Milling material fretting wear prediction method based on small sample

Abstract

The invention discloses a milling material fretting wear prediction method based on a small sample, and relates to the technical field of informatization planning. The method comprises the following steps: obtaining fretting wear samples on the surface of a milling material based on a physical experiment: sequentially changing fretting wear parameters of the surface of the milling material based on a reference test point to obtain four groups of fretting wear samples; based on the relationship between the fretting wear parameters and the fretting wear, establishing a model for predicting the fretting wear; taking a standard deviation between a predicted energy wear rate and a wear measured value as a minimum target, sequentially substituting a cycle index sample, a contact stress sample, a displacement amplitude sample and a frequency sample, and fitting the model to obtain a model capable of predicting fretting wear of the surface of the milling material; and the fretting wear of the surface of the milling material can be predicted based on the model. According to the method, the fretting wear of the milling material can be predicted only through a small number of samples, the cost is low, and the prediction precision is high.

Description

Technical Field

[0001] The present invention relates to the technical field of informatization planning, and particularly to a method for predicting fretting wear of milling materials based on small samples. Background Art

[0002] The milling process plays a key role in manufacturing high-performance components such as GH4169 superalloy turbine disks. It not only determines the geometric accuracy and surface characteristics of the workpiece but also profoundly changes the surface integrity of the material at the micro level. The high strain, high strain rate, and temperature changes during the milling process cause significant changes in the microstructure of the surface and subsurface layers of the material. These changes have a significant impact on the fretting wear behavior of the material by forming metamorphic layers (such as white layers, residual stresses, work hardening). Therefore, it is crucial to clarify the relationship between machining parameters and the fretting wear performance of the workpiece surface, which is related to the performance and durability of superalloy components in high-temperature and harsh environments such as gas turbines. However, due to the high cost of these materials, the prior art needs to predict the specific fretting wear situation according to the processing method to reduce costs, which is conducive to the informatization planning of this industry.

[0003] Currently, there are mainly two methods for fretting wear prediction modeling, namely the finite element method and the machine learning regression prediction method. Machine learning methods, with their powerful data analysis capabilities, can deeply explore the deep non-linear relationships in complex systems, and their prediction results have relatively high accuracy; however, the effectiveness of machine learning models is closely related to the quality and scale of the data, and constructing a high-quality data set is often costly and time-consuming. The finite element method (FEM), with its excellent physical modeling capabilities, has become a key means for predicting the mechanical response of materials. The prior art mainly constructs corresponding simulation models based on the friction coefficient and wear coefficient obtained through fretting wear experiments; however, this method has an important limitation: when using new tribological parameters, the previously constructed simulation model may significantly lose its accuracy. To ensure the reliability of the simulation results, it is necessary to re-enter the wear coefficient and friction coefficient under the current tribological parameters into the model. This strategy that relies on experimental data limits the adaptability of the model and makes it difficult to flexibly respond to changes in different working conditions and material properties. Summary of the Invention

[0004] To solve at least one of the above problems, the present invention proposes a method for predicting fretting wear of milling materials based on small samples, which can predict the fretting wear on the surface of milling materials with only a small number of samples and has a relatively high prediction accuracy.

[0005] The technical solution of the present invention is as follows: A method for predicting fretting wear of milling materials based on small samples, comprising the following steps:

[0006] S1. Obtain fretting wear samples on the surface of the milling material based on physical experiments: Based on a reference test point, successively change its cycle number, contact stress, displacement amplitude, and frequency to obtain 4 groups of fretting wear samples;

[0007] S2. Establish a model for predicting fretting wear based on the relationships among the cycle number, contact stress, displacement amplitude, frequency, and fretting wear: In the formula, α pred represents the predicted energy wear rate; α represents the energy wear rate; p ref , f ref , σ ref , and N ref respectively represent the contact stress, frequency, displacement amplitude, and cycle number of the reference test point; p, f, σ, and N are respectively the contact stress, frequency, displacement amplitude, and cycle number of the fretting wear sample; n p , n f , n σ , n N , a1, a2, a3, a4, b1, b2, b3, and b4 are all variables to be fitted;

[0008] S3. Taking the minimum standard deviation between the predicted energy wear rate and the measured wear value as the goal, successively substitute the cycle number sample, contact stress sample, displacement amplitude sample, and frequency sample into the model in S2 to fit the values of the variables to be fitted, and then obtain a model capable of predicting the fretting wear on the surface of the milling material;

[0009] S4. Based on the model in S3 and combined with the measured fretting wear parameters, the fretting wear on the surface of the milling material can be predicted.

