Artificial intelligence-based tool management tool warehouse layout optimization method

Abstract

The application relates to the technical field of data processing, in particular to a tooling manager storage layout optimization method based on artificial intelligence, which comprises the following steps: constructing a weighted tooling atlas reflecting the historical sharing frequency among various tools; performing multi-scale community discovery scanning on the weighted tooling atlas to generate an initial division set, and calculating the co-occurrence frequency between any two tool nodes in the weighted tooling atlas; constructing a hierarchical stability community tree layer by layer according to a plurality of preset stability thresholds and the co-occurrence frequency; performing optimal segmentation on the hierarchical stability community tree based on a preset storage applicability evaluation function to determine an optimal tool community division scheme, and optimizing the storage layout of the tools based on the optimal tool community division scheme. Through the technical scheme, multi-scale tool groups can be accurately and stably identified, and the reliability of the storage layout scheme is significantly improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to a tool management tool warehouse layout optimization method based on artificial intelligence. BACKGROUND

[0002] In a complex industrial scene, in order to improve the efficiency of material and tool access, one of the leading technical means is to use graph theory and community detection algorithm to optimize the warehouse layout. Specifically, each tool in the warehouse is abstracted as a node in the network graph, and if two tools are often used in the same task, a weighted edge is connected between them, and the weight is proportional to the frequency of joint use. By running the community detection algorithm (such as Louvain algorithm) on this constructed tool relationship graph, the tools with close relationship can be automatically divided into different communities or clusters. In theory, the storage of tools in the same community can be adjacent, which can effectively improve the material preparation efficiency of related operations.

[0003] However, the modularity optimization algorithm represented by the Louvain algorithm has an inherent resolution limit defect. For a given graph, the algorithm may not be able to identify those small-scale and highly cohesive micro-communities. In the tool management scene, this may lead the algorithm to fail to find a special tool group consisting of 3-4 tools for a specific precision maintenance task, but instead to merge it with other tools into a larger and meaningless community.

[0004] At the same time, the existing algorithm often has instability, and even if a few new records are added or a few edges are deleted in the graph, the community division result output by the algorithm may also change dramatically. For the physical warehouse layout that needs to invest a lot of manpower and material resources for adjustment, the algorithm output of such instability and different results each time lacks the reliability and credibility of the decision basis. SUMMARY

[0005] In order to solve the technical problems of existing tool grouping result distortion and unreliability, the present application provides a tool management tool warehouse layout optimization method based on artificial intelligence, which can accurately and stably identify multi-scale tool groups and significantly improve the reliability of the warehouse layout scheme.

[0006] The application provides a tooling management tool warehouse layout optimization method based on artificial intelligence, comprising: obtaining historical taking records of toolings, and constructing a weighted tooling map reflecting historical sharing frequencies between toolings; performing multi-scale community discovery scanning on the weighted tooling map to generate an initial division set containing multiple different granularity division schemes, and calculating co-occurrence frequencies between any two tooling nodes in the weighted tooling map based on the initial division set; constructing a hierarchical stability community tree capable of reflecting hierarchical structures of tooling communities under different stabilities layer by layer according to a plurality of preset stability thresholds in descending order from high to low and the co-occurrence frequencies; performing optimal segmentation on the hierarchical stability community tree based on a preset warehouse applicability evaluation function to determine an optimal tooling community division scheme, and performing warehouse layout optimization on toolings based on the optimal tooling community division scheme.

[0007] The application can simultaneously capture and present the aggregation relationship of toolings under different granularities, ensures that a micro special tool group with only a few pieces or a large general tool set containing dozens of pieces can be stably and accurately identified, solves the resolution limitation problem of community discovery, and realizes accurate identification of multi-scale tool groups.

[0008] In one embodiment, the specific process of performing multi-scale community discovery scanning on the weighted tooling map is to use a Louvain algorithm variant introducing a resolution parameter, select a plurality of different resolution parameter values in a preset range, and perform independent community discovery calculation on the weighted tooling map using each resolution parameter value.

