Machine learning-based shock absorber overhaul tool intelligent management method and system

Abstract

The present application relates to the field of shock absorber maintenance, in particular to a shock absorber maintenance tool intelligent management method and system based on machine learning, comprising: constructing a tool usage set of any historical maintenance work order; using a clustering algorithm to cluster all tool usage sets to obtain clustering clusters, calculating the frequency of each tool usage, and filtering tools according to the frequency to obtain a high-weight tool set and an important tool set; taking the ratio of the number of times that the target tool appears in the high-weight tool set to the number of times that the target tool appears in the important tool set as the contribution strength, calculating the cumulative value of the number of times that each tool appears in the important tool set, and taking the ratio of the number of times that the target tool appears in the important tool set to the cumulative value as the coverage rate; generating the storage code of the target tool according to the contribution strength and the coverage rate, and traversing to obtain the storage code of all tools; and managing the tools according to all the storage codes. The present application can improve the management standardization of shock absorber maintenance tools.

Description

Technical Field

[0001] This invention relates to the field of tooling management. More specifically, this invention relates to a machine learning-based intelligent management method and system for vibration damper maintenance tooling. Background Technology

[0002] During the maintenance of vibration dampers, various tooling tools, such as clamps and measuring instruments, are required. Quickly and accurately locating the necessary tooling directly impacts maintenance efficiency and quality. Traditional techniques typically rely on simple machine learning algorithms (such as decision tree models) to recommend tooling to maintenance personnel. However, because tooling storage often depends on individual habits and lacks unified management, it makes it difficult for others to find the tools, thus affecting maintenance efficiency. Therefore, traditional machine learning-based vibration damper maintenance tooling management methods have limitations. Summary of the Invention

[0003] The main objective of this application is to propose an intelligent management method and system for vibration damper maintenance tooling based on machine learning, aiming to improve the management standardization of vibration damper maintenance tooling.

[0004] To achieve the above objectives, an embodiment of the first aspect of this application proposes an intelligent management method for vibration damper maintenance tooling based on machine learning. The method includes: constructing a tooling usage set for any historical maintenance work order; using a clustering algorithm to cluster all tooling usage sets to obtain clusters; for any cluster, calculating the frequency of use of each tooling; filtering tools based on the frequency to obtain a high-weight tooling set and an important tooling set, wherein the high-weight tooling set is a subset of the important tooling set; selecting any tooling as a target tooling; using the ratio of the number of times the target tooling appears in the high-weight tooling set to the number of times it appears in the important tooling set as a contribution intensity; calculating the cumulative value of the number of times each tooling appears in the important tooling set; using the ratio of the number of times the target tooling appears in the important tooling set to the cumulative value as a coverage rate; generating a storage code for the target tooling based on the contribution intensity and the coverage rate; traversing to obtain the storage codes of all tools; and managing the tooling based on all the storage codes.

[0005] In some embodiments, clustering all tooling usage sets using a clustering algorithm to obtain clusters includes: taking each tooling usage set as an initial cluster, taking any initial cluster as a target cluster, taking any initial cluster other than the target cluster as a reference cluster, calculating the target similarity between the target cluster and the reference cluster, merging the target cluster and the reference cluster to obtain a new cluster in response to the target similarity being greater than a preset upper limit, calculating the new similarity between the new cluster and any cluster other than the target cluster and the reference cluster using the target similarity calculation method, iterating the above process of obtaining a new cluster in response to the new similarity being greater than a preset upper limit, until the similarity between any two clusters is not greater than a preset upper limit, ending the iteration, and taking the cluster after the iteration as the cluster.

[0006] In some embodiments, calculating the target similarity between the target cluster and the reference cluster includes: using the ratio of the intersection of the target cluster and the reference cluster to the union of the target cluster and the reference cluster as the similarity between the target cluster and the reference cluster.

[0007] In some embodiments, filtering tooling based on usage frequency to obtain a high-weight tooling set and an important tooling set includes: for any tooling, in response to the usage frequency being greater than a first preset weight threshold, the tooling is included in the high-weight tooling set; in response to the usage frequency being greater than a second preset weight threshold, the tooling is included in the important tooling set, wherein the first preset weight threshold is greater than the second preset weight threshold.

