Material warehousing coding and inventory screening matching method and system

Abstract

The invention discloses a material warehousing coding and inventory screening matching method and system, and the method comprises the steps: carrying out the similarity calculation of material attributes of inventory materials, the material attributes comprise text attributes and numerical attributes, and carrying out the calculation of the similarity of each attribute for the numerical attributes, then different weights are given according to the importance of the attributes, and finally the weighted attribute similarities are summed to obtain the overall similarity of the materials; for text attributes, calculation is carried out based on the similarity of a vector space model, and target materials are screened and matched by calculating the cosine similarity between two material attribute vectors. According to the method, existing stock material matching and checking with consideration of multiple material attributes can be solved, and the utilization rate of existing stock materials is increased.

Description

Technical Field

[0001] The present invention belongs to the material inventory management technology of industrial enterprises, and specifically relates to a material warehousing coding and inventory screening and matching method and system. Background Art

[0002] In today's digital age, the construction of enterprise informatization continues to advance. Industrial enterprises are also actively engaged in information technology transformation in terms of inventory material management, aiming to comprehensively improve the work efficiency and accuracy of material procurement, storage and use processes, thereby enhancing the overall operational efficiency of the enterprise. However, when faced with a large inventory of material warehouses, many companies still face severe challenges. In particular, the management of non-standard materials has become a thorny problem. Taking the material steel pipe as an example, in the procurement and warehousing link, since the steel pipe has not been used, its size specifications, materials and other information are registered, and it is relatively easy to match it when the material is subsequently taken. However, if a steel pipe has been partially used and registered again, how to accurately match this type of residual material with the target demand material has become a major bottleneck that plagues the company's material management. This not only affects the effective utilization rate of materials, but may also lead to additional procurement costs and inventory backlogs.

[0003] Further in-depth analysis shows that in the process of material screening and matching, the principle of giving priority to existing inventory materials is followed. In order to achieve a better matching effect, it is necessary to comprehensively consider multiple material attributes and accurately determine their proximity to the target material, that is, the comprehensive similarity of the materials. This not only involves complex data analysis, but also requires the establishment of a set of scientific and reasonable calculation rules. At present, although there are some material management methods on the market, there are generally limitations when dealing with multi-attribute material matching, and it is difficult to achieve the optimal target matching under multiple material attributes.

[0004] Therefore, it is urgent to develop an efficient material warehousing coding and inventory screening and matching method and system, which has important practical significance for improving the enterprise inventory management level and reducing operating costs. Summary of the invention

[0005] Purpose of the invention: The present invention aims to provide a material entry coding and inventory screening and matching method and system, which is used to solve the problems that the existing material inventory management retrieval is complex and the matching degree is inaccurate, and the redundant coding leads to chaotic inventory data and cannot be classified.

[0006] Technical solution: A material warehousing coding and inventory screening and matching method, the method performs similarity calculation on the material attributes of the inventory materials, the material attributes include text attributes and numerical attributes, the text attributes refer to keyword information including material name and material, and the numerical attributes refer to digital information including material size specifications;

[0007] Similarity calculation of numerical attributes: Let the weight of the numerical attribute x be w x , and for the required material A, the matching degrees of the inventory material B on this numerical attribute are x b . Define the similarity of material numerical attributes as:

[0008] Sim(A,B) 数值 = w x x b

[0009] Similarity calculation of text attributes: First, assign values to the keyword information in the text attributes of the materials, then construct material attribute vectors from several text attribute values, and calculate their similarity by calculating the cosine similarity between the two material attribute vectors. Specifically:

[0010] Suppose the attributes of the required material A The existing attributes of material B Then their cosine similarity is:

[0011]

[0012] For more than one numerical attribute, different weights are assigned according to the importance of the attributes, and then the similarity of each material attribute is calculated separately, and the weighted attribute similarities are summed to obtain the overall similarity of the materials, thereby realizing the screening and matching of inventory materials;

[0013] The final matching degree of the material is:

[0014] Sim(A,B) = Sim(A,B) 文本 + Sim(A,B) 数值 .

