Cast iron cylinder sleeve structure and alloy component optimization method based on friction and wear performance prediction

Abstract

The invention discloses a cast iron cylinder sleeve structure and alloy component optimization method based on friction and wear performance prediction, and belongs to the technical field of material science and machine manufacturing. The problems that a traditional cylinder sleeve alloy design method is low in efficiency and the material performance of the designed cylinder sleeve alloy is limited are solved. The method comprises the following steps: collecting an existing large-cylinder-diameter engine sample, and carrying out metallographic characterization and microhardness test on an engine cylinder sleeve to obtain cylinder sleeve structure characteristics and alloy component data; the wear volumes of different parts are obtained through a friction wear experiment, and then the wear rate is calculated based on the wear volumes. After tissue characteristic data, alloy component data and frictional wear test data of the cylinder sleeve are collected, the data are imported into different machine learning models, the models learn the corresponding relation between the tissue characteristic data and the frictional wear performance of the existing cylinder sleeve, and optimization of material tissue and alloy components is guided according to the prediction result of the optimal model. The method can be applied to optimization of cylinder sleeve structures and alloy components.

Description

Technical Field

[0001] This invention belongs to the field of materials science and mechanical manufacturing technology, specifically relating to a method for optimizing the microstructure and alloy composition of cast iron cylinder liners based on the prediction of friction and wear performance. Background Technology

[0002] With the rapid development of industrial technology, the engine, as the core component of power equipment, directly affects the efficiency and lifespan of the entire system. The cylinder liner, as a critical component of the engine, directly impacts its operating efficiency and service life. Traditional cylinder liner alloy design methods rely primarily on experience and trial-and-error, resulting in time-consuming and costly processes. Furthermore, predicting friction and wear performance depends on laboratory testing, leading to low data acquisition efficiency and hindering rapid optimization.

[0003] While some optimization methods exist in existing technologies, they often lack a systematic approach and fail to effectively combine material design and performance prediction. For example, some studies focus solely on optimizing alloy composition while neglecting the systematic prediction of friction and wear performance; others, although predicting friction and wear performance, lack efficient machine learning methods to optimize alloy composition. Therefore, there is an urgent need for an efficient and accurate method for optimizing cylinder liner microstructure and alloys to improve design efficiency and material properties. Furthermore, with the increasing complexity and high performance of industrial equipment, the requirements for cylinder liner alloys are also becoming increasingly stringent. Traditional alloy design methods are no longer sufficient to meet the needs of modern industry, especially in terms of high performance, long service life, and low friction loss.

[0004] In summary, in order to solve the problems of low efficiency and limited material properties of traditional cylinder liner alloy design methods, it is urgent to propose a new method for optimizing the microstructure and alloy composition of cast iron cylinder liners. Summary of the Invention

[0005] The purpose of this invention is to address the problems of low efficiency and limited material properties of traditional cylinder liner alloy design methods, and to propose a method for optimizing the microstructure and alloy composition of cast iron cylinder liners based on the prediction of friction and wear performance.

[0006] The technical solution adopted by this invention to solve the above-mentioned technical problems is: a method for optimizing the microstructure and alloy composition of cast iron cylinder liners based on the prediction of friction and wear performance, the method specifically including the following steps:

[0007] Step 1: Select N types of marine low-speed engine cylinder liner materials, and then cut each selected material to obtain a standard sample of each material;

[0008] The standard samples are then sanded and polished using sandpaper and polishing agent to obtain the processed standard samples of various materials.

[0009] Step 2: Collect material characteristic data based on the processed standard sample. The material characteristic data includes metallographic data, hardness data, and content data of each element.

[0010] Step 3: Collect friction and wear data on the processed standard samples, including the friction coefficient and wear rate;

[0011] Step 4: Clean and standardize the raw data collected in Steps 2 and 3 to obtain standardized samples.

[0012] Step 5: Divide the samples obtained in Step 4 into three parts: training set, validation set, and test set, according to a 6:2:2 ratio.

[0013] Each model is trained, validated, and tested using the training set, validation set, and test set, and the best model is selected.

[0014] Step 6: Use the selected optimal model to predict the friction coefficient and wear rate of the material to be optimized, and use the prediction results to guide the optimization of material microstructure and alloy composition.

[0015] Furthermore, the standard sample has dimensions of 30 mm × 30 mm × 6 mm.

