WO2022252061A1

WO2022252061A1 - Knowledge-based assembly process planning method, apparatus and system

Info

Publication number: WO2022252061A1
Application number: PCT/CN2021/097439
Authority: WO
Inventors: 陈雪
Original assignee: 西门子（中国）有限公司
Priority date: 2021-05-31
Filing date: 2021-05-31
Publication date: 2022-12-08
Also published as: CN117355852A

Abstract

The present invention provides a knowledge-based assembly process planning method. The method comprises the following steps: S1, receiving a product design material list, acquiring knowledge information in the product design list, generating a topology model instance on the basis of a topology model template, and matching in a database a historical design material list similar to the product design material list; S2, retrieving a historical assembly plan corresponding to the similar historical design material list from the database, replacing nodes in the historical assembly plan with nodes corresponding to the product design material list, removing the nodes in the historical assembly plan that are not related to the historical design material list and corresponding process steps thereof, and reconstructing same into a product assembly plan; and S3, retrieving a historical assembly line corresponding to the corresponding historical assembly plan from the database, removing workstations corresponding to the process steps removed by step S2 in the historical assembly line, and reconstructing same into a product assembly line. The present invention is convenient to use, highly efficient and intelligent.

Description

Knowledge-based assembly process planning method, device and system

technical field

The invention relates to the digital field, in particular to a knowledge-based assembly process planning method, device and system.

Background technique

In the era of Industry 4.0, many manufacturing plants are facing more and more customer customization needs. How to respond quickly to such changing requirements by providing process planning and resource planning accordingly is the decisive question.

Traditionally, production line planning was performed manually by engineers using their judgment and experience. Some factories define standard process procedures as a starting point to cover customization for a specific product family.

The solutions of the above-mentioned prior art are all highly dependent on the professionalism of people, and are prone to errors and heavy workload. For complex products, the number of possible assembly sequences can be so large that even the most skilled engineer can miscalculate in finding the best possible sequence. In addition, the above solution cannot utilize a large amount of accumulated historical process planning data of historical customization requirements, which will lead to waste of information resources.

For example, the prior art provides a hierarchical method for planning the assembly sequence of connectors. Specifically, the principle is to generate a connector structure hierarchy from the CAD model, then decompose the connector hierarchy hierarchically, retrieve similar connector structures in the database and refer to their assembly plans, and finally merge the socket structure from top to bottom Assembly planning for nodes. The limitation of this scheme is that when retrieving similar connector structures in the database, retrieving historical reference data needs to use the same name to retrieve the same part, and only compares the topological similarity without quantitative comparison.

For another example, compared with the hierarchical scheme of the connector structure, the prior art also provides a graph similarity method to compare the assembly graphs of two production lines. Specifically, the connection information between components and/or sub-assemblies at the edge of the assembly map is described in principle. Graph similarity algorithms for similar assembly retrieval include graph matching (topological) and similarity testing (quantitative). The limitation of this scheme is that compared with the connector-level scheme, the assembly graph has more nodes, and graph matching may require a greater amount of computation.

Contents of the invention

The first aspect of the present invention provides a knowledge-based assembly process planning method, which includes the following steps: S1, receiving the product design material list, obtaining the knowledge information in the product design list and generating a topology model instance based on the topology model template, in matching the historical design bill of materials similar to the design bill of materials of the product in the database; S2, fetching the historical assembly plan corresponding to the similar historical design bill of materials from the database, and converting the nodes in the historical assembly plan to replacing with nodes corresponding to the product design bill of materials, removing nodes and their corresponding process steps in the historical assembly planning that are not related to the historical design bill of materials and reconstructing the product assembly planning; S3, from The historical assembly line corresponding to the corresponding historical assembly plan is retrieved from the database, and the workstations corresponding to the process steps removed by the step S2 in the historical assembly line are removed and reconstructed into a product assembly line.