[0015] Furthermore, in the calculation of numerical attributes, the processing of the matching degree includes empirical assignment and / or calculation based on the Euclidean distance method;

[0016] For a material with n numerical attributes, for the i-th numerical attribute, the numerical attribute value of the required material is denoted as A 0i , and the numerical attribute value of the inventory material is B ji . Calculate the matching degree by using the Euclidean distance method:

[0017]

[0018] Then the overall matching degree of the inventory material B relative to the required material A under this attribute is:

[0019]

[0020] Furthermore, the similarity calculation of text attributes includes processing the calculation of word frequency vectors based on Python code.

[0021] For the similarity calculation results of several related materials in inventory, the method outputs the matching degree between the candidate material and the inventory material according to the size of the similarity.

[0022] Furthermore, the method for material coding is as follows:

[0023] Formulate master data business rules according to the source of materials, management level, basic data items, description and meaning explanation of the generation process of data items in the business environment, the restrictive relationship between data, and the logical rules to be followed in the data generation process; the master data business rules include the coding specifications, classification rules and naming rules of each data item of the master data, and establish a material coding template.

[0024] In this method, the coding rules are classified according to industrial industry standards and / or the actual inventory materials of users, including constants, classification codes and serial numbers. The constants refer to numbers or values that do not need to be modified, including the letter prefix added before the coding of self-made parts of materials for the management of material attributes.

[0025] Based on the implementation of the above method, the present invention can also construct an industrial material information data management system, which encodes materials for warehousing and performs inventory matching and screening according to the above method to realize the information data management of industrial materials.

[0026] Beneficial effects: Compared with the prior art, the present invention combines several material attributes for matching, calculates the similarity according to the preset similarity weight value of the system, and then outputs the inventory material matching results from large to small, which can be applied to the optimal screening in the case of the same type but different sizes, different specifications, etc. Description of the Drawings

[0027] Figure 1 It is the matching situation calculated and output by the system in the embodiment taking the board as an example. Detailed Embodiment

[0028] First of all, the present invention constructs an industrial material information data management system, which is used to establish the standardized management and information entry of the enterprise material library for subsequent efficient calculation and matching of target materials. The implementation process of the system includes:

[0029] S1. Establish master data business rules for material coding: Formulate master data business rules according to the source of materials, management level, basic data items, description and meaning explanation of the generation process of data items in the relevant business environment, the restrictive relationship between data, and the logical rules to be followed in the data generation process; the master data business rules include the coding specifications, classification rules and naming rules of each data item of the master data, and establish a material coding template;

[0030] In this step, we can establish the coding rules for materials according to the user's requirements, or classify them according to industrial industry standards and the user's actual inventory materials, including constants, classification codes, and sequential codes. Combining with the industrial field industry standards, industrial materials are divided into 24 major categories. For example, it includes metal materials and products (10), fasteners (11), seals (13), pipes and pipe fittings (14), general mechanical fittings (15), instruments and meters (16), electrical components (17), electrical engineering materials (18), electronic industrial products (19), oil and gas raw materials (20), chemical raw materials (21), welding materials and flaw detection supplies (22), non-metallic materials and products (24), chemical reagents and auxiliary materials (25), product-specific purchased parts and fittings (40), equipment maintenance fittings (42), cutting tools, measuring tools, and abrasive tools (50), labor protection and safety supplies (52), office cleaning and daily necessities (53), packaging materials and supplies (54), hardware and products (55), decoration materials (56), semi-finished products (60), virtual keys (89).