[0009] By repeatedly scanning the map under different scales, all potential aggregation modes of toolings under different granularities can be captured, laying a data foundation for subsequent seeking of consensus across different analysis dimensions, and avoiding one-sidedness caused by single analysis.

[0010] In one embodiment, the calculation method of the co-occurrence frequency is to construct a co-occurrence frequency matrix , wherein represents the co-occurrence frequency of tooling i and tooling j, and the co-occurrence frequency satisfies the relationship: ; wherein N is the total number of community discovery scanning, and are community identifiers to which tooling i and tooling j belong in the kth scanning, respectively, is a Kronecker function, which is 1 when its two parameters are equal, and 0 otherwise.

[0011] Through the matrix, the stability degree of the association relationship between any two toolings can be clearly quantified, and the stability degree is a consensus across different analysis scales, which can better reflect the essential strength of the internal association than a single sharing frequency or weight.

[0012] In one embodiment, the hierarchical stability community tree is constructed in such a way that, for each preset stability threshold, only the associations between nodes with a co-occurrence frequency not lower than the stability threshold are retained in the weighted tooling graph to form an m-layer stable community corresponding to the stability threshold, where m is the index of the stability level; if an m-layer stable community is a superset of multiple (m-1)-layer stable communities, a corresponding parent-child node relationship is established in the community tree.

[0013] The community tree thus constructed can reconfigure unstable communities into a stable hierarchical tree structure, revealing the hierarchical evolution relationship of tooling communities under different stability requirements, and overcoming the resolution limit problem of existing algorithms.

[0014] In one embodiment, the warehouse suitability evaluation function integrates at least one data stability indicator and at least one warehouse physical constraint indicator.

[0015] By combining the data-level stability indicators with the physical-level constraint indicators, it is ensured that the final community division decision is a balanced result between data optimization and physical feasibility.

[0016] In one embodiment, the warehouse suitability evaluation function is calculated as follows: ; where A is a tooling community division scheme, is the score of the scheme, is the average community stability of the division scheme, is the physical incompatibility penalty term of the division scheme, is the storage capacity overflow penalty term of the division scheme, , , is a preset weight coefficient.

[0017] By quantifying and balancing the three core elements of community stability, physical incompatibility, and storage capacity, a scientific evaluation standard is provided for automatically selecting the best from among numerous possible division schemes.

[0018] In one embodiment, the physical incompatibility penalty term is calculated based on a predefined tooling type incompatibility matrix indicating which types of tooling cannot be stored in the same storage unit.

[0019] In one embodiment, the storage capacity overflow penalty term is calculated based on the preset volume data of each tooling and the preset upper limit of the capacity of the standard storage unit.

[0020] In one embodiment, the average community stability of the division scheme is obtained by calculating the arithmetic mean of the co-occurrence frequency of node pairs within all communities in the division scheme.

[0021] In one embodiment, the storage layout optimization of the tooling based on the optimal tooling community division scheme specifically comprises: planning and storing all toolings belonging to the same optimal community in the optimal tooling community division scheme in physically adjacent or functionally preset associated storage units.

[0022] The technical scheme of the present application has the following beneficial technical effects:

[0023] The present application can capture and present the aggregation relationship of tooling at different granularities by constructing a hierarchical stability community tree through multi-resolution scanning, ensuring that both a micro-special tool set with only a few pieces and a large general tool set containing dozens of pieces can be accurately identified, solving the resolution limitation problem of community discovery and realizing the accurate identification of multi-scale tool sets.

[0024] Further, the defect that the results of traditional algorithms are prone to fluctuation is fundamentally solved, and the final division results have robustness to slight perturbations of data, and the results are more stable and reliable, providing reliable data support for guiding physical layout adjustment with high cost. BRIEF DESCRIPTION OF DRAWINGS

[0025] Figure 1 is a flowchart of a tooling storage layout optimization method based on artificial intelligence according to an embodiment of the present application.

[0026] Figure 2 is an original association network diagram containing 50 toolings, which is illustratively shown.

[0027] Figure 3 is a result schematic diagram of the division by using a single standard community discovery algorithm in the prior art.