Further, the step S1 also includes the following steps: S11, receiving the product design bill of materials, analyzing the data structure of the product design bill of materials based on the topology model template, acquiring the knowledge information in the product design bill and Generate a topology model instance; S22, retrieve the historical design material list from the database, and convert the product design material list and the historical design material list into a product CBS map and a historical CBS map respectively; S23, match the historical CBS map according to the product CBS map , when the product CBS graph matches multiple historical CBS graphs, the Euclidean algorithm is used to calculate the index difference between the product CBS graph and the nodes corresponding to the multiple historical CBS graphs to perform similarity measurement, so that in Selecting a similar CBS spectrum with the smallest difference in the indicators among the plurality of historical CBS spectrums.

Further, it is characterized in that the step S23 also includes the following steps: hierarchically matching historical CBS graphs according to product CBS graphs, when a part of product CBS graphs matches multiple historical CBS graphs, calculate by Euclidean algorithm The product CBS map and the index difference between the nodes corresponding to multiple historical CBS maps are used to perform similarity measurement, and the matched and unmatched parts of the product CBS map are deconstructed and the product CBS map is continued. The unmatched part of the product continues to perform matching and similarity measurement until it is completely matched or there is one unmatched node left. When there is one unmatched node left in the product CBS map, the standard CBS map node in the database is called to match the node. Perform a match.

Further, the nodes of the topology model include various types of entities, the entities include assemblies, subassemblies, parts, and connectors, and the nodes of the CBS map are connectors and multiple assemblies connected to them, Subassemblies, parts, connectors.

Further, the step S22 also includes the following steps: converting the nodes of the topology model into the nodes of the CBS map, so as to convert the product design material list and the historical design material list into a product CBS map and a historical CBS map respectively Atlas.

Further, the database stores design material lists, standard CBS diagrams, historical assembly planning, topology model templates, and historical assembly lines.

A second aspect of the present invention provides a knowledge-based assembly process planning system, comprising: a processor; and a memory coupled to the processor, the memory having instructions stored therein that when executed by the processor use The electronic device performs an action, and the action includes: S1, receiving a list of product design materials, acquiring knowledge information in the product design list and generating a topology model instance based on a topology model template, and matching with the design material of the product in the database list a similar historical design material list; S2, retrieve the historical assembly plan corresponding to the similar historical design material list from the database, and replace the nodes in the historical assembly plan with the product design material list For the corresponding nodes, remove the nodes in the historical assembly plan that are not related to the historical design bill of materials and their corresponding process steps, and reconstruct them into product assembly plans; S3, retrieve the corresponding nodes from the database The historical assembly line corresponding to the historical assembly plan, removing the workstations corresponding to the process steps removed by the action S2 in the historical assembly line and reconstructing it as a product assembly line.

Further, the action S1 also includes the following steps: S11, receiving the product design bill of materials, analyzing the data structure of the product design bill of materials based on the topology model template, acquiring the knowledge information in the product design bill and Generate a topology model instance; S22, retrieve the historical design material list from the database, and convert the product design material list and the historical design material list into a product CBS map and a historical CBS map respectively; S23, match the historical CBS map according to the product CBS map , when the product CBS graph matches multiple historical CBS graphs, the Euclidean algorithm is used to calculate the index difference between the product CBS graph and the nodes corresponding to the multiple historical CBS graphs to perform similarity measurement, so that in Selecting a similar CBS spectrum with the smallest difference in the indicators among the plurality of historical CBS spectrums.

Further, the action S23 also includes: hierarchically matching the historical CBS graph according to the product CBS graph, and when a part of the product CBS graph matches multiple historical CBS graphs, calculating the product CBS graph through the Euclidean algorithm respectively Perform similarity measurement based on the index difference between nodes corresponding to multiple historical CBS graphs, deconstruct the matched and unmatched parts of the product CBS graph and continue to unmatched parts of the product CBS graph Perform matching and similarity measurement until there is a complete match or there is one unmatched node. When there is one unmatched node in the product CBS graph, the standard CBS graph node in the database is called to perform matching on the node.