[0031] Based on the above major categories, for example, the coding identification of fasteners is 11. Under the classification of fasteners, it further includes bolts 1102, nuts 1104, studs 1106, screws 1008, rivets 1110, washers 1112, retaining rings 1114, pins 1116, and keys 1118. The coding of studs consists of 11 (major category) and 02 (medium category) to form 1102. Considering the simplicity of classification, we make the minor categories into templates, which is equivalent to a bottom-level classification containing multiple templates. For example, under bolts, it includes types of bolts such as GB5783 (hexagon head bolts with full thread) and GB5782 (hexagon head bolts). Template management mainly plays a role in connecting the upper and lower levels. It connects with the material classification at the upper level and corresponds to the template attributes at the lower level. The advantage of this is to make similar classifications into templates, reduce classifications, complete the creation of relevant attributes, formulate markings for subsequent material coding attributes, and also facilitate user memory.

[0032] S2. Master data management: Before applying for a material code, first query the material, and retrieve it according to the attribute fields of the queried material; if the corresponding material cannot be retrieved, classify the applied material and apply for a code using the material coding template. After selecting the template, enter the attribute area of the template. The template has already defined the value list and cascading relationship, and select the corresponding value set in the template according to the specific attributes of the applied material; if the code already exists, the system will prompt the existing code and its information. If the code does not exist, enter the code application management interface to complete the relevant application information.

[0033] S3. Data cleaning: Import the data from the original material database into the cleaning table using the DataX tool for cleaning. Before importing into the cleaning table, select the corresponding classification and template, match with the standard data through an Excel table, correct and improve relevant attributes, and after completion of cleaning, import into the system for verification; if the verification is successful, conduct coding review, and if the data fails the verification, clean it again until the verification is successful.

[0034] S4. Material query: The query of materials includes fuzzy query and precise query, and the query methods include attribute query, material description query, and material range query. In this step, the method described in the present invention is executed to calculate the material similarity.

[0035] In this embodiment, taking steel pipes as the required materials, the matching and similarity calculation in the inventory materials are described in combination with the material attributes.

[0036] Table 1. Accurate information of required materials

[0037]

[0038] Table 2. Information of existing inventory materials

[0039]

[0040] Table 3. Schematic diagram of similarity calculation matching corresponding to the method described in the present invention

[0041]

[0042] As shown in Tables 1 - 3, the similarity calculation of the material attributes in the present invention includes:

[0043] (1) Numerical attribute calculation

[0044] (1.1) Material

[0045] Considering the material attribute (steel) of the material, the inventory materials contain the same material, and at the same time, it is refined according to the steel material, and different materials are output for replacement or selection according to actual business needs.

[0046]

[0047] (1.2) Specification

[0048] For the thickness of steel pipe materials, including outer diameter, inner diameter, height / length.

[0049] The closer the wall thickness value and pipe diameter of the pipe material are to the target value, the higher the matching degree. In the case of the same specification, the matching degree is 100%.

[0050]

[0051] (1.3) Diameter pipe length

[0052] The closer the diameter pipe length is to the target value, the higher the matching degree. Under the same specifications, the matching degree is 100%.

[0053]

[0054] It should be noted that: The matching degree processing in the above table includes empirical assignment and / or calculation based on the Euclidean distance method. Empirical assignment refers to analyzing and giving the similarity of two materials in a certain attribute based on business needs and expert experience. For example, if the determined pipe diameter length of the required material is 4800, a length of about 4800 is used for similarity assignment. For a length of 4900 (with a difference of 100 from the required value), the assignment is 95%, and for a length of 4700 (with a difference of 100), the assignment is also 95%. In addition, the calculation of the Euclidean distance is used for the processing of clear numerical data and can also be applied to the calculation of n attributes. The specifications of the proportional pipe materials are recorded as (outer diameter, inner diameter, length) as (168.3, 11, 4800).

