A high-throughput component gradient-varying copper-based material, its preparation method and application

Abstract

The present invention provides a high-throughput component gradient copper-based material, a preparation method thereof, and an application thereof. The preparation method prepares a high-throughput component gradient copper-based material with components varying in a gradient along the axis through the pressing and sintering methods of traditional powder metallurgy. By combining the high-throughput component gradient copper-based material prepared by this preparation method with corresponding friction performance tests, data on component characteristics and friction performance can be obtained efficiently and in large quantities, changing the long-process and low-efficiency process of preparing one sample for each component and conducting one friction experiment in the past, and greatly reducing the development cycle of brake pad materials.

Description

Technical Field

[0001] The present invention belongs to the technical field of brake pads, and particularly relates to a high-throughput component gradient-varying copper-based material, a preparation method thereof, and an application thereof. Background Art

[0002] Copper-based brake pads are prepared by traditional powder metallurgy methods from substances such as copper matrix, hard friction components, and lubricating components, and have become the core key components of the braking systems of trains with a speed of 160 km / h or above. They can physically contact and abrade with the mating disc to generate frictional resistance, ensuring that the train stops within a safe distance. Copper-based brake pads have the following advantages: (1) heat resistance of the friction material; (2) stability of the friction coefficient; (3) anti-bonding property of the friction material; (4) wear resistance of the friction material; (5) strong thermophysical properties; (6) sufficient mechanical strength; (7) meeting environmental requirements; (8) being stable and having low noise during braking, and the friction pair can quickly recover its performance after periodic overload; (9) economy and process reliability during mass production. Currently, copper-based brake pads are equipped on various types of trains such as high-speed EMUs, intercity EMUs, and ordinary EMUs.

[0003] However, the copper-based brake pads used in many train models are also different because the operating environments and braking conditions of these trains often vary greatly, and the required friction coefficient ranges during braking are inconsistent. Therefore, the development of a specific train model often involves the composition design and development of a new generation of copper-based brake pad materials, which increases the development cycle and cost of the new train model. In addition, the components of copper-based brake pads often include more than a dozen composition components. For example, a pure copper matrix ensures excellent thermal conductivity and formability of the brake pad, hard wear-resistant components (such as iron, ferrochrome, and silicon carbide) improve the strength, wear resistance, and resistance between the friction interfaces of the matrix, and lubricating components (such as graphite and molybdenum disulfide) relieve the friction degree at the interface and smooth the friction surface to play a role in reducing wear and stabilizing the friction coefficient. Therefore, the previous methods for improving the performance of copper-based brake pads often optimize more than a dozen components through a large number of experiments. Although this can obtain data results with strong practical reference significance, it requires a large amount of time and resources. Therefore, there is an urgent need for a high-efficiency and low-cost development method for copper-based brake pads.

[0004] However, the preparation of multi-component copper-based brake pads by powder metallurgy and subsequent friction braking performance testing are complex processes involving multiple physical fields and multiple scales. Establishing an accurate relationship between composition, microstructure, and braking performance still faces huge challenges. The emergence of machine learning enables researchers to use computers as tools and strive to simulate the human learning method in real time, which provides a powerful tool for the rapid development, optimization, and iteration of materials. However, machine learning is data-based, and how to obtain sufficient and reliable experimental test data becomes a prerequisite for using machine learning methods.

[0005] In the preparation of traditional friction materials, the path of friction and wear performance testing is relatively inefficient. It is very urgent to provide a preparation method for developing high-throughput materials to obtain a large amount of reliable data through friction experiments. Summary of the Invention

[0006] Aiming at the deficiencies in the prior art, the present invention proposes a high-throughput component gradient-varying copper-based material, its preparation method, and application.

[0007] The technical solution of the present invention is as follows:

[0008] In the first aspect of the present invention, a preparation method of a high-throughput component gradient-varying copper-based material is provided. The preparation method includes: powder mixing treatment, powder pressing treatment, and sintering treatment; wherein, the step of the powder mixing treatment includes: preparing a mixed powder body with a component mass fraction varying along the axis in a powder mixing device; the powder mixing device includes a mold in the shape of a cube or a cuboid, and a baffle arranged along the diagonal direction of the mold, and the baffle is detachable; the baffle divides the mold into two right triangular prism regions, one of the right triangular prism regions is used to load matrix powder, and the other right triangular prism region is used to load a uniform mixture of one or more component powders to be screened; after loading is completed in the two right triangular prism regions, the baffle is removed, and the mold is made to perform a horizontal rotation movement along the central axis parallel to the side length direction, so that the matrix powder and the uniform mixture of one or more component powders to be screened are uniformly mixed only on the cross-section of the mold, and a mixed powder body with a component mass fraction varying along the axis is obtained.