[0028] Figure 4 is a result schematic diagram of the stable community core identified according to the embodiment of the present application. DETAILED DESCRIPTION

[0029] The technical scheme in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments.

[0030] Figure 1 is a flowchart of a tooling storage layout optimization method based on artificial intelligence according to an embodiment of the present application. As Figure 1As shown, the tooling management warehouse layout optimization method based on artificial intelligence includes steps S101 to S104, which are described in detail below.

[0031] S101, obtain the historical usage records of tooling and construct a weighted tooling map that reflects the historical frequency of shared use among various tooling components.

[0032] In one embodiment, the tooling and equipment that needs to be managed in the warehouse can be numbered. For example, if there are 50 different types of tooling and equipment managed in the warehouse, these tooling and equipment can be numbered from nodes v0 to v49. Specifically, all usage records of these tooling and equipment in the past evaluation period, such as one year, can be retrieved through the enterprise warehouse management system or production execution system. Each usage record clearly indicates which tooling and equipment were used for which maintenance task.

[0033] Furthermore, a weighted undirected graph G=(V, E) can be constructed, where the node set V contains 30 nodes (v0 to v29), corresponding to 30 types of tooling. Iterate through all requisition records, and for each record, find any two tooling nodes within the tooling set (for example, ...). and Each of them establishes or updates an edge between them. . Specifically, if and If there is no edge between them, create an edge with a weight of 1; if an edge already exists, increment its weight by 1. After traversing all records, the weight of the edge between any two nodes is equal to the total number of times they were used together during the evaluation period. This yields a weighted tooling graph reflecting the historical frequency of tooling usage.

[0034] In this optional embodiment, analysis shows that these toolkits naturally form three closely related toolkits in actual use, resulting in the following real community structure: small community A = {v0, v1, v2, v3, v4}, small community B = {v5, v6, v7, v8, v9}, and a large community C = {v10, v11, ..., v29}. When constructing the weighted toolkit graph, not only are high-weighted connections established between nodes within each community, but also a small number of connections exist between v4 and v5, and between v3 and v6. Figure 2 The diagram shown is the original network diagram of the completed project, which includes 30 tooling nodes.

[0035] Thus, by transforming discrete and unstructured requisition history into structured and computable weighted spectral data, the necessary data foundation is provided for subsequent analysis.

[0036] S102, performing multi-scale community discovery scanning on the weighted tool map to generate an initial partition set containing multiple different granularity partition schemes, and calculating the co-occurrence frequency between any two tool nodes in the weighted tool map based on the initial partition set.

[0037] In one embodiment, in order to analyze the potential clustering patterns between tools from different analysis scales, multi-scale scanning can be performed on the constructed weighted tool map. Specifically, a Louvain algorithm variant supporting resolution parameters can be used, multiple different resolution parameter values are selected within a preset range, and each resolution parameter value is used to perform an independent community discovery calculation on the weighted tool map.

[0038] For example, two independent community discovery calculations are performed on the same weighted tool map by selecting resolution parameters with values of 0.7 and 1.5, respectively. It is found that when the resolution parameter is 0.7, the algorithm corresponds to discovering larger communities; when the resolution parameter is 1.5, the algorithm corresponds to discovering smaller communities. These two different community partition schemes together constitute the initial partition set. As shown in Figure 3 The scanning result of the prior art using a standard single resolution parameter of 1 can be seen, which incorrectly merges small community A and small community B into one community.

[0039] In one embodiment, the co-occurrence frequency between any two tool nodes and can be calculated to measure the stability of their association. Specifically, a co-occurrence frequency matrix can be constructed, where represents the co-occurrence frequency of tool i and tool j, and the co-occurrence frequency satisfies the relationship:

[0040]

[0041] where N is the total number of community discovery scans, and are the community identifiers to which tool i and tool j belong in the kth scan, respectively, is the Kronecker function, which is 1 when its two parameters are equal, and 0 otherwise.

[0042] For example, for nodes v0 and v1, they are in the same community in both scans, so the co-occurrence frequency is 2 / 2 = 1, while for nodes v4 and v5, they are in the same community only in the low-resolution scan, and are separated in the high-resolution scan, so their co-occurrence frequency is 1 / 2 = 0.5.