[0055] The numerical attributes in the required material attributes are "material", "specification", and "diameter pipe length". For the convenience of calculation, they are represented by the letters C, G, and Z respectively, where:

[0056] The similarity degrees of the three numerical attributes of material b and material c are as follows:

[0057] Sim(C) b = 0.3 × w C = 0.3 × 0.95 = 0.285

[0058] Sim(C)c = 0.3 × w C = 0.3 × 0.8 = 0.24

[0059] Sim(G) b = 0.3 × (w1 + w2) = 0.3 × (0.6 × 1 + 0.4 × 0.9) = 0.288

[0060] Sim(G)c = 0.3 × (w1 + w2) = 0.3 × (0.6 × 0.7 + 0.4 × 0.85) = 0.228

[0061] Sim(Z) b = 0.2 × w Z = 0.2 × = 0.2

[0062] Sim(Z)c = 0.2 × w Z = 0.2 × 0.8 = 0.16

[0063] (2) Text attribute calculation

[0064] The "General Requirements" in the above Table 1 - Table 3 are the keyword information, and the text attribute similarity calculation (2.1) determines the word frequency vector.

[0065] Among the text attributes of the required materials, there is only the item of "General Requirements". For the convenience of calculation, the purchased required materials and inventory materials are represented by letters a, b, and c in sequence, and calculations are carried out separately.

[0066] Taking all the requirement items as the vector length and using the word frequency to measure the vector, the word frequencies of the general requirements for each material are as follows:

[0067] Material a: Good plasticity - 1, Excellent welding performance - 1, Outstanding hot workability - 1, Certain thermal conductivity - 1, Ability to withstand a certain degree of cold bending deformation - 0

[0068] Material b: Good plasticity - 1, Excellent welding performance - 1, Outstanding hot workability - 1, Certain thermal conductivity - 0, Ability to withstand a certain degree of cold bending deformation - 1

[0069] Material c: Good plasticity - 1, Excellent welding performance - 1, Outstanding hot workability - 0, Certain thermal conductivity - 0, Ability to withstand a certain degree of cold bending deformation - 0

[0070] In the calculation programming, its word frequency vector can be expressed in Python code as:

[0071] a1 = np.array([1, 1, 1, 1, 0])

[0072] b1 = np.array([1, 1, 1, 0, 1])

[0073] c1 = np.array([1, 1, 0, 0, 0])

[0074] (2.2) Cosine angle calculation

[0075] The cosine value of the angle based on the "General Requirements" attribute is:

[0076]

[0077] To sum up, the similarity between seamless steel pipe a and seamless steel pipe b is:

[0078] Sim(a, b) = Sim(a, b)1 × 0.2 + Sim(C) b + Sim(G) b + Sim(Z) b

[0079] = 0.75 × 0.2 + 0.285 + 0.288 + 0.2

[0080] = 0.9230

[0081] Sim(a, c) = Sim(a, c)1×0.2 + Sim(C) c + Sim(G) c + Sim(Z) c

[0082] = 0.707×0.2 + 0.24 + 0.228 + 0.16

[0083] = 0.7694

[0084] That is, the similarity between the seamless steel pipe b in the existing inventory and the seamless steel pipe a in the procurement demand is 92.3%; the similarity between the seamless steel pipe c in the existing inventory and the seamless steel pipe a in the procurement demand is 76.94%.

[0085] The above embodiments give the calculation process of the numerical attributes (material, specification, diameter pipe length) and text attributes (calculating similarity after keyword assignment) of materials, and obtain the similarity calculation structure of the target material and the inventory material. For this result, we can realize the verification and matching of the incoming materials and the existing inventory materials, and can also be used for the rapid screening and matching of the required materials and the existing inventory materials to find the optimal target material (the closest residual material). Based on this result, we can also appropriately adjust the requirements of the business design to meet the use of the existing materials, realize two-way selection, improve the utilization rate of the existing inventory materials and the secondary reuse of the residual materials, and at the same time take into account the optimal design and operation management costs of the enterprise.

[0086] Taking the sheet as an example: the smaller the distance between the sheet thickness value and the target thickness value, the higher the matching degree, and the larger the number after the decimal point. Under the same material, if the thickness value is the same as the target value, the matching degree is 100%. The closer the thickness value is to the target value, the closer it is to 100%. On the contrary, under the same material, it is closer to 99%. The tentative range of the sheet thickness is: t - 2 to t + 6 (t represents the bottom thickness, and t - 2 is greater than zero). Under the same material, the matching degrees of t - 1 and t + 1 are the same, and the matching degrees of t - 2 and t + 2 are the same, and so on. As long as x (the number added or subtracted to t) satisfies 2 > x >= 0, the matching degree of t - x is equal to the matching degree of t + x. The system calculates the matching value as Figure 1 shown.