[0010] Beneficial effects: The present invention only needs a small number of samples to realize the prediction of the fretting wear of the milling material, with low cost; and its prediction accuracy is relatively high, and R 2 reaches 0.96, and MAPE reaches 9.06%. Description of the Drawings

[0011] Figure 1 is the diagram for selecting the fretting wear sample in this embodiment;

[0012] Figure 2 is the diagram of the single-item fitting result of the model in this embodiment;

[0013] Figure 3 is the diagram of the prediction result of the model in this embodiment;

[0014] Figure 4 is the diagram of the prediction result error of the model in this embodiment. Detailed Embodiment

[0015] The specific embodiments of the present invention will be clearly and completely described below with reference to examples. Obviously, the described examples are only a part of the embodiments of the present invention, rather than all of them.

[0016] A method for predicting fretting wear of milling materials based on small samples includes the following steps:

[0017] S1. Based on physical experiments, obtain fretting wear samples on the surface of milling materials: Based on a reference test point, sequentially change its cycle number, contact stress, displacement amplitude, and frequency to obtain 4 groups of fretting wear samples;

[0018] In this embodiment, the milling material used is GH4169, and a tangential fretting wear device is used to test its fretting wear. Under different fretting parameter conditions, 4 groups of fretting wear samples are obtained.

[0019] The "reference test point" mentioned in this step is a point artificially set to facilitate obtaining 4 groups of fretting wear samples, which can be set according to actual situations. In this point, the cycle number, contact stress, displacement amplitude, and frequency are all reference values in the 4 groups of fretting wear samples. When obtaining fretting wear samples, the single variable method is used: that is, in each group of fretting wear samples, three of the fretting wear parameters are fixed as constants, and the other fretting wear parameter is used as a variable, and this variable needs to change before and after the reference test point.

[0020] In this embodiment, considering the particularity of the variable, the sample size of each group of fretting wear samples is not less than 3. This is mainly because: when the sample size is less than 3, its fitting lines are all straight lines, but according to the actual experience of the inventor, this is obviously not in line with the actual situation. Therefore, only when its sample size is not less than 3, the finally fitted model has better accuracy. In some cases, when the sample size of each group of fretting wear samples is 3 - 10, it has better effects: the sample size is small, and at the same time, the accuracy is high; in particular, the inventor found that when the sample size is 3, the finally fitted model still has good effects.

[0021] In this embodiment, the relevant fretting wear parameters of the selected reference test point are: cycle number N ref = 5×10 4 times, contact stress p ref = 40 MPa, displacement amplitude δ ref = 150 μm, and frequency f ref = 10 Hz.

[0022] Each group of fretting wear samples is set with 3 samples, specifically as Figure 1 shown.

[0023] S2. Based on the relationships between the number of cycles, contact stress, displacement amplitude, frequency, and fretting wear, a model for predicting fretting wear is established: In the formula, α pred represents the predicted energy wear rate; α represents the energy wear rate; p ref , f ref , σ ref and N ref represent the contact stress, frequency, displacement amplitude, and number of cycles of the reference test points respectively; p, f, σ, and N are the contact stress, frequency, displacement amplitude, and number of cycles of the fretting wear samples respectively; n p , n f , n σ , n N , a1, a2, a3, a4, b1, b2, b3, and b4 are all variables to be fitted;

[0024] The model of the embodiment of the present invention is based on the current weighted friction energy method. On the basis of this method, this embodiment extends it to the surface-to-surface contact fretting wear test, comprehensively considers the influence of all load conditions, and at the same time combines a large number of previous tests of the inventor to establish it on this basis.

[0025] First of all, based on the weighted friction energy method, considering the surface-to-surface contact fretting wear test, and at the same time combining a large number of experiments of the inventor, it can be known that the wear volume is proportional to the contact stress, frequency, and displacement amplitude, and inversely proportional to the fretting cycle number. Therefore, under the condition of considering the additional contributions of all load conditions, a preliminary prediction model is obtained:

[0026]

[0027] For the energy wear rate α, it can be calculated by the following formula: α = V / E d , where V represents the wear volume and Ed represents the cumulative dissipated energy; In the formula, Ed(i) represents the dissipated energy of a single fretting cycle

[0028] However, in the actual prediction process, we found that when using the above preliminary prediction model for prediction, we found that some weights are greater than 1, which does not conform to the monotonicity reflected by the large number of previous experimental results of the inventor: According to the monotonicity reflected by the large number of experimental results of the inventor, it is found that the fitting exponents n p , n f , n σ , n NThe range should be within [-1, 1]; at the same time, the coefficient of determination of the fitting is low and does not reach the ideal level, mainly because the sample size used in this embodiment is small. Therefore, in order to change this situation, the inventor adjusted the above-mentioned preliminary prediction model, added multiple bias terms, and increased the complexity of the model to better adapt to the prediction of the fretting wear rate of the milling material. The adjusted model is as follows: This model has high precision.