The third aspect of the present invention provides a knowledge-based assembly process planning device, wherein the knowledge-based assembly process planning device further includes: a similarity retrieval device, which receives a product design material list and acquires the knowledge in the product design list information and generate a topology model instance based on the topology model template, and match the historical design bill of materials similar to the design bill of material of the product in the database; assembly planning generating means, which transfers the similar historical design bill of material from the database For the corresponding historical assembly planning, replace the nodes in the historical assembly planning with nodes corresponding to the product design bill of materials, remove the nodes in the historical assembly planning that are not related to the historical design bill of materials and The corresponding process steps are reconstructed into a product assembly plan; an assembly line generation device, which retrieves the historical assembly line corresponding to the corresponding historical assembly plan from the database, and removes the historical assembly line from the historical assembly line by the assembly plan generation device The workstations corresponding to the removed process steps are restructured into product assembly lines.

A fourth aspect of the present invention provides a computer program product tangibly stored on a computer-readable medium and comprising computer-executable instructions which, when executed, cause at least one processor to perform the The method described in the first aspect of the present invention.

A fifth aspect of the present invention provides a computer readable medium having stored thereon computer executable instructions which, when executed, cause at least one processor to perform the method according to the first aspect of the present invention.

The present invention uses the semantic graph as the carrier for describing assembly, assembly line and assembly planning, which has strong descriptive ability, easy to understand, easy to apply inference rules, and its entity naming is not limited. Therefore, the knowledge-based assembly process planning device provided by the present invention easy to use.

Moreover, the present invention combines topological and quantitative methods to perform graph similarity measurement, wherein, ontology comparison is the first rough filtering, and quantitative similarity measurement is a precise filtering based on the first rough filtering, so computing Quantity decreased. Quantitative measures increase the accuracy of similarity comparisons and thus increase recommendation accuracy compared to pure topological comparisons. Therefore, the present invention is more efficient and more precise.

In addition, the knowledge in the database is accumulated over a long period of time as historical data, so the recommendation ability is increased, and the present invention is more intelligent and self-evolving.

Description of drawings

Fig. 1 is a schematic diagram of a topology model according to a specific embodiment of the present invention;

Fig. 2 is a schematic diagram of the relationship between a product and a process topology model according to a specific embodiment of the present invention;

Fig. 3 is a schematic diagram of the recommended assembly planning and assembly line in the product design bill of materials according to a specific embodiment of the present invention;

4 is a schematic structural diagram of a knowledge-based assembly process planning system according to a specific embodiment of the present invention;

Fig. 5 is a schematic diagram of converting the design bill of materials into a CBS structure according to a specific embodiment of the present invention;

Fig. 6 is a schematic diagram of database data according to a specific embodiment of the present invention;

Fig. 7 is a schematic diagram of converting a product design bill of materials into a CBS structure according to yet another specific embodiment of the present invention;

Fig. 8 is a schematic diagram of a list of historical design materials according to another specific embodiment of the present invention;

Fig. 9 is a schematic diagram of hierarchical deconstruction and map matching according to a specific embodiment of the present invention;

Fig. 10 is a schematic diagram of similarity comparison according to a specific embodiment of the present invention.

Detailed ways

The specific embodiments of the present invention will be described below in conjunction with the accompanying drawings.

The present invention provides a knowledge-based assembly process planning (assembly process planning) method, device and system, which reduces the dependence on human expertise and increases planning quality and efficiency. The invention greatly reduces the calculation amount of database retrieval, provides a way of knowledge expression, and the construction of knowledge base is also more convenient. Furthermore, the present invention enables quantitative similarity measurement with rich geometric attributes.

