is as follows: Fig.13 As shown, the coefficient of determination is R 2 =0.031899, residual graph, such as Fig.14 When the full sample data is used, the total sample size is S = 359959, the sampling number is s = 349159, and the fitting result is y = 0.0000018x-48.53. The fitted curve is as follows Fig.15 As shown, the coefficient of determination is R 2 =0.029999, residual graph, such as Fig.16 shown.

[0212] Method 2: Quadratic polynomial fitting method

[0213] Quadratic polynomial fitting is a statistical method that approximates the trend in a data set through a quadratic polynomial function. Unlike linear fitting, quadratic polynomial fitting can better capture nonlinear relationships in the data. Quadratic polynomial fitting usually uses the least squares method to find the best fit curve. Its goal is to minimize the sum of the squares of the vertical distances from all data points to the fitted curve.

[0214] Specifically, when half sample size data is used, the total sample size is S = 359959, the number of samples is s = 176,380, and the result after fitting is y = 0.000000000042x 2 -0.0000041x-48.44, the fitted curve is as follows Fig.17 As shown, the coefficient of determination is R 2 =0.040893, residual graph, such as Fig.18When the full sample data is used, the total sample size is S = 359959, the sampling number is s = 349159, and the fitting result is

[0215] y=0.000000000042x 2 -0.0000041x-48.44, the fitted curve is as follows Fig.19 As shown, the coefficient of determination is R 2 =0.034731, residual graph, such as Fig. 20 shown.

[0216] Method 3: Cubic polynomial fitting method

[0217] Cubic polynomial fitting is a statistical method that approximates the trend of a data set by a cubic polynomial. Compared with linear and quadratic polynomial fitting, cubic polynomial fitting can capture more complex nonlinear relationships and curvilinear changes in data. Cubic polynomial fitting usually uses the least squares method to find the best fit curve to minimize the sum of the squares of the vertical distances from the data points to the fitted curve.

[0218] Specifically, when half sample size data is used, the total sample size is S = 359959, the number of samples is s = 176,380, and the result after fitting is y = -0.00000000000000007x 3 +0.00000000024x 2 -0.000019x-48.2, the fitted curve is as follows Fig.21 As shown, the coefficient of determination is R 2 =0.045997, residual graph, such as Fig. 22 When the full sample data is used, the total sample size is S = 359959, the sampling number is s = 349159, and the fitting result is y = -0.000000000000000005651x 3 +0.000000000022x 2 +0.00000029x-48.56, the fitted curve is as follows Fig.23 As shown, the coefficient of determination is R 2 =0.036602, residual graph, such as Fig.24 shown.

[0219] Method 4: Sine function model fitting method

[0220] Sine function fitting is a technique used to fit data into a sinusoidal form, and is often used to analyze data with periodic or fluctuating characteristics. Its purpose is to find appropriate sine function parameters so that the function can best describe the periodic characteristics of the data. A typical sine function can be expressed as:

[0221] y=A / sin(Bx+C)+D

[0222] Where A is the amplitude, which indicates the height of the wave; B is the angular frequency; C is the phase offset, which controls the horizontal movement of the waveform on the time axis; and D is the vertical offset, which indicates the overall movement of the waveform in the vertical direction.

[0223] Specifically, when half sample size data is used, the total sample size is S = 359959, the sampling number is s = 176,380, and the fitted curve is as follows: Fig.25 As shown, the coefficient of determination is R 2 =0.03849.

[0224] Method 5: Custom model fitting

[0225] Custom model fitting, the expression is:

[0226] f(x)=a*(sin(x-pi))+b*((x-10) 2 )+c

[0227] Where a represents the amplitude of the sine function; b represents the coefficient of the quadratic term; and c represents the coefficient of the quadratic term.

[0228] Specifically, when half sample size data is used, the total sample size is S = 359959, the number of samples is s = 176,380, and the result after fitting is y = a*(sin(x-pi))+b*((x-10) 3 )+c*x 2 +d*x+e, where a=0.0004233, b=-7.036e-16, c=2.355e-10, d=-1.858e-05, e=-48.2. The fitted curve is as follows Fig.26 As shown, the coefficient of determination is R 2 =0.046.