[0008] In some embodiments, generating the storage code of the target tooling based on the contribution intensity and the coverage includes: obtaining the ranking of the contribution intensity of the target tooling among all contribution intensities; in response to the contribution intensity being not less than a first preset contribution threshold and the coverage being greater than a preset lower limit, using a first region as the storage area of ​​the target tooling, and concatenating the area code of the first region and the ranking to obtain the storage code; in response to the contribution intensity being less than the first preset contribution threshold and greater than a second preset contribution threshold and the coverage being not greater than a preset lower limit, using a second region as the storage area of ​​the target tooling, and concatenating the area code of the second region and the ranking to obtain the storage code; in response to the contribution intensity being not greater than the second preset contribution threshold and the coverage being not greater than a preset lower limit, using a second region as the storage area of ​​the target tooling, and concatenating the area code of the second region and the ranking to obtain the storage code.

[0009] In some embodiments, in response to the target tooling appearing in the important tooling set 0 times, the area code of the spare area is used as the storage code of the target tooling.

[0010] In some embodiments, the machine learning-based intelligent management method for vibration damper maintenance tools further includes: for any set of tooling usage, obtaining the maintenance type and maintenance object corresponding to the set of tooling usage, using the maintenance type and maintenance object as features and the important tooling sets of the cluster to which the set of tooling usage belongs as labels to construct samples, traversing to obtain samples corresponding to each set of tooling usage to construct a sample set, and using a machine learning algorithm to construct a tooling recommendation model based on the sample set.

[0011] The second aspect of this application proposes a machine learning-based intelligent management system for vibration damper maintenance fixtures. The system includes a processor and a memory, wherein the memory stores computer program instructions. When the computer program instructions are executed by the processor, the above-described machine learning-based intelligent management method for vibration damper maintenance fixtures is implemented.

[0012] The beneficial effects of this invention are:

[0013] This invention constructs tool usage sets based on the usage of tools in historical maintenance work orders, where each historical maintenance work order corresponds to one tool usage set. A clustering algorithm is used to classify all obtained tool usage sets into clusters. Each cluster contains several similar tool usage sets to define a tool usage scenario. For any cluster, the frequency of each tool's use within that cluster is calculated. Higher frequency indicates more frequent use of the tool in the corresponding tool usage scenario. Tools are selected based on frequency to obtain high-weight tool sets and important tool sets. Since the high-weight tool sets are subsets of the important tool sets, tools in the high-weight tool sets will also be in the important tool sets. At this point, each cluster (tool usage scenario) has one high-weight tool set and one important tool set. Next, this invention introduces a contribution rate to quantify the importance of tools: for any tool, the number of times it appears in all high-weight tool sets is counted. The frequency of each tool appearing in all important tool sets is used, and the ratio of these two frequencies is taken as the contribution strength. The contribution strength is no greater than 1, and the closer the contribution strength is to 1, the higher the weight of the tool in most of its associated clusters (tool use scenarios). In this case, the tool's status is that it will be used very frequently. Conversely, the closer the contribution strength is to 0, the higher the weight of the tool is only in a small number of its associated clusters (tool use scenarios). That is, the tool's status is that it is used very frequently only in a few scenarios. In addition, this invention introduces coverage to quantify the universality of the tool: the frequency of each tool appearing in all important tool sets is counted, and the cumulative value of the frequency is obtained. The frequency of each tool appearing in all important tool sets and the cumulative value of the frequency are taken as the coverage of the tool. The higher the coverage, the stronger the universality of the tool compared to other tools. Finally, a corresponding storage code is generated based on the contribution strength and coverage of each tool, and the tool is managed based on the storage code. Therefore, this invention can manage tooling according to its importance and versatility, making tooling management more standardized and helping maintenance personnel to quickly and accurately locate tooling, thereby improving the maintenance efficiency and quality of vibration dampers. Attached Figure Description

[0014] Figure 1 This is a flowchart of steps S1-S3 in the intelligent management method for vibration damper maintenance tooling based on machine learning in an embodiment of the present invention. Detailed Implementation

[0015] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0016] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0017] Reference Figure 1 The intelligent management method for vibration damper maintenance tooling based on machine learning includes steps S1-S3, as follows:

[0018] Step S1: Construct the tooling set for any historical maintenance work order.