[0009] Further, the matrix powder includes matrix pure copper powder; the uniform mixture of one or more component powders to be screened includes one or more mixtures of solid lubricating components and hard friction components; wherein, the range of the solid lubricating components includes one or more of graphite, metal sulfides, and metal fluorides; the range of the hard friction components includes one or more of hard metals and ceramics.

[0010] Further, in the mixed powder with the component mass fraction varying in a gradient along the axis, the contents of the following components are all in parts by mass: 5-20 parts of the graphite, 1-10 parts of the metal sulfide, 1-10 parts of the metal fluoride, 20-35 parts of the hard metal, and 0-4 parts of the ceramic; optionally, the graphite includes one or more of granular graphite, flaky graphite, and colloidal graphite; wherein, the particle size of the granular graphite is 75 μm - 180 μm, the particle size of the flaky graphite is 210 μm - 500 μm, and the particle size of the colloidal graphite is 1 μm - 10 μm; optionally, the particle size of the metal sulfide is 10 μm - 25 μm; optionally, the metal sulfide includes one or more of molybdenum disulfide, tungsten disulfide, iron sulfide, zinc sulfide, and copper sulfide; optionally, the particle size of the metal fluoride is 1 μm - 6.5 μm; optionally, the metal fluoride includes one or more of calcium fluoride, barium fluoride, lanthanum trifluoride, and cerium trifluoride; optionally, the powder particle size of the hard metal is 53 μm - 150 μm; optionally, the hard metal includes one or more of iron, chromium, and ferrochrome; optionally, the powder particle size of the ceramic is 0.5 μm - 2.6 μm; optionally, the ceramic includes one or more of silicon dioxide, silicon carbide, alumina, and molybdenum trioxide.

[0011] Further, the rotation speed of the powder mixing process is 50 r / min - 200 r / min, and the time is 0.5 h - 2 h; optionally, the powder pressing process includes placing the mixed powder with the component mass fraction varying in a gradient along the axis in a pressing mold of the same shape and size for pressing to obtain a green body; the pressing pressure is 150 MPa - 500 MPa, and the pressure holding time is 30 s - 120 s; optionally, the sintering process includes sintering the green body in an atmosphere hot pressing sintering furnace, the sintering temperature is 800 °C - 980 °C, the sintering pressure is 1 MPa - 3 MPa, the pressure holding sintering time is 1 h - 3 h, and the sintering atmosphere is H2.

[0012] The second aspect of the present invention provides a high-throughput component gradient-varying copper-based material, which is prepared by the preparation method of the high-throughput component gradient-varying copper-based material as described above; the composition of the high-throughput component gradient-varying copper-based material varies in a gradient along the axis.

[0013] The third aspect of the present invention provides the application of the high-throughput component gradient-varying copper-based material as described above in the fields of preparing brake pads and brake pad composition design.

[0014] Further, the method for designing the composition of the brake pad includes: Step 1, performing friction performance tests on the high-throughput component gradient-varying copper-based material to obtain a component feature-friction performance data set; Step 2, performing machine learning on the component feature-friction performance data set to generate and train a model to obtain a brake pad composition combination; wherein, Step 1 includes: performing linear reciprocating friction perpendicular to the component gradient direction on the high-throughput component gradient-varying copper-based material to obtain friction coefficient and stability data caused by the same material; performing linear reciprocating friction parallel to the component gradient direction on the high-throughput component gradient-varying copper-based material to obtain friction coefficient change data caused by the change in the component mass fraction; and obtaining the component feature-friction performance data set from the friction coefficient and stability data caused by the same material and the friction coefficient change data caused by the change in the component mass fraction.

[0015] Further, Step 1 is specifically carried out on a pin-on-disc friction testing machine, wherein the material of the disc in the pin-on-disc friction testing machine is the same as that of the high-throughput component gradient-varying copper-based material, and the material of the pin is one of cast iron, cast steel, and forged steel.