[0087] Based on the above examples, those skilled in the art should understand that this method can comprehensively consider the influence of multiple attributes on the similarity of materials, and can also handle the situation of mixed text and numerical attributes well, giving system matching and recommendation results. And in the actual process, it can also adjust the preset similarity weights of different material attributes according to specific business requirements and scenario requirements to achieve the most matching results of two-way selection between the required materials and the existing materials in stock. Since it is sensitive to the allocation of preset similarity weights for attributes, it is necessary for material managers to determine the preset similarity weights of attributes in advance in combination with the opinions of material experts when entering them into the system for the first time.

Claims

1. A method for encoding incoming materials and screening and matching inventory, characterized in that, The method calculates the similarity of the material attributes of the inventory materials. The material attributes include text attributes and numerical attributes. The text attributes refer to the keyword text information including the material name, material, etc., and the numerical attributes refer to the digital information including the material size specifications; Calculation of the similarity of numerical attributes: Let the weight of the numerical attribute x be w x , for the required material A, the matching degree of the inventory material B on this numerical attribute is x b , define the similarity of the material numerical attribute as: Sim(A,B) 数值 = w x x b Calculation of the similarity of text attributes: First, assign values to the keyword information in the text attributes of the materials, then construct a material attribute vector with several text attribute values, and calculate the similarity by calculating the cosine similarity between two material attribute vectors. Specifically: Assume the attributes of the required material A The existing attributes of material B Then their cosine similarity is: For more than one numerical attribute, different weights are assigned according to the importance of the attributes, and then the similarity of each material attribute is calculated separately, and the weighted attribute similarities are summed to obtain the overall similarity of the materials, thereby realizing the screening and matching of inventory materials; The final matching degree of the materials is: Sim(A,B) = Sim(A,B) 文本 + Sim(A,B) 数值 .

2. The material warehousing coding and inventory screening and matching method according to claim 1, wherein In the calculation of numerical attributes, the processing of the matching degree includes empirical assignment and / or calculation based on the Euclidean distance method; For the material, there are n numerical attributes. For the i-th numerical attribute, the numerical attribute value of the required material is denoted as A 0i , and the numerical attribute value of the inventory material is B ji . By using the Euclidean distance method to calculate the matching degree: Then the overall matching degree of the inventory material B relative to the required material A under this attribute is:

3. The method for encoding incoming materials and screening and matching inventory according to claim 1, characterized in that, The calculation of the similarity of text attributes includes processing the calculation of word frequency vectors based on Python code.

4. The material warehousing coding and inventory screening and matching method according to claim 1, wherein The method outputs the matching degree between the candidate materials and the inventory materials according to the similarity calculation results of several related inventory materials according to the size of the similarity.

5. The material warehousing coding and inventory screening and matching method according to claim 1, characterized in that The method for the material coding is as follows: Formulate the master data business rules according to the source of the materials, management level, basic data items, description and meaning explanation of the data item generation process in the business environment, the restrictive relationship between data, and the logical rules to be followed in the data generation process; the master data business rules include the coding specifications, classification rules and naming rules of each data item of the master data, and establish a material coding template.

6. The material warehousing coding and inventory screening and matching method according to claim 5, characterized in that The coding rules are classified according to industrial industry standards and / or the actual inventory materials of users, including constants, classification codes and serial numbers. The constants refer to numbers or values that do not need to be modified, including the letter prefix added before the self-made part code of the materials for the management of material attributes.

7. An industrial material information-based data management system, characterized in that, The system performs material warehousing coding and inventory matching screening according to the method described in any one of the above claims 1-6.

Images