[0019] It should be noted that the maintenance information included in the historical maintenance work orders includes the vehicle model to be maintained, the object to be maintained, and a list of tools used. The vehicle model to be maintained is, for example, the Fuxing CR4004AF EMU, the object to be maintained is, for example, the model of the Fuxing CR4004AF EMU's primary vertical shock absorber, and the list of tools used is, for example, fixtures. Tools that exist in the list of tools used are included in the tool usage set.

[0020] Step S2: Use a clustering algorithm to cluster all tooling usage sets to obtain clusters. For any cluster, calculate the frequency of each tooling usage. Based on the frequency, filter the tooling to obtain a high-weight tooling set and an important tooling set. The high-weight tooling set is a subset of the important tooling set.

[0021] Understandably, clustering all tooling usage sets using a clustering algorithm to obtain clusters involves: taking each tooling usage set as an initial cluster, taking any initial cluster as a target cluster, taking any initial cluster other than the target cluster as a reference cluster, calculating the target similarity between the target cluster and the reference cluster, merging the target cluster and the reference cluster to obtain a new cluster when the target similarity is greater than a preset upper limit, calculating the new similarity between the new cluster and any cluster other than the target cluster and the reference cluster using the target similarity calculation method, iterating the above process of obtaining a new cluster when the new similarity is greater than a preset upper limit, until the similarity between any two clusters is no greater than a preset upper limit, ending the iteration, and taking the cluster after the iteration as the cluster.

[0022] It should be noted that the preset upper limit for similarity is 0.7 (this threshold can be adjusted according to the requirements of the maintenance scenario for "consistency of tooling requirements": if it is necessary to strictly ensure the similarity of tooling used within the cluster, the threshold can be increased; if a certain degree of difference is allowed, the threshold can be decreased).

[0023] It should be noted that after the iteration, each cluster contains several tooling usage sets, and each cluster represents a tooling usage scenario. For any cluster, the frequency of occurrence of each tooling is counted, and this frequency is the tooling usage frequency.

[0024] It is understandable that calculating the target similarity between the target cluster and the reference cluster involves using the ratio of the intersection of the target cluster and the reference cluster to the union of the target cluster and the reference cluster as the similarity between the target cluster and the reference cluster.

[0025] For example, the similarity is calculated as follows:

[0026]

[0027] In the formula middle, Let A be the similarity between cluster A and cluster B. The intersection of cluster A and cluster B. Let be the union of cluster A and cluster B.

[0028] It is understandable that the process of filtering tooling based on usage frequency to obtain a set of high-weight tooling and a set of important tooling includes: for any tooling, in response to a usage frequency greater than a first preset weight threshold, the tooling is included in the set of high-weight tooling, and in response to a usage frequency greater than a second preset weight threshold, the tooling is included in the set of important tooling, wherein the first preset weight threshold is greater than the second preset weight threshold.

[0029] It should be further explained that for any cluster, all tools within the cluster are sorted by usage frequency from low to high. A high percentile (e.g., 70%) is selected as the first preset weight threshold, and tools with usage frequencies greater than this threshold are included in the high-weight tool set. Simultaneously, a low percentile (e.g., 50%) is selected as the second preset weight threshold, and tools with usage frequencies greater than this threshold are included in the important tool set. The high-weight tool set is a subset of the important tool set.

[0030] It should be noted that each cluster has a set of high-weight tools and a set of important tools. The first preset weight threshold may be different for different clusters, and the second preset weight threshold may also be different. This allows for adaptation to different tool usage scenarios.

[0031] Step S3: Take any tooling as the target tooling, and use the ratio of the number of times the target tooling appears in the high-weight tooling set to the number of times it appears in the important tooling set as the contribution intensity. Calculate the cumulative value of the number of times each tooling appears in the important tooling set, and use the ratio of the number of times the target tooling appears in the important tooling set to the cumulative value as the coverage. Generate the storage code of the target tooling based on the contribution intensity and coverage. Iterate through and obtain the storage codes of all tooling. Manage the tooling based on all storage codes.