[0016] Further, Step 2 includes: preprocessing and feature selection on the component feature-friction performance data set to obtain a filtered training data set; performing standardization processing on the filtered training data set to obtain a standardized training data set; establishing and evaluating a model with the standardized training data set to obtain a trained standard prediction model; inputting the required conditional parameters and the desired friction coefficient value or stability range into the trained standard prediction model to obtain a brake pad composition combination; wherein, the conditional parameters include one or more of speed, pressure, and temperature; and the brake pad composition combination includes the type, mass fraction, and particle size of the components.

[0017] Furthermore, the step 2 specifically includes: using the component feature-friction performance data set to divide the machine learning training data set and the test data set, using the Pearson correlation coefficient to analyze the correlation between the feature data X, and removing redundant features with a correlation greater than 0.9; applying the ReliefF algorithm to further screen the features that have an important impact on the target attribute Y to obtain a screened training data set; wherein the feature data X includes the speed, pressure and temperature during the friction test, the type, mass fraction and particle size of the component, and the target attribute Y includes the friction coefficient value and stability; the screened training data set is standardized using the Z-score method to obtain a standardized training data set; random forest regression, artificial neural network, XGBoost regression, gradient boosting regression tree Four models are built, and the key parameters of each model are searched and optimized using the GridSearchCV automatic parameter adjustment tool, including the number of trees in the random forest, the hidden layer size and activation function of the artificial neural network, the learning rate of XGBoost, and the maximum depth of the tree; each model is trained using the standardized training data set and adjusted to the optimal parameter configuration; an evaluation standard for model prediction accuracy is constructed, and mean square error, root mean square error, and R² score are used as evaluation indicators; based on the evaluation results, a trained standard prediction model is obtained; the required conditional parameters and the expected friction coefficient value or stability range are input into the trained standard prediction model, and the model is used to predict the copper-based brake pad component combination that meets specific goals; wherein the specific goals include optimal friction performance.

[0018] Furthermore, the ratio of training dataset to test dataset is 6:4 to 8:2.

[0019] Compared with the prior art, the technical solution provided by the present invention has at least the following advantages:

[0020] The present invention provides a high-throughput component gradient-changing copper-based material and a preparation method and application thereof. The preparation method prepares a high-throughput component gradient-changing copper-based material whose components change gradiently along an axis by a traditional powder metallurgy pressing and sintering method. The high-throughput component gradient-changing copper-based material prepared by the preparation method is combined with corresponding friction performance tests to efficiently and in large quantities obtain component characteristics and friction performance data, changing the long and inefficient process of preparing a sample for each component and conducting a friction test. Secondly, the correlation between performance and composition is fully considered, and a machine learning prediction model for brake pad composition is established and trained from a data-driven perspective. The obtained training model is used to predict the new brake pad composition required to obtain the target friction coefficient size and its stability, greatly reducing the cycle of brake pad material development. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] One or more embodiments are illustrated by way of example in the accompanying drawings, which illustrations do not constitute a limitation of the embodiments, and unless otherwise stated, the figures in the drawings do not constitute a scale limitation.

[0022] Figure 1 Schematic structural diagram of the powder mixing device provided by the embodiment of the present invention;

[0023] Figure 2 Flow chart of the brake pad composition design method provided by the embodiment of the present invention;

[0024] Figure 3 Schematic diagram of the high-throughput component gradient change copper-based material sample and friction performance test provided by the embodiment of the present invention, where arrow 1 and arrow 2 in the figure respectively represent the directions of the friction experiments;

[0025] Figure 4 Copper-25 wt.% iron composite material with a gradient change in composition along the axis prepared by the embodiment of the present invention;

[0026] Figure 5 Graph showing the relationship between the friction coefficient and time during the friction experiment along the direction of continuous change of the copper-8 wt.% molybdenum disulfide composition provided by the embodiment of the present invention;

[0027] Figure 6 Graph showing the relationship between the friction coefficient and time obtained during the friction experiment along the direction of continuous change of the copper-25 wt.% iron composition provided by the embodiment of the present invention;

[0028] Figure 7 Graph showing the relationship between the friction coefficient and time obtained during the friction experiment along the direction of continuous change of the copper-22 wt.% chromium composition provided by the embodiment of the present invention;

[0029] Figure 8 Graph showing the relationship between the friction coefficient and time obtained during the friction experiment along the direction of continuous change of the copper-30 wt.% (graphite and CrFe) premixed powder composition provided by the embodiment of the present invention;

[0030] Figure 9 Graph of the average friction coefficient of the brake pad composition combination obtained by the brake pad composition design method based on high-throughput experiments and machine learning provided by the embodiment of the present invention. Detailed implementation manners

[0031] The present invention will be described in detail below in conjunction with the specific implementation manners, taking copper powder as the matrix powder as an example.