When performing assembly process planning using the present invention, what is important is the product structure and relationships between components, not the overall shape of the assembly model. Wherein, similar connections correspond to similar assembly process steps. That is, two assemblies may not be very related and have a similar assembly process even though their overall visual appearance is not quite the same.

For manufacturers who typically have customization requests, they may have accumulated historical assembly plans for a large number of customized products. Some of these products belong to the same product type, only some parts are different in appearance, including the number of parts, materials and decorations, etc. For example, some electronic control cabinets that include doors include glass windows, some control cabinets have pure metal doors, some control cabinets have 10 internal switches, and some control cabinets only have 8 internal switches. Therefore, for new customer needs, if these historical data can be used as a reference for new assembly process planning, efficiency and accuracy will be greatly improved.

The first aspect of the present invention provides a knowledge-based assembly process planning method.

Fig. 4 is a schematic structural diagram of a knowledge-based assembly process planning system according to a specific embodiment of the present invention. The knowledge-based assembly process planning system 200 includes a similarity retrieval device 210 , an assembly planning generating device 220 and an assembly line generating device 230 . Wherein, the similarity retrieval device 210 is used to decompose the product design material list Q based on the new customer demand, and use the product design material list Q to search the knowledge database KD (Knowledge Database). Specifically, the similarity retrieval device 210 includes a history extraction device 211 , a graph structure construction device 212 and a similarity measurement device 213 . The assembly plan generating means 220 includes a plan extracting means 221 , a node replacing means 222 and an assembly plan forming means 223 . The assembly line generating device 230 includes a planning line extracting device 231 , a workstation adding and removing device 232 and an assembly line forming device 233 .

First, step S1 is executed. The similarity retrieval device 210 receives the product design material list, acquires the knowledge information in the product design list, generates a topology model instance based on the topology model template, and matches the history similar to the product design material list in the database. Design bill of materials.

When receiving a new product order, the product designer usually provides a CAD model and its design material list (EBOM, Engineering Bill of Materials). The design bill of materials should have all parts, sub-assemblies (sub-assemblies), hierarchical structures (hierarchical structures) and connections between the above parts.

Among them, the topology model (Ontology model) of the assembly structure will be stored in the knowledge base (Knowledge Base) after it is generated, which conforms to the ISO10303 standard. ISO10303 standard part 44 provides some limited assembly design representations (assembly design representations), which capture the assembly structure and dynamic joint information (kinematic joint information) in the design process. Specifically, the assembly model establishes a neutral representation of the product assembly, which includes multiple sets of components. In this model, a complete product is called an "assembly", the lowest-level inseparable components in the assembly are called "parts", and two or more of the parts assembled together are called "subassemblies". body". The model focuses on product hierarchies, and the location and orientation between parts.

As shown in FIG. 1 , the topology model 100 includes concepts, parts, sub-assemblies and connectors related to the product and its assembly. Among them, the connector type corresponds to a specific assembly operation (assembly operation). The relationship between the above concepts, the data properties of each concept are described as attributes. Specifically, the topology model 100 includes an assembly, a subassembly 1 , a subassembly 2 , a subassembly 3 , a subassembly 4 , a subassembly 5 , a part 1 , a part 2 , a part 3 and a part 4 . The circles in FIG. 1 indicate the relationship between the above components.

Fig. 2 is a schematic diagram of the relationship between product and process topology models according to a specific embodiment of the present invention. As shown in FIG. 2 , a portion 310 of a topological model of a product includes assemblies, parts, subassemblies, connectors, and connectors. Among them, parts and subassemblies are connected by connectors. Connections include glue bonding, bolting, press fitting and snap fitting. Process 320 includes assembly operations including gluing, bolting, press-fitting, and snap-fitting operations. Wherein, the above-mentioned processes and connectors correspond one-to-one, and the relationship between the processes and connectors is usage. Specifically, "glue bonding" is used for "glue bonding operation", "bolt connection" is used for "bolt operation", "press fitting" is used for "press fitting operation", and "snap fitting operation" is used for "snap fitting operation". ".