[0043] Thus, by using multi-resolution scanning and co-occurrence frequency calculation, the strength of the association between nodes can be transformed from a single number of shared occurrences into a stability dimension that better reflects the nature of their association for evaluation.

[0044] S103, based on multiple preset stability thresholds and co-occurrence frequencies decreasing from high to low, constructs a hierarchical stability community tree that reflects the hierarchical structure of the tooling community under different stability conditions.

[0045] In one embodiment, a stability threshold sequence decreasing from high to low can be set, for example, {1.0, 0.9, 0.8}, and combined with co-occurrence frequency, a hierarchical stability community tree that can reflect the hierarchical structure of the tooling community under different stability conditions can be constructed layer by layer.

[0046] Specifically, at the highest stability threshold of 1.0, only associations with a co-occurrence frequency of 1 are retained in the weighted tooling graph. The resulting independent connected components are the leaf nodes at the bottom layer of the community tree. For example... Figure 4 As shown, three core communities can be identified at this time: small community A={v0, v1, v2, v3, v4}, small community B={v5, v6, v7, v8, v9}, and large community C={v10, v11, ..., v29}.

[0047] At the next stability threshold of 0.9, associations with a co-occurrence frequency of not less than 0.9 are retained. This allows nodes in a community to connect with other nodes, thus forming a larger community. Specifically, for each stability threshold, only associations between nodes with a co-occurrence frequency not lower than that threshold are retained in the weighted tooling graph, forming m-level stable communities corresponding to that stability threshold, where m is the index of the stability level. If an m-level stable community is a superset of multiple (m-1)-level stable communities, a corresponding parent-child node relationship is established in the community tree. For example, if, at a stability threshold of 0.9, all nodes in community C1 are also connected with another free node, forming a larger community C2, then in the community tree structure, C2 is the parent node of C1. By successively lowering the threshold and analyzing the merging and evolutionary relationships of communities, a complete hierarchical stability community tree is finally constructed.

[0048] Thus, by constructing a hierarchical stability community tree, we can preserve the stability community information of the tooling network at all scales in a structured way, reveal the hierarchical relationships between them, and solve the resolution limitation problem of existing technologies.

[0049] S104, performing optimal partitioning on the hierarchical stability community tree based on a preset warehouse suitability evaluation function to determine an optimal tool community partitioning scheme, and performing warehouse layout optimization on the tools based on the optimal tool community partitioning scheme.

[0050] In one embodiment, in order to find an optimal partitioning scheme that is not only reasonable in data but also physically feasible from the obtained hierarchical stability community tree, a preset warehouse suitability evaluation function can be used for evaluation, and the calculation formula of the warehouse suitability evaluation function is:

[0051]

[0052] wherein A is a tool community partitioning scheme, is the score of the scheme, is the average community stability of the partitioning scheme, is the physical incompatibility penalty term of the partitioning scheme, is the storage capacity overflow penalty term of the partitioning scheme, , , is a preset weight coefficient, and an example can be , , .

[0053] Further, the physical properties of the tools can be predefined, such as v0-v4 for hydraulic tools, v5 to v9 for precision instruments, and v10 to v29 for electric tools. Among them, the precision instruments and hydraulic tools are defined as physically incompatible, and the volume of each tool can also be defined, and the capacity upper limit of a standard storage unit (such as a tool cabinet) is set to 1.0 cubic meters. Thus, by performing optimal partitioning on the hierarchical stability community tree in combination with the preset warehouse suitability evaluation function, an optimal tool community partitioning scheme can be determined.

[0054] For example, for scheme one, when the stability threshold is 1, the internal co-occurrence frequency is set to 1, there is no physical incompatibility type in the community, the penalty for this item is 0, and the storage capacity check is all over the limit, the penalty for this item is 0, and the final score is ; and for scheme two, the stability threshold is 0.8, Ca and Cb communities are merged, the corresponding average community stability is 0.85, there is no physical incompatibility type in the community, the penalty for this item is 0, the total volume of the merged community is 1.1 cubic meters, which exceeds the upper limit, the penalty for this item is 0.1, and the final score is , so it can be determined that scheme one is the optimal tool community partitioning scheme.