[0032] Embodiment

[0033] The present invention provides a method for preparing a high-throughput component gradient-varying copper-based material, and the preparation method includes: powder mixing treatment, powder pressing treatment, and sintering treatment. Among them, the steps of the powder mixing treatment include: preparing a mixed powder with the component mass fraction varying along the axis in a powder mixing device.

[0034] As Figure 1 shown, the powder mixing device includes a mold in the shape of a cube or a cuboid, and a baffle arranged along the diagonal direction of the mold, and the baffle is detachable; the baffle divides the mold into two right triangular prism regions, one of the right triangular prism regions is used for loading matrix powder, and the other right triangular prism region is used for loading a uniform mixture of one or more component powders to be screened.

[0035] After the two right triangular prism regions are loaded, the baffle is removed, and the mold is horizontally rotated along the central axis parallel to the side length direction (the rotation method is as shown by the arrow in Figure 1 ), so that the matrix powder and the uniform mixture of one or more component powders to be screened are only uniformly mixed on the cross-section of the mold, and no mixing occurs along the axis of the cubic mold, and a mixed powder with the component mass fraction varying along the axis is obtained.

[0036] Among them, the matrix powder includes matrix pure copper powder; the uniform mixture of one or more component powders to be screened includes one or more mixtures of solid lubricating components and hard friction components; among them, the range of the solid lubricating components includes one or more of graphite, metal sulfides, and metal fluorides; the range of the hard friction components includes one or more of hard metals and ceramics.

[0037] In a specific embodiment, graphite is 5 to 20 parts, metal sulfide is 1 to 10 parts, metal fluoride is 1 to 10 parts, hard metal is 20 to 35 parts, and ceramic is 0 to 4 parts.

[0038] Specifically, graphite includes one or more of granular graphite, flake graphite, and colloidal graphite; the particle size of the granular graphite is 75 μm to 180 μm, the particle size of the flake graphite is 210 μm to 500 μm, and the particle size of the colloidal graphite is 1 μm to 10 μm. The particle size of the metal sulfide is 10 μm to 25 μm; the metal sulfide includes one or more of molybdenum disulfide, tungsten disulfide, iron sulfide, zinc sulfide, and copper sulfide. The particle size of the metal fluoride is 1 μm to 6.5 μm; the metal fluoride includes one or more of calcium fluoride, barium fluoride, lanthanum trifluoride, and cerium trifluoride. The powder particle size of the hard metal is 53 μm to 150 μm; the hard metal includes one or more of iron, chromium, and ferrochrome. The powder particle size of the ceramic is 0.5 μm to 2.6 μm; optionally, the ceramic includes one or more of silicon dioxide, silicon carbide, aluminum oxide, and molybdenum trioxide.

[0039] More specifically, the rotation speed of the powder mixing process is 50 r / min to 200 r / min, and the time is 0.5 h to 2 h. The powder pressing process includes placing the mixed powder with the component mass fraction varying along the axis gradient in a pressing die of the same shape and size for pressing to obtain a green body; the pressing pressure is 150 MPa to 500 MPa, and the pressure holding time is 30 s to 120 s. The sintering process includes sintering the green body in an atmosphere hot pressing sintering furnace, the sintering temperature is 800 °C to 980 °C, the sintering pressure is 1 MPa to 3 MPa, the pressure holding and sintering time is 1 h to 3 h, and the sintering atmosphere is H2.

[0040] Such as Figure 2 , the embodiment of the present invention also provides a design method for the brake pad composition, specifically a design method for the brake pad composition based on high-throughput experiments and machine learning. This design method includes three main steps:

[0041] S1. Preparation of high-throughput component gradient change copper-based materials: Mix copper powder and another component powder in a mold to make a powder with continuously changing composition, and then make the required high-throughput component gradient change copper-based material sample through two steps of pressing and hot pressing sintering.