Specifically, each design BOM information is described as an RDF or LGP graph, which conforms to the topology model shown in FIG. 2 . The present invention will try to find an assembly similar to the new customized assembly, and generate a new assembly plan by modifying the assembly plan of the historical similar assembly.

Specifically, step S1 includes sub-step S11, sub-step S12 and sub-step S13.

In sub-step S11, the history extraction device 211 receives the product design bill of materials, analyzes the data structure of the product design bill of materials based on the topology model template, obtains the knowledge information in the product design bill, and generates a topology model instance based on the topology model template . Specifically, the history extracting means 211 is used to search the list of historical design materials in the graph database, and the list of historical design materials is described by RDF graph or LGP graph. If the graphs in the database are classified by product category, since the products involved in the product design material list are considered to belong to a specific category, the history extracting means 211 only needs to perform a search in this category. Otherwise, all design BOM patterns would need to be searched through the entire database. Among them, the graph retrieval is identified in the application program interface in the graph database.

Fig. 3 is a schematic diagram of the recommended assembly plan and assembly line in the product design bill of materials according to a specific embodiment of the present invention. As shown in Figure 3, in the product design bill of materials G1, node a is an assembly, node B is a part, node c is a subassembly, and c1, c2, and c3 are parts or connectors. The topological model of the product and the product design BOM G1 are sent to the similarity retrieval device 210, and the similarity retrieval device 210 searches the database D for historical design BOMs similar to the product design BOM.

Wherein, the database KD includes historical design material list, assembly plan and assembly line. In the linked assembly planning graph, P1, P2, and P3 nodes are process step nodes, and material nodes are connected to process nodes, which represent that materials can be consumed or manufactured in the process. Preferably, the material node is a material flow node. In addition to materials, process nodes also have some data values to describe their properties, such as process time and process name. Furthermore, in the linked assembly line graph, WS nodes represent workstations, and connections between WS nodes represent the sequence of process steps that occur at the workstations. Each WS node has some data values to describe its attributes, such as height, width, length, cost, workstation name and workstation number, etc.

In sub-step S22, the map structure construction device 212 retrieves the historical design material list from the database KD, and converts the product design material list and historical design material list into a product CBS map and a historical CBS map respectively. The step S22 also includes the following step: converting the nodes of the topology model into the nodes of the CBS map, so as to convert the product design material list and the historical design material list into a product CBS map and a historical CBS map respectively.

Specifically, the graph structure construction device 212 is used to convert the RDF or LPG graph in the product design bill of materials and historical design bill of materials in the database into a connector-based structure (CBS, Connector-Based Structure) graph, referred to as a CBS graph. Among them, the structure based on the connector is described as C[P1, P2...Pn] format, C represents the name of the connector, P1, P2...Pn represents the part or subassembly connected by the connector C. Connectors and their parts or subassemblies can be named by any name desired by the designer. If there are more than one connector of the same type, the present invention combines them into a connector group, and the connector group is represented by (C1, C2...Cn)[P1, P2...Pn]. For example, 4 screws are required to secure a cover and box.

Exemplarily, as shown in FIG. 5 , the design bill of materials 410 may be a product design bill of materials or a historical design bill of materials. Design bill of materials 410 includes subassembly B, connector b, connector S1, connector R1, connector R2, part N, part N1, part N2, subassembly N3, subassembly N4, part N5, and part N6 , the aforementioned parts, subassemblies, and connectors are represented as nodes. The graph structure constructing device 212 converts the product design bill of materials 410 into a CBS graph 420 . For example, the part N1 is a table leg, the connector S1 is a screw, and the part N2 is a table top. In the design bill of materials 410, the part N1, the connector S1 and the part N2 are all nodes, but in the CBS diagram 420 they will be converted into a node S1 [N1,N2]. Similarly, in the CBS diagram 420, connector b, subassembly B and part N are converted to node b[B,N], connector R1, subassembly N3 and subassembly N4 are replaced by node R1[N3, N4], link R2, part N5 and part N6 are transformed into node R2[N5,N6].