[0229] S6. Calculate the error between the predicted value in each fitting method and the actual stress value in step S3 to verify the fitting quality of each fitting method, specifically, calculate the determination coefficient between the predicted value of the regression model and the actual stress value, and its expression is:

[0230]

[0231] In the formula, y i represents the actual stress value, represents the predicted value, represents the average value of the true observation, and n represents the number of samples;

[0232] S7. According to the preset fitting model, the optimal data evaluation value expression of the gear pump valve core specimen stress value is determined as follows:

[0233]

[0234] In the formula, w1 represents the residual weight coefficient, w2 represents the root mean square error weight coefficient, and w3 represents the determination coefficient weight coefficient. Represents the average of the absolute values ​​of the residuals, and RMSE stands for root mean square error, which provides an overall measure of the prediction error, emphasizing large errors because it squares the error. A smaller RMSE indicates a good overall prediction performance of the model. 2 R represents the coefficient of determination, which measures the model's ability to explain data changes. The closer it is to 1, the stronger the model's ability to explain the data. By multiplying the weight c, we can flexibly adjust R according to specific needs. 2 The comprehensive evaluation formula can effectively integrate the residual value, RMSE and R 2 The information provided by the algorithm provides valuable guidance for model selection and optimization. In this embodiment, w1=3, w2=4, and w3=3 are selected for evaluation.

[0235] S8. Predict the stress value of the same type of gear pump valve core specimen at any time point through a determined fitting method, so as to predict the stress relaxation amount of the gear pump valve core specimen in advance.

[0236] The present invention integrates a variety of data fitting methods to select the most suitable fitting model according to different data features. In this way, the changing trends of each stage in the gear pump valve core relaxation process can be captured more accurately, thereby improving the accuracy of data analysis. At the same time, a multi-model fitting strategy is adopted, so that the system can adapt to gear pump valve core relaxation test data of different types and materials. Whether it is linear, nonlinear or exponential changes, the present invention can provide a suitable fitting scheme to increase the flexibility and scope of application of data processing. In addition, the ability to automatically evaluate and select the best fitting model makes the entire data processing process more efficient. The process of manually selecting models is reduced, the possibility of human errors is reduced, and overall work efficiency is improved. Through optimized algorithm design and preprocessing steps, the present invention can quickly process large amounts of data, significantly reduce data processing time, and provide technical support for the research on performance degradation models of key components of gear pumps and data expansion methods.

[0237] The above embodiments are only descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A method for predicting stress relaxation of a gear pump valve core, characterized in that: It includes the following steps: S1. Select a plurality of identical gear pump valve core test pieces to be tested, and divide them into a first group of gear pump valve core test pieces, a second group of gear pump valve core test pieces, and an Nth group of gear pump valve core test pieces; S2, installing the selected gear pump valve core specimens on the loading fixture in the stress relaxation testing machine, starting the loading fixture, and starting the stress test of the first group of gear pump valve core specimens; S3, using a data acquisition and processing module in the stress relaxation testing machine, real-time acquisition of stress values ​​of the gear pump valve core specimen measured at each time point during the loading time, and dividing the data measured during the loading time into a training phase data set and a test phase data set; S4. According to the training phase data set recorded in step S3, the load loss rate of the gear pump valve core specimen at each time point in the training phase is calculated, and the expression is: Where, P0 represents the initial stress value of the gear pump valve core specimen, P t represents the stress value at time t; S5. The least square method is used to fit the logarithmic regression model of the load loss rate and relaxation time of the gear pump valve core specimen, and the expression is: In the formula, represents the load loss rate, t R represents the relaxation time, A and B represent the parameters of the regression model; S6, according to the regression model in step S5, determining the parameter values ​​of A and B in the regression model of the first group of gear pump valve core specimens; S7. Calculate the stress prediction value of the first group of gear pump valve core specimens during the test phase according to the fitted regression model; S8, calculating the error between the stress prediction value of the first group of gear pump valve core specimens during the test phase and the actual stress value of the test phase recorded in step S3, and obtaining the root mean square difference to verify the fitting quality of the regression model; S9, repeating steps S2 to S8, carrying out stress tests on the second group of gear pump valve core specimens and the Nth group of gear pump valve core specimens in turn, and selecting a group of gear pump valve core specimen regression models with the smallest root mean square difference as the optimal regression model; S10. Through the optimal regression model, the stress value of the same type of gear pump valve core specimen at any time point can be predicted, so that the stress relaxation amount of the gear pump valve core specimen can be predicted.