[0042] S2. Obtain the component feature-friction performance data set through high-throughput friction experiments: Use a linear reciprocating friction testing machine, and the pin conducts friction experiments along the direction of continuous change or uniform direction of the sample composition.

[0043] S3. Use machine learning methods for composition design: Select the components required for the copper-based brake pad, and then establish and train a machine learning model using the corresponding experimental data set. Input the desired friction coefficient value or stability range, and use the finally optimized model to predict the copper-based brake pad composition combination that meets specific goals.

[0044] In a specific embodiment, the component to be screened is a hard friction component, specifically iron as an example. In step S1, in the copper-iron mixed powder, the proportion of iron added by mass fraction is 25%, the particle size of the iron powder is 75 μm, the rotation speed of the mold is 80 r / min, and the mixing time is 1 h. Transfer this powder to a pressing die of the same shape and size for pressing, the pressing pressure is 400 MPa, and the pressure holding time is 80 s. Transfer the pressed green body to an atmosphere hot pressing sintering furnace for sintering, the sintering temperature is 940 °C, the sintering pressure is 2 MPa, the pressure holding and sintering time is 2 h, and the sintering atmosphere is H2. After sintering, continuously introduce the atmosphere until the material cools to room temperature. The prepared sample is as Figure 4 .

[0045] In a specific embodiment, the component to be screened is a solid lubricant component, specifically molybdenum disulfide as an example. In step S1, in the copper-molybdenum disulfide mixed powder, the proportion of molybdenum disulfide added by mass fraction is 8%, the particle size of iron powder is 18 μm, the rotation speed of the mold is 50 r / min, and the mixing time is 1.5 h. Transfer this powder to a pressing mold with the same shape and size for pressing, the pressing pressure is 150 MPa, and the pressure holding time is 70 s. Transfer the pressed green body to an atmosphere hot pressing sintering furnace for sintering, the sintering temperature is 800 °C, the sintering pressure is 1 MPa, the pressure holding and sintering time is 1 h, and the sintering atmosphere is H2. After sintering is completed, continuously introduce the atmosphere until the material cools to room temperature.

[0046] In a specific embodiment, the component to be screened is a hard friction component, specifically chromium as an example. In step S1, in the copper-chromium mixed powder, the proportion of chromium added by mass fraction is 22%, the particle size of chromium powder is 106 μm, the rotation speed of the mold is 100 r / min, and the mixing time is 0.5 h. Transfer this powder to a pressing mold with the same shape and size for pressing, the pressing pressure is 500 MPa, and the pressure holding time is 120 s. Transfer the pressed green body to an atmosphere hot pressing sintering furnace for sintering, the sintering temperature is 920 °C, the sintering pressure is 2.5 MPa, the pressure holding and sintering time is 3 h, and the sintering atmosphere is H2. After sintering is completed, continuously introduce the atmosphere until the material cools to room temperature.

[0047] In a specific embodiment, the component to be screened is a uniform mixture of a solid lubricant component and a friction component, specifically a pre-mixed powder of graphite and CrFe in a weight ratio of 1:1 as an example, where the particle size of flaky graphite is 350 μm and the particle size of CrFe is 105 μm. In step S1, in the copper-pre-mixed powder, the proportion of the pre-mixed powder added by mass fraction is 30%, the rotation speed of the mold is 200 r / min, and the mixing time is 2 h. Transfer this powder to a pressing mold with the same shape and size for pressing, the pressing pressure is 320 MPa, and the pressure holding time is 30 s. Transfer the pressed green body to an atmosphere hot pressing sintering furnace for sintering, the sintering temperature is 980 °C, the sintering pressure is 3 MPa, the pressure holding and sintering time is 2 h, and the sintering atmosphere is H2. After sintering is completed, continuously introduce the atmosphere until the material cools to room temperature.

[0048] In step S2, it includes: performing linear reciprocating friction perpendicular to the composition gradient direction on the high-throughput component gradient-varying copper-based material to obtain the friction coefficient and stability data caused by the same material; performing linear reciprocating friction parallel to the composition gradient direction on the high-throughput component gradient-varying copper-based material to obtain the friction coefficient change data caused by the change in component mass fraction; and obtaining the component characteristic-friction performance data set from the friction coefficient and stability data caused by the same material and the friction coefficient change data caused by the change in component mass fraction. As Figure 3As indicated by the arrows 1 and 2. The material compositions of Path 1 are the same, so the friction coefficients and their stabilities caused by the same material can be obtained. The material compositions of Path 2 show continuous changes, and the changes in friction coefficients caused by the changes in the mass fractions of components can be obtained at one time, thus establishing a complete dataset of component characteristics - friction properties.