[0032] Understandably, generating the storage code for the target tooling based on contribution intensity and coverage includes: obtaining the ranking of the target tooling's contribution intensity among all contribution intensities; in response to a contribution intensity not less than a first preset contribution threshold and coverage greater than a preset lower limit, using the first region as the storage area for the target tooling, and concatenating the region number and ranking of the first region to obtain the storage code; in response to a contribution intensity less than the first preset contribution threshold but greater than a second preset contribution threshold and coverage not greater than a preset lower limit, using the second region as the storage area for the target tooling, and concatenating the region number and ranking of the second region to obtain the storage code; in response to a contribution intensity not greater than the second preset contribution threshold and coverage not greater than a preset lower limit, using the second region as the storage area for the target tooling, and concatenating the region number and ranking of the second region to obtain the storage code.

[0033] For example, the formula for calculating contribution intensity is as follows:

[0034]

[0035] In the formula middle, For the first The contribution intensity of the tooling For the first The number of times a tool appears in a high-weight tool set. For the first The number of times a tool appears in an important tool set.

[0036] For example, the coverage calculation formula is as follows:

[0037]

[0038] In the formula middle, For the first The coverage rate of various tooling, For the first The number of times a certain type of tooling appears in a set of important tooling. , This is the cumulative value representing the number of times each tool appears in the set of important tooling. This represents the number of tooling types.

[0039] For the above formula It should be further explained that if the target tool appears in the high-weight tool set of a certain cluster, it means that the target tool is used very frequently in the tool use scenario corresponding to that cluster (the frequency of use is higher than the first preset weight threshold). The closer the contribution strength is to 1, the more likely the target tool is a high-weight tool in most of its associated clusters (tool use scenarios), meaning that the target tool will be used very frequently whenever it is used. Conversely, the closer the contribution strength is to 0, the more likely the target tool is a high-weight tool only in a small number of its associated clusters (tool use scenarios), meaning that the target tool is used very frequently only in a few scenarios (the target tool tends to be specialized).

[0040] For the above formula It should be added that the greater the coverage, the stronger the versatility of the target tooling compared to other tooling.

[0041] For example, the first preset contribution threshold is 0.7, the second preset contribution threshold is 0.3, the preset lower limit is 0.5, and the area code of the first region is 01. The contribution intensity of the target tooling is 0.8, the coverage rate is 0.7, the contribution intensity of the target tooling ranks 5th among all tooling contribution intensities, and the storage code of the target tooling is 01005.

[0042] Understandably, the machine learning-based intelligent management method for shock absorber maintenance tooling also includes: in response to the target tooling appearing in the set of important toolings 0 times, using the area code of the spare area as the storage code of the target tooling.

[0043] It should be noted that a count of 0 indicates that the target tool is not used frequently in various maintenance scenarios. In this case, the target tool can be placed in the spare area.

[0044] Understandably, the intelligent management method for vibration damper maintenance tools based on machine learning also includes: for any set of tools used, obtaining the maintenance type and maintenance object corresponding to any set of tools used, using the maintenance type and maintenance object as features, and using the important tool sets of the cluster to which any set of tools used as labels to construct samples, traversing to obtain the samples corresponding to each set of tools used to construct a sample set, and using machine learning algorithms to construct a tool recommendation model based on the sample set.

[0045] It should be noted that the inspection type can be the vehicle model to be inspected. The vehicle model and the object to be inspected are input into the intelligent management system for shock absorber inspection tools. The intelligent management system for shock absorber inspection tools obtains a set of important tools through the tool recommendation model. The intelligent management system for shock absorber inspection tools outputs the set of important tools and the storage code of each tool in the set.

[0046] This invention also provides an intelligent management system for vibration damper maintenance tooling based on machine learning. The system includes a processor and a memory, the memory storing computer program instructions. When the processor executes the computer program instructions, it implements the intelligent management method for vibration damper maintenance tooling based on machine learning according to the first aspect of this invention. The system also includes other components well-known to those skilled in the art, such as a communication bus and a communication interface; their configuration and functions are known in the art and will not be described further here.

[0047] It should be noted that the preferred embodiments of this application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of this application. For those skilled in the art, various modifications and improvements can be made without departing from the concept of the invention, and these all fall within the protection scope of the invention. Therefore, the protection scope of this patent should be determined by the appended claims.