2. The method for predicting stress relaxation of a gear pump valve core according to claim 1, characterized in that: In step S2, the selected gear pump valve core specimen is installed on a loading fixture in the stress relaxation testing machine, the loading fixture is started, and the stress test of the first group of gear pump valve core specimens is started. The specific steps include: S21, installing the first set of gear pump valve core test pieces on the loading fixture, and setting the initial test parameters; S22, raising the temperature of the incubator to a predetermined test temperature and performing heat preservation; S23, starting the loading push rod to apply a loading force to the first group of gear pump valve core test pieces so that the gear pump valve core test pieces maintain a certain elastic displacement; S24, setting the loading temperature and loading time of the gear pump valve core test piece; S25, recording in real time the elastic force generated by the gear pump valve core test piece at each time point through a force sensor; S26. When the elastic force of the gear pump valve core specimen decays to the loading time, the stress test of the first group of gear pump valve core specimens is stopped.

3. The method for predicting stress relaxation of a gear pump valve core according to claim 2, characterized in that: The displacement control accuracy of the loading tool in step S23 is greater than or equal to ±0.1%.

4. The method for predicting stress relaxation of a gear pump valve core according to claim 1, characterized in that: The step S3 divides the data measured during the loading time into a training phase data set and a test phase data set, and specifically includes the following steps: S31, performing equidistant sampling on the stress value data of the gear pump valve core test piece measured at each time point during the loading time collected in real time; S32, dividing the data after equidistant sampling into data sets; S33, using 60% of the sampled data as a training phase data set, which can be used to fit the regression model of the gear pump valve core test piece; S34. 40% of the sampled data are used as a test phase data set to test the error between the stress value predicted by the regression model and the actual stress value in the test phase.

5. The method for predicting stress relaxation of a gear pump valve core according to claim 4, characterized in that: The sampling frequency of the data acquisition processing module in step S3 is greater than or equal to 1 Hz.

6. A gear pump valve core performance degradation model data expansion method, characterized in that: It includes the following steps: S1. Select the gear pump valve core specimen to be tested; S2. Install the selected gear pump valve core specimen on the loading fixture in the stress relaxation testing machine, start the loading fixture, and start the gear pump valve core specimen stress test; S3, collecting the initial stress value of the gear pump valve core specimen and the stress values ​​measured at each time point in real time through the data acquisition and processing module in the stress relaxation tester, and storing the collected stress values ​​in the data acquisition and processing module; S4, preprocessing the collected stress value data of the gear pump valve core test piece; S5, using a variety of fitting methods to fit the stress value data preprocessed in step S4; S6, calculating the error between the predicted value in each fitting method and the actual stress value in step S3, and verifying the fitting quality of each fitting method; S7. According to the preset fitting model, the optimal data evaluation value expression of the gear pump valve core specimen stress value is determined as follows: Where w1 represents the residual weight coefficient, w2 represents the root mean square error weight coefficient, w3 represents the determination coefficient weight coefficient, e| represents the average value of the absolute value of the residual, RMSE represents the root mean square error, R 2 represents the coefficient of determination; S8. By determining the optimal fitting method, the stress value of the same gear pump valve core specimen at any time point is expanded, so as to predict the stress relaxation amount of the gear pump valve core degradation model.

7. The gear pump valve core performance degradation model data expansion method according to claim 6, characterized in that: In step S4, the collected stress value data of the gear pump valve core test piece are preprocessed, which specifically includes the following steps: S41, the stress values ​​of the gear pump valve core specimen collected during the heating stage and the cooling stage of the insulation box during the removal test; S42, using a wavelet transform method to perform denoising on the collected stress values ​​of the gear pump valve core specimen; S43, normalizing the denoised stress value data to ensure consistency of subsequent processing, and the expression is: Where, X i represents the stress value data after denoising, X min Represents the minimum value of the data set, X max Indicates the maximum value of the data set.

8. The gear pump valve core performance degradation model data expansion method according to claim 6, characterized in that: The multiple fitting methods in step S5 include a linear regression fitting method, a polynomial regression fitting method and a nonlinear regression fitting method.

Images