[0049] Step S2 is specifically carried out on a pin-on-disk friction testing machine. Among them, the material of the disk in the pin-on-disk friction testing machine is the same as that of the high-throughput component gradient-varied copper-based material, and the material of the pin is one of cast iron, cast steel, and forged steel.

[0050] In a specific embodiment, in Step S2, a reciprocating friction experiment is carried out along the continuous change direction of copper - 8 wt.% molybdenum disulfide. The counterpiece is a pin made of forged steel, and the time is 30 seconds. The change of the friction coefficient with time is as Figure 5 . The reciprocating friction process brings about periodic changes in the friction coefficient. The continuous increase in the content of molybdenum disulfide makes the friction coefficient decrease from about 0.24 to about 0.05.

[0051] In a specific embodiment, in Step S2, a reciprocating friction experiment is carried out along the continuous change direction of copper - 25 wt.% iron. The counterpiece is a pin made of forged steel, and the time is 30 seconds. The change of the friction coefficient with time is as Figure 6 . The reciprocating friction process brings about periodic changes in the friction coefficient. The continuous increase in the iron content makes the friction coefficient increase from about 0.1 to about 0.3 - 0.35, and the fluctuation of the friction coefficient is larger than that of the lubricating component molybdenum disulfide.

[0052] In a specific embodiment, in Step S2, a reciprocating friction experiment is carried out along the continuous change direction of copper - 22 wt.% chromium. The counterpiece is a pin made of forged steel, and the time is 30 seconds. The change of the friction coefficient with time is as Figure 7 . The reciprocating friction process brings about periodic changes in the friction coefficient. The continuous increase in the chromium content makes the friction coefficient increase from about 0.13 to about 0.5 - 0.55, resulting in a large fluctuation of the friction coefficient.

[0053] In a specific embodiment, in Step S2, a reciprocating friction experiment is carried out along the continuous change direction of copper - 30 wt.% (graphite and CrFe) premixed powder composition. The counterpiece is a pin made of forged steel, and the time is 30 seconds. The change of the friction coefficient with time is as Figure 8 . The reciprocating friction process brings about periodic changes in the friction coefficient. The continuous increase in the graphite and CrFe mixed powder makes the friction coefficient increase from about 0.03 to about 0.23, and the fluctuation of the friction coefficient is significantly reduced compared with that when only friction components are added.

[0054] In a specific embodiment, in step S3, the data preprocessing and feature selection process is as follows: using high-throughput experimental data to divide the machine learning training data set and the test data set, and the ratio between the two is 7:3. Use the Pearson correlation coefficient to analyze the correlation between the feature data X (speed, pressure and temperature during the friction experiment, types, mass fractions and particle sizes of components), remove redundant features with a correlation greater than 0.9, and use the standard score (Z-score) method to standardize the filtered training data set. Select four models for modeling: random forest regression, artificial neural network, extreme gradient boosting (XGBoost) regression, and gradient boosting regression tree (GBRT). Use the grid search cross-validation (GridSearchCV) automatic parameter tuning tool to search and optimize the key parameters of each model, including the number of trees in the random forest, the hidden layer size and activation function of the artificial neural network, the learning rate of XGBoost, and the maximum depth of the tree. Use the training data set to train each model and adjust it to the optimal parameter configuration. Construct an evaluation criterion for the prediction accuracy of the model, and use the root mean square error and R² score as evaluation indicators. The smaller the root mean square and the closer R 2 is to 1, the higher the prediction accuracy of the model. Based on the evaluation results, select the model with the highest prediction accuracy as the final model. Input the required conditional parameters (speed, pressure), as well as the desired friction coefficient value or stability range. Use the model to predict the composition combination of the copper-based brake pad that meets specific objectives, including the types, mass fractions and particle sizes of components.

[0055] The components of the copper-based brake pad are screened out by the above method, as shown in Table 1.

