Material data processing method and system

Abstract

The invention relates to the field of automobile part data processing, in particular to a material data processing method and system. The method comprises the following steps: acquiring material data of a plurality of automobile parts; and clustering the material data to obtain a final cluster, and managing the material data of the automobile parts according to the final cluster. According to the technical scheme, the accuracy of material data clustering of the automobile parts can be improved, and the precision of a data processing result is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of automobile parts data processing. More specifically, the present application relates to a material data processing method and system. BACKGROUND

[0002] With the rapid development of the automotive industry, the material data processing method and system of automobile parts become increasingly important. Automobile manufacturing involves the production and assembly of thousands of parts, and ensuring the quality and supply chain management of parts is crucial for the performance, safety and cost control of the whole vehicle. In order to cope with these challenges, more and more automobile manufacturers and parts suppliers begin to introduce advanced technologies such as Internet of Things, artificial intelligence and big data analysis, and build intelligent material data processing systems. These systems not only improve the work efficiency and supply chain transparency, but also help enterprises maintain an advantage in fierce market competition and promote the progress towards lean production and continuous innovation.

[0003] The existing patent application file with the publication number CN108733003A discloses a rotary part process time prediction method and system based on kmeans clustering algorithm, wherein the application file adopts an initial cluster center calculation method, a distance calculation method and an update cluster center method to obtain clustering results and cluster centers under different K values, calculates SP / DB indexes under different K values, takes the K value corresponding to the minimum SP / DB, the cluster center and the clustering result, which is the final K value, the final clustering result and the final cluster center.

[0004] However, when analyzing the material data of automobile parts by kmeans clustering, there may be many deviations in the data collection process, and the data with large deviation degree will cause the center point in the clustering algorithm to deviate, thereby affecting the clustering result and further leading to inaccurate data processing result. SUMMARY

[0005] To solve the problem of inaccurate data processing result, the present application proposes a material data processing method and system.

[0006] In a first aspect, the present invention discloses a material data processing method, comprising: acquiring material data of a number of automobile parts; clustering the material data to obtain a final clustering cluster, and managing the material data of the automobile parts according to the final clustering cluster; wherein obtaining the final clustering comprises: presetting the number of cluster center points, obtaining the position of each cluster center point, taking any material data as a target point, and for any iteration, calculating the degree of discreteness according to a preset neighborhood range of the target point, and calculating the attribution possibility of the target point in any iteration; taking the ratio of the attribution possibility to the degree of discreteness as the influence weight of the target point, traversing to obtain the influence weight of each material data, updating the position of the cluster center point, and obtaining the final clustering cluster in response to the iteration satisfying the stopping condition.

[0007] Cluster analysis is used to manage automotive parts material data. The influence weights of target points are calculated based on their likelihood of belonging and their degree of discreteness, effectively enabling adaptive adjustment of data points to cluster centers. In each iteration, material data is dynamically evaluated based on the discreteness and likelihood of belonging of its neighborhood, ensuring that cluster centers gradually align with the true distribution of the data. This weighted adjustment approach results in clusters that more accurately reflect the actual characteristics of automotive parts material data, enabling more precise classification and optimization in material management. This improves efficiency in inventory management, production scheduling, and quality control, while reducing resource waste and potential management errors.

[0008] Preferably, obtaining the position of each cluster center point includes: taking the center point of all material data as the first cluster center point, taking the material data farthest from the first cluster center point as the first reference point, selecting the midpoint between the first cluster center point and the reference point as the second cluster center point, obtaining the third cluster center point based on the first cluster center point and the second cluster center point, and similarly obtaining all cluster center points; in response to the number of cluster center points reaching the preset number of cluster center points, obtaining the position of each cluster center point.

[0009] It not only improves the accuracy of clustering results, but also helps to better classify, analyze and control the distribution of materials.

[0010] Preferably, obtaining the third cluster center point includes: taking the material data with the largest sum of distances from the first cluster center point and the second cluster center point as the second reference point; and selecting the midpoint of the first cluster center point, the second cluster center point and the second reference point as the third cluster center point.

[0011] Preferably, the discrete degree is obtained by: obtaining a neighborhood range of the target point, taking the material data in the neighborhood range except the target point as first neighborhood points, and taking the neighborhood points of the cluster center point of the cluster to which the target point belongs as second neighborhood points; calculating the absolute difference value between the number of the first neighborhood points and the number of the second neighborhood points, and taking the ratio of the absolute difference value to the number of the second neighborhood points as the discrete degree.

[0012] By calculating the difference value and the ratio of the number of the first neighborhood points and the number of the second neighborhood points, the stability of the target point in the cluster and the deviation degree from the cluster center can be quantified. Such a calculation method helps to identify abnormal data points deviating from the cluster center, reduces their influence on the clustering result, and thus improves the accuracy and robustness of the clustering result. Not only can the clustering process be optimized, but also the distribution characteristics of the material data in actual application can be better reflected.