[0016] Furthermore, the process of using sandpaper and polishing agent to grind and polish the standard sample specifically involves:

[0017] For any standard specimen, the standard specimen is polished in sequence using sandpaper with abrasive grit of 400#, 800#, 1500#, 2000# and 3000#, and then polished in sequence using polishing agents with thicknesses of 2.5μm, 1.0μm and 0.5μm to obtain the treated standard specimen.

[0018] Each standard sample is then processed separately to obtain the processed standard samples of various materials.

[0019] Furthermore, the metallographic data acquisition method is as follows: a 3×3 dot matrix is ​​selected on the surface of each processed standard sample, with a distance of 3 mm between adjacent points on the same surface. The selected points are etched using a 4% nitric acid alcohol solution for 10 seconds. Metallographic sampling at 100x magnification is performed at each etched point to obtain a metallographic photograph at each point. Then, each acquired metallographic photograph is processed, specifically as follows:

[0020] For a metallographic image, the area ratio of graphite phase, pearlite matrix and high hardness phase in the metallographic image is statistically analyzed. The area ratio of graphite phase is taken as the content of graphite phase, the area ratio of pearlite matrix is ​​taken as the content of pearlite matrix, and the area ratio of high hardness phase is taken as the content of high hardness phase.

[0021] For a standard sample of a material after processing, the graphite phase content of 9 metallographic images corresponding to the standard sample is averaged, and the average value is taken as the graphite phase content of the material. The pearlite matrix content of 9 metallographic images corresponding to the standard sample is averaged, and the average value is taken as the pearlite matrix content of the material. The high hard phase content of 9 metallographic images corresponding to the standard sample is averaged, and the average value is taken as the high hard phase content of the material.

[0022] Furthermore, the hardness data is acquired in the following way:

[0023] For a standard sample of a material after treatment, 100 points are selected on the surface of the standard sample to perform microhardness testing on the pearlite matrix. The average hardness data of the pearlite matrix at the 100 points is then taken as the pearlite matrix hardness of the material. Similarly, 100 points are selected on the surface of the standard sample to perform microhardness testing on the high-hardness phase. The average hardness data of the high-hardness phase at the 100 points is then taken as the high-hardness phase hardness of the material.

[0024] Furthermore, the method for collecting the content data of each element is as follows:

[0025] For a standard sample of a material after treatment, an inductively coupled plasma mass spectrometer is used to test the standard sample to obtain the alloy composition of the material, that is, the content of C, Si, Mn, P, S, Cu, Sn and Cr elements in the material.

[0026] Furthermore, the wear rate is calculated as follows:

[0027] (1)

[0028] in, Let P be the wear volume of the material, P be the applied load, and L be the sliding distance.

[0029] Furthermore, the specific process of step four is as follows:

[0030] Step 4: Through the processing in Step 2 and Step 3, a set of data consisting of metallographic data, hardness data, content data of each element, friction coefficient and wear rate was obtained for the standard sample of each material. It was then determined whether there were any outliers in the data corresponding to each material, and all data in the group containing the outliers were removed.

[0031] Step 42: Standardize the remaining data sets:

[0032] If the remaining data sets are normally distributed, then the standardization method used is Z-score standardization.

[0033] If the remaining data sets are distributed in other ways, then the standardization method used is Min-Max standardization.

[0034] Furthermore, the process of selecting the optimal model is as follows:

[0035] The mean squared error and coefficient of determination of each model are compared, and the model with the smallest mean squared error and the largest coefficient of determination is selected as the best model.

[0036] The beneficial effects of this invention are:

[0037] This invention collects existing large-bore engine samples, then performs metallographic characterization and microhardness testing on the engine cylinder liners to obtain microstructural characteristics and alloy composition data for different parts of the cylinder liner. Wear volumes for different parts are obtained through friction and wear experiments, and the wear rate is calculated based on these volumes. After collecting the microstructural characteristics, alloy composition, and friction and wear test data of the cylinder liners, the collected data is cleaned and then imported into different machine learning models. Machine learning is used to train and learn the correspondence between the existing microstructural characteristics of cylinder liners and their friction and wear performance, thereby obtaining the laws governing the friction and wear performance of different cylinder liner microstructural characteristics. The selected optimal model is used to predict material performance, avoiding the inefficiency of obtaining friction and wear performance through experimental methods. Furthermore, the parameters predicted by the machine learning model can guide the optimization of material design, thereby improving material design efficiency and further optimizing material performance, ultimately enhancing the friction and wear performance and service life of marine engine cylinder liners. Attached Figure Description

[0038] Figure 1 This is a flowchart of a method for optimizing the microstructure and alloy composition of cast iron cylinder liners based on the prediction of friction and wear performance, according to the present invention.