[0056] Table 1 Composition of the Predicted Copper-Based Brake Pad

[0057] Component Cu Fe CrFe Cr Granular graphite Flake graphite <![CDATA[MoS2]]> <![CDATA[Al2O3 fiber]]> <![CDATA[SiO2]]> Content 55 18 6 2 3 9 3 3 1

[0058] Conduct a simulated braking experiment with a braking speed range of 50 - 350 km / h on the samples prepared according to the components designed by machine learning, as Figure 9 . The dotted line is the upper and lower limits of the average friction coefficient specified in the "Interim Technical Conditions for Brake Pads of Multiple Unit Trains" (TJ / CL 307 - 2019). The results show that the prepared brake pads have excellent friction coefficient magnitude and stability under various braking conditions.

[0059] Those of ordinary skill in the art can understand that the above embodiments are specific embodiments for implementing the present application. In actual applications, various changes can be made in form and details without departing from the spirit and scope of the present application. Any person skilled in the art can make their respective changes and modifications without departing from the spirit and scope of the present application. Therefore, the protection scope of the present application should be subject to the scope defined by the claims.

Claims

1. A method for preparing a high-throughput component gradient-changing copper-based material, characterized in that: The preparation method comprises: powder mixing process, powder pressing process and sintering process; The step of powder mixing treatment includes: preparing a mixed powder in which the mass fraction of the components changes gradiently along the axis in a powder mixing device; The powder mixing device comprises a cube or rectangular parallelepiped mold, and a baffle arranged along the diagonal direction of the mold, and the baffle is detachable; the baffle divides the mold into two right triangular prism areas, one of the right triangular prism areas is used to load the matrix powder, and the other right triangular prism area is used to load a uniform mixture of one or more component powders that need to be screened; After the two straight triangular prism regions are loaded, the baffle is removed, and the mold is horizontally rotated along the central axis parallel to the length direction, so that the uniform mixture of the matrix powder and one or more component powders to be screened is uniformly mixed only on the cross section of the mold to obtain a mixed powder with a component mass fraction gradiently changing along the axis; Among them, the high-throughput component gradient change copper-based material is used for friction performance testing, and the friction performance test includes performing linear reciprocating friction perpendicular to the component gradient direction to obtain the friction coefficient and stability data caused by the same material; and performing linear reciprocating friction parallel to the component gradient direction to obtain the friction coefficient change data caused by the change in component mass fraction; the component characteristic-friction performance data set is obtained from the friction coefficient and stability data caused by the same material and the friction coefficient change data caused by the change in the component mass fraction.

2. The method for preparing a high-throughput component gradient-changing copper-based material according to claim 1, characterized in that: The matrix powder includes matrix pure copper powder; The uniform mixture of one or more component powders to be screened includes one or more of solid lubricating components and hard friction components; wherein the solid lubricating component range includes one or more of graphite, metal sulfide and metal fluoride; The hard friction component range includes one or more of hard metals and ceramics.

3. The method for preparing a high-throughput component gradient-changing copper-based material according to claim 2, characterized in that: In the mixed powder in which the mass fraction of the components changes along the axis gradient, the contents of the following components are all in mass fractions: The graphite is 5 to 20 parts, the metal sulfide is 1 to 10 parts, the metal fluoride is 1 to 10 parts, the hard metal is 20 to 35 parts, and the ceramic is 0 to 4 parts; The graphite includes one or more of granular graphite, flake graphite and colloidal graphite; wherein the particle size of the granular graphite is 75 μm to 180 μm, the particle size of the flake graphite is 210 μm to 500 μm, and the particle size of the colloidal graphite is 1 μm to 10 μm; The particle size of the metal sulfide is 10 μm to 25 μm; the metal sulfide includes one or more of molybdenum disulfide, tungsten disulfide, ferrous sulfide, zinc sulfide and copper sulfide; The particle size of the metal fluoride is 1 μm to 6.5 μm; the metal fluoride includes one or more of calcium fluoride, barium fluoride, lanthanum trifluoride and cerium trifluoride; The particle size of the hard metal powder is 53 μm to 150 μm; the hard metal includes one or more of iron, chromium and chromium iron; The powder particle size of the ceramic is 0.5 μm to 2.6 μm; the ceramic includes one or more of silicon dioxide, silicon carbide, aluminum oxide and molybdenum trioxide.