[0013] Preferably, the discrete degree is obtained by: obtaining a neighborhood range of the target point, taking the material data in the neighborhood range except the target point as first neighborhood points, and taking the neighborhood points of the cluster center point of the cluster to which the target point belongs as second neighborhood points; taking the range of the material data in the first neighborhood points as a first range, and taking the range of the material data in the second neighborhood points as a second range, calculating the absolute difference value between the first range and the second range, and taking the ratio of the absolute difference value to the second range as the discrete degree.

[0014] By calculating the discrete degree of the target point and the material data in its neighborhood, the differences between the material data in different regions are quantified, which can effectively reveal the change and fluctuation trend of the material data.

[0015] Preferably, the belonging possibility satisfies the following relationship:

[0016] G m,i represents the belonging possibility of the target point i after the mth iteration, d m,i represents the Euclidean distance between the cluster center point of the cluster to which the target point i belongs and the target point i after the mth iteration, represents the mean value of the Euclidean distances between all the material data in the cluster to which the target point i belongs and the cluster center point after the mth iteration, E m,i,j and E m-1,i,j respectively represent the value of the cluster center point of the cluster to which the target point i belongs in the jth dimension after the mth iteration and after the (m-1)th iteration, D i,j represents the value of the target point i in the jth dimension, M represents the total number of iterations, and J represents the total number of dimensions of the material data.

[0017] The formula for the probability of belonging is optimized through multiple iterative adjustments of the clustering results. Specifically, as the iterations proceed, each iteration calculates the distance between the target point and the current cluster center, and dynamically adjusts the probability of belonging based on the position of the target point within the cluster and the changes in the cluster center.

[0018] Preferably, the stopping condition is that after two adjacent iterations are completed, the material data in the cluster does not change.

[0019] In a second aspect, the present invention discloses a material data processing system, comprising: a processor; and a memory, wherein the memory stores computer instructions, and when the computer instructions are executed by the processor, the system executes the above-mentioned material data processing method.

[0020] Beneficial effects of the present invention:

[0021] By presetting the number of cluster centers and iteratively updating their positions, the present invention gradually optimizes clustering results, ensuring that each material data point is rationally distributed based on its distance from the cluster center and its neighborhood characteristics. By calculating influence weights, the importance of each material data point in the clustering process is effectively reflected, and the positions of cluster centers are continuously optimized through iteration to achieve a stable clustering effect. This clustering-based processing method improves the accuracy of material data classification and effectively enhances the efficiency and quality of material data processing. Furthermore, by measuring the degree of discreteness, the flexibility and adaptability of the clustering process are further enhanced, preventing deviations in clustering results. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood by reading the following detailed description with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present invention are shown in an illustrative and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0023] Figure 1 The present invention is a flowchart of a material data processing method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work shall fall within the scope of protection of the present invention.

[0025] It should be understood that when the claims, the specification, and the drawings of the present application use terms such as "first", "second", etc., these terms are only used to distinguish different objects, and are not used to describe a specific order. The terms "include" and "contain" used in the specification and claims of the present application indicate the presence of the described features, whole, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, whole, steps, operations, elements, components and / or sets thereof.

[0026] The present application provides a material data processing method. As shown in Figure 1 The material data processing method comprises steps S1-S2, which are described in detail below.

[0027] S1, obtaining material data of a plurality of automobile parts.

[0028] It should be noted that the quality evaluation standards of parts in different positions of the vehicle will be different, which will be determined according to the function of the parts in the vehicle and the use environment. For example, engine-related components usually need to take torque, strength and high-temperature performance as the main standards, because these components bear greater pressure and high-temperature environment; while the body structure components such as doors, windows, etc. may pay more attention to impact resistance, corrosion resistance and tensile strength to ensure safety and durability; some electrical components may focus more on electrical performance and cold and heat resistance to meet the use requirements under different climate conditions.

[0029] In one embodiment, the material data of the automobile parts is collected, which can be single-dimensional data or multi-dimensional data. For example, the single-dimensional data is length or weight or torque. The multi-dimensional data is length and weight, or length and torque, or length, weight and torque.

[0030] S2, clustering the material data to obtain a final clustering cluster, and managing the material data of the automobile parts according to the final clustering cluster.

[0031] It should be noted that when the material data of the automobile parts is clustered by the Kmeans algorithm, the abnormal data may have a great influence on the calculation of the cluster center point, causing the center point to deviate from the true position of the clustered data points, so that the final clustering result is inaccurate. In order to solve this problem, the dispersion degree of each material data can be calculated to measure its stability in the clustering cluster, so as to determine its attribution degree. The material data with higher attribution degree has greater contribution to the cluster center point, while the material data with lower attribution degree (i.e. abnormal data) has relatively smaller influence. Based on this, a weight value can be assigned to each material data. On this basis, the cluster center point is recalculated by weighted average to improve the accuracy of clustering.