[0039] Figure 2 This is a schematic diagram showing the microstructure characteristics of the cylinder liner material;

[0040] Figure 3It is a graph of the friction coefficients of the three wear tracks;

[0041] Figure 4 This is the outline of the first wear mark;

[0042] Figure 5 This is the outline of the second wear mark;

[0043] Figure 6 This is the outline of the third wear mark;

[0044] Figure 7 These are the determination coefficients obtained from the ET model's prediction of the friction coefficient;

[0045] Predicted COF represents the predicted coefficient of friction, while Experimental COF represents the experimentally measured coefficient of friction.

[0046] Figure 8 The coefficient of determination is obtained by the ET model for predicting wear rate;

[0047] The unit for wear rate is 10. -7 mm 3 ·N -1 ·m -1 Predicted wear rate refers to the predicted wear rate, while Experimental wear rate refers to the experimentally measured wear rate.

[0048] Figure 9 These are the determination coefficients obtained from the RF model's prediction of the friction coefficient;

[0049] Figure 10 These are the coefficients of determination obtained from the RF model for predicting wear rate;

[0050] Figure 11 The coefficient of determination is obtained by predicting the friction coefficient using the SVM model.

[0051] Figure 12 The coefficients of determination are obtained from the SVM model for predicting wear rate.

[0052] Figure 13 This is a comparison chart of the performance indicators of the three models in predicting the friction coefficient;

[0053] Evaluation criterion represents the index value;

[0054] Figure 14 This is a comparison chart of the performance indicators of the three models for predicting wear rate. Detailed Implementation

[0055] Specific implementation method one: Combining Figure 1This embodiment describes a method for optimizing the microstructure and alloy composition of cast iron cylinder liners based on tribological performance prediction. The method specifically includes the following steps:

[0056] Step 1: Select N types of cylinder liner materials for marine low-speed engines. The microstructure characteristics of the cylinder liner materials are as follows: Figure 2 As shown, each selected material is then cut to obtain a standard sample for each material; the standard samples obtained in this invention are all 30mm × 30mm × 6mm in size;

[0057] The standard samples were then polished using sandpaper and polishing compound. Specifically, the standard samples were polished with sandpaper of abrasive grits of 400#, 800#, 1500#, 2000# and 3000# in sequence, and then polished with polishing compounds of thicknesses of 2.5μm, 1.0μm and 0.5μm in sequence. After treating the standard samples of each material separately, the standard samples of each material were obtained.

[0058] Sanding with sandpaper can remove impurities and oxide layers from the sample surface, avoiding the influence of different sample surface roughness on the friction and wear test results. Polishing agent is then used to polish the sample. The polished sample is observed under a metallographic microscope until there are basically no scratches, and then the processed standard sample is obtained.

[0059] Step 2: Collect material characteristic data based on the processed standard sample. The material characteristic data includes metallographic data, hardness data, and the content data of each element. Among them, the metallographic data and hardness data are the microstructure characteristics of the material, and the content data of each element are the composition data of the alloy.

[0060] The metallographic data acquisition method is as follows: A 3×3 dot matrix is ​​selected on the surface of each processed standard sample, with a distance of 3 mm between adjacent points on the same surface. The selected points cannot be located within 1.5 mm of the sample edge to exclude invalid areas caused by mechanical polishing. The surface of the selected points is etched using a 4% nitric acid-alcohol solution for 10 seconds. Metallographic sampling at 100x magnification is performed at each etched point to obtain a metallographic photograph at each point. The collected metallographic photographs are then processed as follows:

[0061] For a metallographic image, the area ratio of graphite phase, pearlite matrix and high hardness phase in the metallographic image is statistically analyzed. The area ratio of graphite phase is taken as the content of graphite phase, the area ratio of pearlite matrix is ​​taken as the content of pearlite matrix, and the area ratio of high hardness phase is taken as the content of high hardness phase.

[0062] For a standard sample of a material after processing, the graphite phase content of 9 metallographic images corresponding to the standard sample is averaged, and the average value is taken as the graphite phase content of the material. The pearlite matrix content of 9 metallographic images corresponding to the standard sample is averaged, and the average value is taken as the pearlite matrix content of the material. The high hard phase content of 9 metallographic images corresponding to the standard sample is averaged, and the average value is taken as the high hard phase content of the material.