The present invention converts nodes such as products, assemblies, parts, sub-assemblies, and connectors in the design material list into the form of connectors and the above-mentioned components connected in the CBS diagram, reducing the number of nodes and the amount of calculation.

In sub-step S23, the similarity measurement device 213 matches the historical CBS graphs according to the product CBS graphs. The similarity measurement is performed based on the index difference between the nodes corresponding to the CBS graphs, so as to select a similar CBS graph with the smallest index difference among the multiple historical CBS graphs.

Specifically, the step S23 also includes the following steps: Hierarchically matching the historical CBS graph according to the product CBS graph, when a part of the product CBS graph matches a plurality of historical CBS graphs, calculate the product CBS graph through the Euclidean algorithm The index difference between the graphs and the nodes corresponding to multiple historical CBS graphs is used to perform similarity measurement, deconstruct the matched and unmatched parts of the product CBS graph and continue to analyze the unmatched parts of the product CBS graph Partially continue to perform matching and similarity measurement until there is a complete match or there is one unmatched node. When there is one unmatched node in the product CBS map, the standard CBS map node in the database is called to perform matching on the node.

Wherein, the similarity measurement device 213 is used to perform hierarchical map similarity measurement using a map similarity algorithm, wherein the map similarity algorithm includes two incremental steps: a map matching step and a similarity measuring step.

The following describes the spectrum matching principle of the present invention. Graph matching uses the VF2 graph isomorphism algorithm (VF2 graph isomorphism algorithm), plus semantics to use similar topological structures to query historical design bill of materials graphs. When executing the graph isomorphism algorithm, node types are assigned to serve as input attributes for the algorithm. At this time, the same connector nodes do not need to use the same name, they can also be identified as the same. Since semantically they belong to the same type of connectors as long as they belong to the same type in the topology model.

The hierarchical structure principle of map matching is as follows. If the query based on the connector structure map cannot find a match in the historical database at one time, it is necessary to structure the structure hierarchically and perform a search for the match.

In one embodiment of the present invention, the product design BOM graph and the historical design BOM graph (historical EBOM graph) in the database cannot be judged to be isomorphic by comparing their connector structure graphs. Assuming that the query CBS map cannot be matched to the historical CBS map, it is necessary to deconstruct and perform the comparison recursively as shown in the figure.

For each structure, the graph becomes two parts by removing the connection between the upper node and the rest of the graph. For each part, a matching map or sub-map is searched in both databases. Some known CBS structures and their assembly plans are stored in a standard database. The historian stores subassembly-level or product-level CBS structures and their plans. If similar structures are found, their existing assembly plans can be extracted to serve as a starting point for the product's assembly process planning.

If there are still some CBD nodes that cannot be matched in the database after the lowest structure, heuristic rules can be implemented to infer assembly features from basic geometric features, and then infer assembly operations from assembly features. Common-sense rules can be combined in the topology model of the SWRL language and driven by a reasoner. Exemplarily, the reasoner includes Pellet or Hermit. If it is difficult to generate rules from human experience, the CBS that cannot be matched will be executed manually, which will not affect the workflow of the entire system.

Specifically, in one embodiment, as shown in FIG. 7 , a new customer order requires assembly of an assembly as described in the product design BOM 510 . Among them, the product design material list 510 includes the following nodes: assembly, bolt, box base, backplane, HMI interface, bolt, shell assembly, PCBA, bolt, front assembly, PCBA1, PCBA2, card slot, PCB, LCD and nuts. Therefore, the above product design material list 510 is converted into a product CBS map 520, wherein the product CBS map 520 includes the following nodes: the first node N ₁₁ bolts [box base, back plate], the second node N ₁₂ bolts [HMI Interface, shell assembly], the third node N ₁₃ bolts [PCBA, PCBA], the fourth node N ₁₄ card slots [PCBA1, PCBA2], the fifth node N ₁₅ bolt connection [PCB, LCD].