Claims

1. A machine learning-based intelligent management method for vibration damper maintenance tooling, characterized in that, include: Construct a tooling set for any historical maintenance work order; Clustering algorithms are used to cluster all the tooling usage sets to obtain clusters. For any cluster, the frequency of use of each tooling is calculated. Based on the frequency, tooling is filtered to obtain a high-weight tooling set and an important tooling set. The high-weight tooling set is a subset of the important tooling set. Take any tooling as the target tooling, and use the ratio of the number of times the target tooling appears in the high-weight tooling set to the number of times it appears in the important tooling set as the contribution intensity. Calculate the cumulative value of the number of times each tooling appears in the important tooling set, and use the ratio of the number of times the target tooling appears in the important tooling set to the cumulative value as the coverage rate. Generate the storage code of the target tooling based on the contribution intensity and the coverage rate, and iterate to obtain the storage codes of all tooling. Manage the tooling according to all the described storage codes.

2. The intelligent management method for vibration damper maintenance tooling based on machine learning according to claim 1, characterized in that, Clustering algorithms were used to cluster all the aforementioned tooling into sets, resulting in the following clusters: Each tooling set is used as an initial cluster, any initial cluster is used as a target cluster, and any initial cluster other than the target cluster is used as a reference cluster. The target similarity between the target cluster and the reference cluster is calculated. In response to the target similarity being greater than a preset upper limit, the target cluster and the reference cluster are merged to obtain a new cluster. The new similarity between the new cluster and any cluster other than the target cluster and the reference cluster is calculated using the target similarity calculation method. In response to the new similarity being greater than a preset upper limit, the process of obtaining a new cluster is iterated until the similarity between any two clusters is no greater than the preset upper limit. The iteration ends, and the clusters after the iteration are taken as clustering clusters.

3. The intelligent management method for vibration damper maintenance tooling based on machine learning according to claim 2, characterized in that, Calculating the target similarity between the target cluster and the reference cluster includes: The similarity between the target cluster and the reference cluster is defined as the ratio of the intersection of the target cluster and the reference cluster to the union of the target cluster and the reference cluster.

4. The intelligent management method for vibration damper maintenance tooling based on machine learning according to claim 1, characterized in that, The tooling set filtered by usage frequency yields a high-weight tooling set and an important tooling set, including: For any tooling, in response to the usage frequency being greater than a first preset weight threshold, the tooling is included in a high-weight tooling set; in response to the usage frequency being greater than a second preset weight threshold, the tooling is included in an important tooling set, wherein the first preset weight threshold is greater than the second preset weight threshold.

5. The intelligent management method for vibration damper maintenance tooling based on machine learning according to claim 1, characterized in that, The generation of the target tooling storage code based on the contribution intensity and the coverage includes: Obtain the ranking of the contribution intensity of the target tooling among all contribution intensities; In response to the fact that the contribution intensity is not less than a first preset contribution threshold and the coverage is greater than a preset lower limit, the first region is used as the storage area of ​​the target tooling, and the storage code is obtained by concatenating the area code of the first region and the ranking. In response to the situation where the contribution intensity is less than a first preset contribution threshold and greater than a second preset contribution threshold and the coverage rate is not greater than a preset lower limit, the second region is used as the storage area for the target tooling, and the storage code is obtained by concatenating the area code of the second region and the ranking. In response to the fact that the contribution intensity is not greater than the second preset contribution threshold and the coverage is not greater than the preset lower limit, the second region is used as the storage area of ​​the target tooling, and the storage code is obtained by concatenating the area code of the second region and the ranking.

6. The intelligent management method for vibration damper maintenance tooling based on machine learning according to claim 1, characterized in that, Also includes: In response to the target tooling appearing in the important tooling set 0 times, the area code of the spare area is used as the storage code of the target tooling.

7. The intelligent management method for vibration damper maintenance tooling based on machine learning according to claim 1, characterized in that, Also includes: For any set of tooling usage, obtain the maintenance type and maintenance object corresponding to the set of tooling usage, use the maintenance type and maintenance object as features, and use the important tooling sets of the cluster to which the set of tooling usage belongs as labels to construct samples. Iterate through and obtain samples corresponding to each set of tooling usage to construct a sample set. Based on the sample set, use machine learning algorithms to construct a tooling recommendation model.

8. A machine learning-based intelligent management system for vibration damper maintenance tooling, characterized in that, include: A processor and a memory, wherein the memory stores computer program instructions that, when executed by the processor, implement the intelligent management method for vibration damper maintenance fixtures based on machine learning according to any one of claims 1-7.

Images