[0063] The hardness data is collected in the following way:

[0064] For a standard sample of a material after treatment, 100 points are selected on the surface of the standard sample for microhardness testing of the pearlite matrix. The load during measurement is 300g and the holding time is 10s. The pearlite matrix hardness data at the 100 points are then averaged and the average value is taken as the pearlite matrix hardness of the material. For the high-hardness phase, 100 points are selected on the surface of the standard sample after treatment for microhardness testing of the high-hardness phase. The high-hardness phase hardness data at the 100 points are then averaged and the average value is taken as the high-hardness phase hardness of the material.

[0065] The data collection method for the content of each element is as follows:

[0066] For a standard sample of a material after treatment, the standard sample is tested using inductively coupled plasma mass spectrometry (ICP-MS) (0.1-1.0g of sample is selected from the sample for testing) to obtain the alloy composition of the material, that is, the content of C, Si, Mn, P, S, Cu, Sn and Cr elements in the material.

[0067] It should be noted that for each standard sample after grinding, polishing, and etching, a 100x metallographic photograph is first taken under a metallographic microscope. This process does not damage the surface. Then, a microhardness test is performed. The microhardness test will leave a microhardness test indentation on the sample surface. However, the ICP-MS test only requires a very small amount of sample (about 0.1g) to measure the alloy composition. Therefore, the entire process can be completed using one standard specimen for each material. Finally, the sample is ground and polished again before a friction and wear test is performed.

[0068] Step 3: Collect friction and wear data on the processed standard samples, including the friction coefficient and wear rate;

[0069] The specific process of step three is as follows:

[0070] The coefficient of friction (COF) data of the sample is obtained through a friction and wear testing machine, such as... Figure 3As shown, the friction coefficient values ​​of each group were averaged, and the average value was used as the COF data for that group. The circular wear track generated by the friction and wear testing machine was divided into three equal parts according to its circumference to obtain three wear tracks. The three wear tracks were observed and measured using a white light interferometer to obtain the wear track morphology on the sample surface. Then, the wear track morphology on the sample surface was measured and analyzed using Gwyddion software, and the contours of the three wear tracks were measured. The contours of the three wear tracks are shown below. Figure 4 , Figure 5 and Figure 6 As shown, the average wear area of ​​the three wear marks is calculated, and then the wear volume of the entire circular wear mark is calculated based on the average wear area and the integral method.

[0071] The method for calculating the wear rate is as follows:

[0072] (1)

[0073] in, The wear volume of the material (in mm) 3 P is the applied load (in N), and L is the sliding distance (in m).

[0074] It should be noted that the friction and wear test was conducted using a Lanzhou Zhongke Kaihua HT-1000 high-temperature friction and wear testing machine. During the experiment, the sample was fixed to the machine chassis, and the friction method was ball-disc friction with a friction radius of 3 mm. The lower sample was loaded at the center of the chassis, and the upper sample was clamped eccentrically. After the experiment, the wear debris was collected, placed in a centrifuge tube, and a protective gas was introduced. GCr15 balls with a diameter of 6 mm and a hardness of 62 HRC were selected for the wear pair.

[0075] Step 4: Clean and standardize the raw data collected in Steps 2 and 3 to obtain standardized samples; specifically:

[0076] Step 4: Through the processing in Step 2 and Step 3, a set of data consisting of metallographic data, hardness data, content data of each element, friction coefficient and wear rate was obtained for the standard sample of each material. It was then determined whether there were any outliers in the data corresponding to each material (here, data that deviates significantly from the normal range, such as the value of a certain alloy component far exceeding the physically reasonable range). All data in the groups containing outliers were removed.

[0077] Step 42: Standardize the remaining data sets:

[0078] If the remaining data sets are normally distributed (or nearly normally distributed), then the standardization method used is Z-score standardization (i.e., mean 0 and standard deviation 1).

[0079] If the remaining data sets are distributed in other ways, the standardization method used is Min-Max standardization (mapping the feature values ​​to the [0,1] interval).

[0080] Standardization ensures that all feature values ​​are within the same scale range, thus eliminating the impact of feature scale differences on model performance.

[0081] Step 5: Divide the samples obtained in Step 4 into three parts according to a 6:2:2 ratio: training set, validation set, and test set. The training set is used for model training, the validation set is used for model parameter tuning and model selection, and the test set is used to evaluate the generalization performance of the model to ensure that the model performs reliably on unseen data.

[0082] Each model is trained, validated, and tested using the training set, validation set, and test set, and the best model is selected.