[0032] In one embodiment, the number of cluster center points is preset, and the position of each cluster center point is obtained. This includes: taking the center point of all material data as the first cluster center point, taking the material data farthest from the first cluster center point as the first reference point, selecting the midpoint between the first cluster center point and the reference point as the second cluster center point, and obtaining the third cluster center point based on the first cluster center point and the second cluster center point. Obtaining the third cluster center point includes:

[0033] The material data with the largest sum of distances from the first and second cluster centers is selected as the second reference point. The midpoint between the first, second, and second reference points is selected as the third cluster center. The third cluster center is actually the midpoint of the triangle formed by the first, second, and second cluster centers.

[0034] The material data with the largest sum of distances from the first, second, and third cluster centers is selected as the third reference point. The midpoint of the first, second, third, and third reference points is selected as the fourth cluster center. The fourth cluster center is actually the midpoint of the quadrilateral formed by the first, second, third, and third cluster centers, and the third reference point.

[0035] Similarly, all cluster centers are obtained.

[0036] In response to the number of cluster center points reaching a preset number of cluster center points, the position of each cluster center point is obtained.

[0037] Take any material data as the target point. For any iteration, obtain the neighborhood range of the target point, take the material data other than the target point in the neighborhood range as the first neighborhood point, and take the neighborhood point of the cluster center point of the cluster to which the target point belongs as the second neighborhood point; calculate the absolute difference between the number of the first neighborhood points and the number of the second neighborhood points, and take the ratio of the absolute difference to the number of the second neighborhood points as the degree of discreteness.

[0038] By calculating the difference and ratio between the number of first and second neighboring points, we can quantify the stability of the target point within the cluster and its degree of deviation from the cluster center. This calculation method helps identify anomalous data points that deviate from the cluster center, reducing their impact on the clustering results, thereby improving the accuracy and robustness of the clustering results. This not only optimizes the clustering process but also better reflects the distribution characteristics of material data in real applications.

[0039] Exemplarily, the neighborhood range is set to the result of mapping the dimension number of the material data through an exponential function, and the specific value can be set by those skilled in the art.

[0040] Calculate the probability of the target point belonging in any iteration. The probability of belonging satisfies the relationship:

[0041] G m,i represents the possibility of belonging of target point i after the mth iteration, d m,i It represents the Euclidean distance between the cluster center of the cluster to which the target point i belongs and the target point i after the mth iteration. It represents the mean of the Euclidean distances between all material data and the cluster center in the cluster to which the target point i belongs after the mth iteration, E m,i,j and E m-1,i,j Respectively represent the value of the cluster center point of the cluster to which the target point i belongs in dimension j after the mth iteration and the m-1th iteration, D i,j Represents the value of target point i in dimension j, M represents the total number of iterations, and J represents the total number of dimensions of material data.

[0042] The probability of belonging formula is refined through multiple iterations. Specifically, as each iteration progresses, the distance between the target point and the current cluster center is calculated. The probability of belonging is dynamically adjusted based on the target point's position within the cluster and the changes in the cluster center. This process captures the changes in the target point's belonging across iterations, reflecting the stability of the target point relative to the cluster center. As the number of iterations increases, the cluster center gradually stabilizes, and the accuracy of the clustering results improves accordingly.

[0043] The ratio of the likelihood of belonging to the target point to the degree of dispersion is used as the influence weight of the target point. The influence weight of each material data point is obtained through iteration. The position of the cluster center point is updated based on the influence weight. The final cluster is obtained when the iteration meets the stopping condition. The stopping condition is that the material data within the cluster has not changed after two consecutive iterations.

[0044] Specifically, during each iteration, each item data point is first assigned a weight, known as an influence weight, based on the current clustering. This weight reflects the influence of that item data point within the cluster. Next, based on the influence weight of each item data point, a weighted average of all data points within the same cluster is calculated to obtain the weighted cluster center. Then, with each iteration, the influence weight of each item data point is recalculated, and the weighted cluster center is updated based on the new influence weight. This continuous adjustment of the cluster center position results in a more stable and accurate final cluster.

[0045] Based on the final clusters, we can extract the material data for auto parts belonging to the same category, and then conduct in-depth group analysis on the materials in each category. By analyzing the characteristics of the material data in each category, combined with historical data and quality standards, we can predict the quality performance of each group of auto parts.