[0029] In the model of this embodiment, the bias terms a (a1, a2, a3, a4) better adapt to the changes in the microstructures of the machined surface and material properties by adjusting the sensitivity of the exponential part; the bias terms b (b1, b2, b3, b4) further correct the baseline of the model so that it can more accurately reflect the initial wear state of fretting wear on the machined surface.

[0030] S3. Taking the minimum standard deviation between the predicted energy wear rate and the measured wear value as the goal, substituting the cyclic number samples, contact stress samples, displacement amplitude samples, and frequency samples into the model of S2 in sequence for fitting, obtaining the values of the variables to be fitted, and further obtaining a model capable of predicting the fretting wear on the surface of this milling material;

[0031] This process is to substitute the 4 groups of fretting wear samples obtained in S1 into the above model for fitting, so as to obtain the specific values of the fitting index and bias terms.

[0032] The fitting result of this embodiment is as Figure 2 shown. From Figure 2 it can be seen that the prediction model of this embodiment has high precision.

[0033] S4. Based on the model of S3 and combined with the measured fretting wear parameters, the fretting wear on the surface of the milling material can be predicted.

[0034] To illustrate the actual effect of the model after fitting in this embodiment, the inventor gave 13 groups of fretting wear tests under different frictional parameters, as shown in Table 1. Then, the model of this invention embodiment and the preliminary prediction model were used to predict the fretting wear, and the final results are as Figure 3 and Figure 4 shown.

[0035] Table 1 Initial sample data table

[0036]

[0037] Figure 3 For the prediction result graph, it can be seen from the graph that the method of this invention embodiment has high prediction precision.

[0038] Figure 4It is a prediction result error graph. From Figure 4 it can be seen that for the method of the embodiment of the present invention, the error is small, and the highest prediction result error of 15 groups of samples is 15.4%, and the MAPE is 9.06%.

[0039] The above are only the preferred embodiments of the present invention, and do not impose any form of limitation on the present invention. Although the present invention has been disclosed above with the preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to equivalent embodiments by using the above-disclosed technical content within the scope of the technical solution of the present invention. However, as long as it does not depart from the content of the technical solution of the present invention, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A method for predicting fretting wear of milling materials based on small samples, characterized in that: The following steps are involved: S1. Based on physical experiments, obtain the fretting wear samples of the milling material surface: based on a reference test point, change the number of cycles, contact stress, displacement amplitude and frequency in turn to obtain 4 groups of fretting wear samples; S2. Based on the relationship between the number of cycles, contact stress, displacement amplitude, frequency and fretting wear, a model for predicting fretting wear is established: In the formula, α pred represents the predicted energy wear rate; α represents the energy wear rate; p ref 、f ref , σ ref and N ref represents the contact stress, frequency, displacement amplitude and number of cycles of the reference test point; p, f, σ and N represent the contact stress, frequency, displacement amplitude and number of cycles of the fretting wear sample; n p 、n f 、n σ 、n N , a1, a2, a3, a4, b1, b2, b3 and b4 are all variables to be fitted; S3, taking the standard deviation between the predicted energy wear rate and the measured wear value as the minimum target, sequentially substituting the cycle number sample, contact stress sample, displacement amplitude sample and frequency sample into the model of S2, obtaining the value of the variable to be fitted, and then obtaining a model that can predict the fretting wear of the milling material surface; S4. Based on the model of S3 and combined with the measured fretting wear parameters, the fretting wear on the surface of the milled material can be predicted.

2. The method according to claim 1, characterized in that In S1, the sample size of each group of fretting wear samples is not less than 3.

3. The method according to claim 2, characterized in that In S1, the sample size of each group of fretting wear samples ranged from 3 to 10.

4. The method according to claim 1 or 2, characterized in that: The sample size of each group of fretting wear samples in S1 is 3.

5. The method according to claim 1, characterized in that In S1, four groups of fretting wear samples were obtained using the single variable method.

6. The method according to claim 1, characterized in that In S2, the energy wear rate is calculated by the following formula: α = V / E d , where V represents the wear volume, E d represents the cumulative dissipated energy; Where Ed(i) represents the dissipated energy of a single micromotion cycle, i represents the micromotion cycle number, and i=1,2,…,N.

7. The method according to claim 1, characterized in that In S2, the standard deviation between the predicted energy wear rate and the measured wear value is as follows: In the formula, SD represents the standard deviation; n represents the number of samples.

Images