At this point, as shown in Figure 8, a historical CBS map 530 is stored in the database KD, which includes a plurality of nodes: the first node N ₂₁ bolts [PCBA1, front assembly], the second node N ₂₂ bolt connections [PCB, LCD], third node N ₂₃ glue bonding [sheet, frame].

As shown in FIG. 9 , first try to perform the first round of matching on the product CBS graph 520 , and match the historical CBS graph 530 to a node first node N ₁₁ , specifically, the first node N 11 and the first node N ₁₁ of the product CBS graph 520 and The first node N ₂₁ of the historical CBS map 530 has been matched well, and the others have not been matched well. Therefore, the product CBS map 520 is deconstructed, and the first node N ₁₁ and the second node N ₁₂ , the second node N ₁₂ , The third node N ₁₃ and the second node N ₁₄ are disassembled. Then the second round of matching is performed on the unmatched part, and a node second node N ₁₂ is matched with the historical CBS graph 530, specifically, the second node N ₁₂ of the product CBS graph 520 and the first node of the historical CBS graph 530 N ₂₁ is matched well, but others are not matched well, so the product CBS map 520 is deconstructed, and the second node N ₁₂ is disassembled from the second node N ₁₂ , the third node N ₁₃ , and the second node N ₁₄ . Then, the third round of matching is performed on the unmatched part, and two nodes are matched with the historical CBS map 530: the third node N ₁₃ and the fifth node N ₁₅ . Wherein, the third node N ₁₃ and the fifth node N ₁₅ of the product CBS map 520 are respectively matched with the first node N ₂₁ and the second node N ₂₂ of the historical CBS map 530 . Wherein, the above matches are all matched with the historical CBS map in the database KD, and there is still a single node, the fourth node N ₁₄ , which has not been matched well.

Furthermore, the matching of a single node in the present invention usually invokes a standard CBS map node in the database, and the standard CBS map node is a single node. As shown in FIG. 9 , the database includes standard CBS map nodes N ₃₃ , N ₃₄ and N ₃₅ . Finally, in this embodiment, the single node fourth node N ₁₄ is matched to the standard CBS map node N ₃₃ .

The principle of similarity measurement is that if more than one match is found, components inside the CBS node need to further perform quantitative comparisons. FIG. 10 is a schematic diagram of similarity comparison according to a specific embodiment of the present invention. A product CBS graph 610 matches two

historical CBS graphs

620 and 630 . Specifically, the product CBS map 610 includes three nodes, namely D[B, K, E], s[k1, k2, k3] and b[E1, E2]. The two

historical CBS graphs

620 and 630 are similar in structure to the product CBS graph 610, wherein the historical CBS graph 620 includes three nodes S1[N1,N2], R1[N3,N4] and R2[N5,N6]. 630 includes three nodes d[x,y], s2[t1,t2] and p[k1,k2]. The node b[B, N] in the historical CBS map 620 is not matched.

In order to compare the

historical CBS graphs

620 and 630 which are more similar and match the product CBS graph 610 , it is necessary to calculate and execute the index difference between the corresponding nodes through the Euclidean algorithm. For example, it is necessary to compare the index differences between nodes D[B, K, E] and nodes S1[N1, N2] and d[x, y] respectively, and to compare node s[k1, k2, k3] with nodes Index difference between R1[N3,N4] and s2[t1,t2]. Optionally, the indicators include model number, features, dimensions, quality, materials, and the like.

Exemplarily, the distance between part B and part N1 can be calculated by the following algorithm:

Among them, Xi is an index, a and b refer to two corresponding components, and i is a natural number.