4. The method for preparing a high-throughput component gradient-changing copper-based material according to claim 1, characterized in that: The powder mixing process is performed at a rotation speed of 50 r / min to 200 r / min for a time of 0.5 h to 2 h; The powder pressing process comprises placing the mixed powder whose mass fraction of the components changes gradually along the axis into a pressing mold of the same shape and size for pressing to obtain a green body; the pressing pressure is 150MPa to 500MPa, and the holding time is 30s to 120s; The sintering process comprises sintering the green body in an atmosphere hot pressing sintering furnace, the sintering temperature is 800° C. to 980° C., the sintering pressure is 1 MPa to 3 MPa, the pressure holding sintering time is 1 h to 3 h, and the sintering atmosphere is H2.

5. A high-throughput component gradient-changing copper-based material, characterized in that: The high-flux component gradient-changing copper-based material is prepared according to the preparation method of the high-flux component gradient-changing copper-based material as described in any one of claims 1 to 4; the composition of the high-flux component gradient-changing copper-based material changes gradiently along the axis.

6. Application of the high-flux component gradient-changing copper-based material as claimed in claim 5 in the field of preparing brake pads and designing brake pad components.

7. The use according to claim 6, characterized in that: The method for designing the brake pad composition includes: Step 1: testing the friction performance of the high-throughput component gradient-changing copper-based material to obtain a component characteristic-friction performance data set; Step 2: Perform machine learning on the component feature-friction performance data set, generate a model, and train it to obtain a brake pad component combination; Wherein, the step 1 comprises: The high-flux component gradient-changing copper-based material is subjected to linear reciprocating friction perpendicular to the component gradient direction to obtain friction coefficient and stability data caused by the same material; The high-flux component gradient-changing copper-based material is subjected to linear reciprocating friction parallel to the component gradient direction to obtain friction coefficient change data caused by component mass fraction change; The component characteristic-friction performance data set is obtained from the friction coefficient and stability data caused by the same material and the friction coefficient change data caused by the change in the mass fraction of the component.

8. The use according to claim 7, characterized in that: The step 1 is specifically carried out on a pin-disc friction testing machine, wherein the material of the disk in the pin-disc friction testing machine is consistent with the material of the high-throughput component gradient-changing copper-based material, and the material of the pin is one of cast iron, cast steel and forged steel.

9. The use according to claim 7, characterized in that: The second step comprises: Preprocessing and feature selecting the component feature-friction performance data set to obtain a screened training data set; Standardizing the screened training data set to obtain a standardized training data set; Establishing and evaluating a model using the standardized training data set to obtain a trained standard prediction model; The required condition parameters and the expected friction coefficient value or stability range are input into the trained standard prediction model to obtain a brake pad component combination; wherein the condition parameters include one or more of speed, pressure and temperature; and the brake pad component combination includes the type, mass fraction and particle size of the components.

10. The use according to claim 9, characterized in that: The step 2 specifically includes: The component feature-friction performance data set is used to divide the machine learning training data set and the test data set, and the Pearson correlation coefficient is used to analyze the correlation between the feature data X, and the redundant features with a correlation greater than 0.9 are removed; the ReliefF algorithm is used to further screen the features that have an important impact on the target attribute Y to obtain the screened training data set; wherein the feature data X includes the speed, pressure and temperature during the friction test, the type, mass fraction and particle size of the component, and the target attribute Y includes the friction coefficient value and stability; The screened training data set is standardized using a Z-score method to obtain a standardized training data set; Four models, random forest regression, artificial neural network, XGBoost regression, and gradient boosting regression tree, were selected for modeling. The GridSearchCV automatic parameter adjustment tool was used to search and optimize the key parameters of each model, including the number of trees in the random forest, the size of the hidden layer and activation function of the artificial neural network, the learning rate of XGBoost, and the maximum depth of the tree. Each model was trained using the standardized training data set and adjusted to the optimal parameter configuration. An evaluation standard for model prediction accuracy was constructed, using mean square error, root mean square error, and R 2 The score is used as an evaluation indicator; based on the evaluation results, a trained standard prediction model is obtained; After the trained standard prediction model is input with the required condition parameters and the expected friction coefficient value or stability range, the model is used to predict the copper-based brake pad component combination that meets specific goals; wherein the specific goals include optimal friction performance.

Images