[0055] In the optional embodiment, finally, the toolings belonging to the same community can be planned into the physically adjacent storage units in the warehouse according to the optimal tooling community division scheme, so as to realize the storage layout optimization of the toolings.

[0056] In this way, by introducing the evaluation function containing the physical constraint for optimal segmentation, it is ensured that the community division scheme finally output is not only optimal in the data level, but also realistic, feasible, safe and reasonable in the physical storage level, and the substantial improvement of the storage efficiency is realized.

[0057] It should be noted that, for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. An artificial intelligence-based tool management tool warehouse layout optimization method, characterized by, The method comprises the following steps: acquiring historical usage records of tooling and constructing a weighted tooling graph reflecting historical co-usage frequencies between toolings; performing multi-scale community finding scanning on the weighted tooling graph to generate an initial partition set containing multiple different granularity partition schemes, and calculating the co-occurrence frequency between any two tooling nodes in the weighted tooling graph based on the initial partition set; the co-occurrence frequency is calculated by constructing a co-occurrence frequency matrix wherein represents the co-occurrence frequency of tooling i and tooling j, and the co-occurrence frequency satisfies the relationship: where N is the total number of community discovery scans, and are the community identifiers to which tool i and tool j belong in the kth scan, respectively, is the Kronecker delta function, which is 1 when its two arguments are equal and 0 otherwise. constructing a hierarchical stability community tree capable of reflecting hierarchical structures of tooling communities at different stabilities according to a plurality of stability thresholds arranged in descending order and the co-occurrence frequencies; performing optimal splitting on the hierarchical stability community tree based on a preset storage suitability evaluation function to determine an optimal tooling community division scheme and performing storage layout optimization on toolings based on the optimal tooling community division scheme; the storage suitability evaluation function comprehensively considers at least one data stability index and at least one physical constraint index of storage, and a calculation formula of the storage suitability evaluation function is: wherein A is a tool community partitioning scheme, a score for the scheme, an average community stability for the partitioning scheme, a physical incompatibility penalty term for the partitioning scheme, a storage capacity overflow penalty term for the partitioning scheme, , , is a preset weight coefficient. 2.The AI-based tooling manager warehouse layout optimization method of claim 1, wherein, The specific process of the multi-scale community discovery scanning on the weighted tooling graph is to use a Louvain algorithm variant with a resolution parameter, select a plurality of different resolution parameter values within a preset range, and perform an independent community discovery calculation on the weighted tooling graph using each resolution parameter value. 3.The AI-based tooling manager warehouse layout optimization method of claim 1, wherein, The hierarchical stability community tree is constructed in the following manner: for each preset stability threshold, only the associations between nodes with a co-occurrence frequency not lower than the stability threshold are retained in the weighted tooling graph to form an m-layer stable community corresponding to the stability threshold, where m is the index of the stability level; if an m-layer stable community is a superset of a plurality of (m-1)-layer stable communities, a corresponding parent-child node relationship is established in the community tree. 4.The AI-based tooling manager warehouse layout optimization method of claim 1, wherein, The physical incompatibility penalty term is calculated based on a predefined tooling type incompatibility matrix indicating which types of toolings cannot be stored in the same storage unit.

5. The AI-based tooling manager warehouse layout optimization method of claim 1, wherein, The storage capacity overflow penalty term is calculated based on preset volume data of each tooling and a preset upper limit of the capacity of a standard storage unit.

6. The AI-based tooling manager warehouse layout optimization method of claim 1, wherein, The average community stability of the division scheme is obtained by calculating the arithmetic mean of the co-occurrence frequencies of node pairs within all communities in the division scheme. 7.The AI-based tooling manager warehouse layout optimization method of claim 1, wherein, The storage layout optimization on toolings based on the optimal tooling community division scheme specifically comprises planning and storing all toolings belonging to the same optimal community in the optimal tooling community division scheme in storage units that are physically adjacent or functionally preset associated.

Images