[0083] This invention selected three models for comparison: ET (Extremely Random Tree), RF (Random Forest), and SVM (Support Vector Machine). The ET model can significantly improve training speed and reduce variance while sacrificing a small amount of bias; the Random Forest model has strong robustness and anti-overfitting ability, and is suitable for processing medium and small-scale data; and the SVM performs well on small sample data and is suitable for processing linear or nonlinear problems.

[0084] The performance of the three models on the validation set was compared experimentally, and the best-performing model was selected as the final model. The selected model was trained using the training set, and its convergence was monitored. The algorithm and learning rate of the Adam optimizer were configured. During training, the loss values ​​of the model on the training and validation sets were recorded periodically to observe for overfitting. If the validation set performance continued to decline, training should be terminated promptly, and model parameters (such as regularization strength and learning rate) should be adjusted.

[0085] When evaluating the generalization performance of a model using a test set, the following metrics are used:

[0086] Mean Squared Error (RMSE): Measures the average squared error between the predicted and actual values. The smaller the RMSE, the more accurate the model prediction.

[0087] Coefficient of determination (R²): measures the model’s ability to explain changes in the data. The closer R² is to 1, the better the model fits the data.

[0088] The coefficient of determination obtained by the ET model for predicting the friction coefficient is as follows: Figure 7 As shown, the coefficient of determination obtained by the ET model for predicting wear rate is as follows: Figure 8As shown, the coefficient of determination obtained by the RF model for predicting the friction coefficient is as follows: Figure 9 As shown, the coefficient of determination obtained by the RF model for predicting wear rate is as follows: Figure 10 As shown, the coefficient of determination obtained by the SVM model for predicting the friction coefficient is as follows: Figure 11 As shown, the coefficient of determination obtained by the SVM model for predicting wear rate is as follows: Figure 12 As shown.

[0089] The optimal model is selected by comparing the RMSE and R² of the three models. When there is no model with the smallest mean squared error and the largest coefficient of determination, the model with the largest coefficient of determination is selected as the best model.

[0090] The performance indicators of the three models for predicting the friction coefficient are compared as follows: Figure 13 As shown, the performance indicators of the three models for predicting wear rate are compared as follows: Figure 14 As shown.

[0091] It should be noted that the input data for the model consists of the alloy composition data and microstructure characteristics of the material. The data is standardized before being input into the model. The output data of the model is the predicted values ​​of friction and wear performance (friction coefficient, wear rate). The model prediction results are compared with the experimental results. If the prediction results deviate significantly from the experimental results, the data preprocessing process needs to be reviewed. For example, for groups with outlier data, it is possible to choose not to remove them, but instead modify outliers exceeding the upper limit to the upper limit value and outliers below the lower limit to the lower limit value. In addition, sample data enhancement processing can be performed to avoid the problem of unbalanced sample numbers in the high / low composition regions.

[0092] Step Six: Using the selected optimal model, predict the friction coefficient and wear rate of the material to be optimized. Based on the prediction results, guide the optimization of the material's microstructure and alloy composition: Adjust the alloy composition design based on the model prediction results and feature importance analysis, for example, by increasing the proportion of highly important components to improve the material's wear resistance; adjust the material microstructure design, for example, by changing the heat treatment and casting methods to alter the material's microstructure characteristics, thereby improving the material's tribological performance and service life. Apply the optimized alloy composition and modified material microstructure to actual production to guide material preparation and performance improvement.

[0093] The above examples of the present invention are merely illustrative of the computational model and process of the present invention, and are not intended to limit the implementation of the present invention. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is impossible to exhaustively list all possible implementations here. Any obvious variations or modifications derived from the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A method for optimizing the microstructure and alloy composition of a cast iron cylinder liner based on friction and wear performance prediction, characterized by, The method specifically comprises the following steps: Step one, selecting N kinds of marine cylinder materials, and cutting each selected material to obtain a standard sample of each material; Then, polishing the standard sample with sandpaper and polishing agent to obtain the standard sample of each material after treatment; Step two, collecting material characteristic data according to the standard sample after treatment, wherein the material characteristic data comprises metallographic data, hardness data and content data of each element; Step three, collecting friction and wear data of the standard sample after treatment, wherein the friction and wear data comprises friction coefficient and wear rate; Step four, cleaning and standardizing the original data collected in steps two and three to obtain each sample after standardization; Step five, dividing the sample obtained in step four into three parts of training set, validation set and test set; Training, validating and testing each model by using the training set, validation set and test set respectively, and selecting the best model; Step six, predicting the friction coefficient and wear rate of the material to be optimized by using the selected best model, and guiding the optimization of material organization and alloy composition according to the prediction result.