[0046] In another embodiment, the degree of discreteness also includes: obtaining the neighborhood range of the target point, taking the material data other than the target point within the neighborhood range as the first neighborhood point, and taking the neighborhood point of the cluster center point of the cluster to which the target point belongs as the second neighborhood point; taking the range of the material data in the first neighborhood point as the first range, and the range of the material data in the second neighborhood point as the second range, calculating the absolute difference between the first range and the second range, and taking the ratio of the absolute difference to the second range as the degree of discreteness.

[0047] By calculating the discrete degree of material data in the target point and its neighborhood, the differences in material data between different areas are quantified, which can effectively reveal the changes and fluctuation trends of material data.

[0048] An embodiment of the present invention further discloses a material data processing system, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, a material data processing method according to the present invention is implemented.

[0049] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are known in the art and will not be described in detail here.

[0050] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium may be any suitable magnetic storage medium or magneto-optical storage medium, such as resistive random access memory (RRAM), dynamic random access memory (DRAM), static random access memory (SRAM), enhanced dynamic random access memory (EDRAM), high-bandwidth memory (HBM), hybrid memory cube (HMC), etc., or any other medium that can be used to store the required information and can be accessed by an application, module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device.

[0051] While this specification has shown and described several embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Numerous modifications, variations, and alternatives will occur to those skilled in the art without departing from the concept and spirit of the present invention. It should be understood that various alternatives to the embodiments of the present invention described herein may be employed in practicing the present invention.

[0052] The above are all preferred embodiments of the present invention, and are not intended to limit the scope of protection of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A material data processing method, characterized in that: include: Get material data of several auto parts; Cluster the material data to obtain the final clusters, and manage the material data of automobile parts according to the final clusters; The final clustering is obtained by presetting the number of cluster centers, obtaining the position of each cluster center, taking any material data as the target point, calculating the degree of dispersion according to the preset neighborhood range of the target point for any iteration, and calculating the probability of the target point belonging to any iteration; The ratio of the attribution possibility to the discrete degree is used as the influence weight of the target point. The influence weight of each material data is traversed to update the position of the cluster center point. In response to the iteration meeting the stopping condition, the final cluster is obtained.

2. A material data processing method according to claim 1, characterized in that: Obtaining the position of each cluster center point includes: The center point of all material data is used as the first cluster center point, and the material data farthest from the first cluster center point is used as the first reference point. The midpoint between the first cluster center point and the reference point is selected as the second cluster center point. The third cluster center point is obtained based on the first cluster center point and the second cluster center point. Similarly, all cluster centers are obtained. In response to the number of cluster center points reaching a preset number of cluster center points, the position of each cluster center point is obtained.

3. A material data processing method according to claim 2, characterized in that: The obtaining of the third cluster center point includes: The material data with the largest sum of distances from the center point of the first cluster and the center point of the second cluster is used as the second reference point; The midpoint of the first cluster center, the second cluster center, and the second reference point is selected as the third cluster center.

4. A material data processing method according to claim 1, characterized in that: Acquiring the discrete degree includes: Get the neighborhood range of the target point, take the material data except the target point in the neighborhood range as the first neighborhood point, and take the neighborhood point of the cluster center point of the cluster to which the target point belongs as the second neighborhood point; The absolute difference between the number of first neighborhood points and the number of second neighborhood points is calculated, and the ratio of the absolute difference to the number of second neighborhood points is used as the degree of dispersion.

5. A material data processing method according to claim 1, characterized in that: Acquiring the discrete degree further includes: Get the neighborhood range of the target point, take the material data except the target point in the neighborhood range as the first neighborhood point, and take the neighborhood point of the cluster center point of the cluster to which the target point belongs as the second neighborhood point; The range of the material data in the first neighborhood point is taken as the first range, and the range of the material data in the second neighborhood point is taken as the second range. The absolute difference between the first range and the second range is calculated, and the ratio of the absolute difference to the second range is taken as the degree of dispersion.

6. A material data processing method according to claim 1, characterized in that: Obtaining the attribution possibility satisfies the relationship: G m,i represents the possibility of belonging of target point i after the mth iteration, d m,i It represents the Euclidean distance between the cluster center of the cluster to which the target point i belongs and the target point i after the mth iteration. It represents the mean of the Euclidean distances between all material data and the cluster center in the cluster to which the target point i belongs after the mth iteration, E m,i,j and E m-1,i,j Respectively represent the value of the cluster center point of the cluster to which the target point i belongs in dimension j after the mth iteration and the m-1th iteration, D i,j Represents the value of target point i in dimension j, M represents the total number of iterations, and J represents the total number of dimensions of material data.

7. A material data processing method according to claim 1, characterized in that: The stopping condition is that after two adjacent iterations are completed, the material data in the cluster does not change.

8. A material data processing system, characterized in that: include: processor; and A memory storing computer instructions, wherein when the computer instructions are executed by a processor, the system executes a material data processing method according to any one of claims 1 to 7.

Images