2. The method for predicting the microstructure and alloy composition of a cast-iron cylinder liner based on friction and wear properties according to claim 1, characterized by, The size of the standard sample is 30 mm*30 mm*6 mm.

3. The method for predicting the microstructure and alloy composition of a cast-iron cylinder liner based on friction and wear properties according to claim 2, characterized by, The polishing of the standard sample with sandpaper and polishing agent is specifically: For any standard sample, the standard sample is polished with sandpaper with grits of 400#, 800#, 1500#, 2000# and 3000# in sequence, and then the sample is polished with polishing agent with thicknesses of 2.5 μm, 1.0 μm and 0.5 μm in sequence to obtain the standard sample after treatment; Then, each standard sample is treated to obtain the standard sample of each material after treatment.

4. The method for predicting the microstructure and alloy composition of a cast-iron cylinder liner based on friction and wear properties according to claim 3, characterized by, The metallographic data collection method is: On the surface of each standard sample after treatment, 3*3 dot matrix is selected, the distance between adjacent dots on the same surface is 3 mm, 4% nitric acid alcohol solution is used to corrode the surface of the selected dots, the corrosion time is selected to be 10 seconds, 100 times of metallographic sampling is performed at each corroded dot to obtain the metallographic photos collected at each dot; then each collected metallographic photo is processed, and the specific processing method is: For a metallographic photo, the area ratio of graphite phase, pearlite matrix and high-hard phase in the metallographic photo is counted, the area ratio of graphite phase is taken as the content of graphite phase, the area ratio of pearlite matrix is taken as the content of pearlite matrix, and the area ratio of high-hard phase is taken as the content of high-hard phase; For the standard sample after treatment of a kind of material, the graphite phase contents of the 9 metallographic photos corresponding to the standard sample are averaged, the obtained average value is taken as the graphite phase content of the material, the pearlite matrix contents of the 9 metallographic photos corresponding to the standard sample are averaged, the obtained average value is taken as the pearlite matrix content of the material, and the high-hard phase contents of the 9 metallographic photos corresponding to the standard sample are averaged, the obtained average value is taken as the high-hard phase content of the material.

5. The method for cast iron cylinder liner structure and alloy composition optimization based on friction and wear performance prediction according to claim 4, characterized in that, The hardness data collection method is: For a treated standard sample of a material, 100 points on the surface of the treated standard sample are selected for pearlite matrix microhardness testing, the pearlite matrix hardness data at the 100 points are averaged, the obtained average value is taken as the pearlite matrix hardness of the material, 100 points on the surface of the treated standard sample are selected for high-hardness phase microhardness testing, the high-hardness phase hardness data at the 100 points are averaged, and the obtained average value is taken as the high-hardness phase hardness of the material.

6. The method for cast iron cylinder liner structure and alloy composition optimization based on friction and wear performance prediction according to claim 5, characterized in that, The collection mode of the content data of each element is: For a treated standard sample of a material, the treated standard sample is tested by using an inductively coupled plasma mass spectrometer, and the alloy components in the material are obtained, that is, the contents of C element, Si element, Mn element, P element, S element, Cu element, Sn element and Cr element in the material are obtained.

7. The method for cast iron cylinder liner structure and alloy composition optimization based on friction and wear performance prediction according to claim 6, characterized in that, The calculation method of the wear rate is: （1） wherein, is the wear volume of the material, P is the external load, and L is the sliding distance.

8. The method for cast iron cylinder liner structure and alloy composition optimization based on friction and wear performance prediction according to claim 7, characterized in that, The specific process of step four is: Step four one, through the processing of step two and step three, the standard sample of each material obtains a group of data composed of metallographic data, hardness data, content data of each element, friction coefficient and wear rate, and whether there is an abnormal value in the data corresponding to each group of materials is judged, and the data of the group with the abnormal value is removed; Step four two, the remaining data of each group is standardized: If the distribution of the remaining data of each group is normal distribution, the standardization processing mode is Z-score standardization; If the distribution of the remaining data of each group is other distribution type, the standardization processing mode is Min-Max standardization.

9. The method for cast-iron cylinder liner structure and alloy composition optimization based on friction and wear performance prediction according to claim 8, characterized in that, The specific process of selecting the best model is: The mean square error and the determination coefficient of each model are compared respectively, and the model with the smallest mean square error and the largest determination coefficient is taken as the best model.

Images