Then step S2 is executed, the history extracting means 211 retrieves the historical assembly plan corresponding to the similar historical design material list from the database, and the node replacement means 222 replaces the nodes in the historical assembly plan with the Designing the nodes corresponding to the BOM, the assembly plan forming unit 223 removes the nodes in the historical assembly plan that are not related to the historical design BOM and their corresponding process steps and reconstructs them into a product assembly plan. The assembly plan generation device 220 is used to generate assembly plans for new products, wherein the matched CBS graph structures can come from different assemblies, so there can be more than one assembly plan for reference. Therefore, it is first necessary to extract the assembly plans that match the CBS diagram of the product, then remove the unmatched parts of these assembly plans, and finally replace the parts and sub-assemblies for the matching parts of the parts/subassemblies in the product design bill of materials, and then in The materials in the product design bill of materials are combined with process parts according to the hierarchical relationship.

The assembly line generation device 230 is used to generate assembly lines for new products, wherein there are more than one assembly lines that can be referred to, so these assembly lines are first extracted with matching parts, and then the workstations without matching parts or redundant process steps are removed, and then in the process steps Merge these workstations according to the priority relationship. As shown in Figure 3, in this embodiment, the historical material list matched in step S1 is G2, which includes nodes A, B, C, D, E, c1, c2, c3, E1, and the historical material The historical assembly plan corresponding to list G2 is SP1. Wherein, the historical assembly planning includes process steps P1, P2 and P3, and the process steps P1 and P2 are executed before the process step P3. The material flow before process step P1 is described by node E1, the material flow after process step P1 is described by node E, the material flow before process step P2 is described by node E, and the material flow after process step P3 is described by node A. Description, the material flow before process step P2 is described by nodes c1, c2, c3, the material flow after process step P2 is described by node C, the material flow before process step P2 is described by nodes E, B, C, in The material flow after process step P2 is described by node A. Therefore, after removing the irrelevant nodes E1 and E and their related process steps, the product assembly plan SP2 is obtained.

Finally, step S3 is executed, the planned line extracting device 231 retrieves the historical assembly line corresponding to the corresponding historical assembly plan from the database, and the workstation adding and removing device 232 removes the process steps removed by the step S2 in the historical assembly line Corresponding to the workstation, the assembly line configuration device 233 is reconfigured into a product assembly line. As shown in FIG. 3 , the historical assembly line corresponding to the corresponding historical assembly plan SP1 is SL1 , wherein the assembly line includes workstations executed by each process step of the assembly plan. Specifically, the process step P1 is performed in the workstation WS1, the process step P2 is performed in the workstation WS2, and the process step P3 is performed in the workstation WS3. Therefore, the historical assembly line SL1 includes the workstations WS1, WS2 and WS2, wherein, the Said workstations WS1 and WS2 perform process steps P1 and P2, respectively, before said workstation WS3 performs process step P3. Therefore, after removing the workstation WS1 corresponding to the process step P1 removed in step S2, the product assembly line SP3 is obtained.

Further, the database stores design material lists, standard CBS diagrams, historical assembly planning, topology model templates, and historical assembly lines. As shown in FIG. 6, the standard CBS map is a node N33, which includes a card slot, and the card slot includes a box and a board. Therefore, the assembly plan SP3 generated for the standard CBS map by using the present invention includes a process step P, and the process step P is a clamping process. The material flow before the execution of the process step P is described by nodes N41 and N42, and the material flow after the execution of the process step P is described by the node N43, wherein, the node N41 is "a box with a slot", and the node N42 is "Board", the node N43 is "the box with the board connected".

In addition, the present invention also includes the following steps: storing the generated product design material list, product assembly plan and product assembly line in the database.

Moreover, the present invention combines topological and quantitative methods to perform graph similarity measurement, wherein the ontology comparison is the first rough filtering, and the quantitative similarity measurement is further refined filtering based on the first rough filtering, so computing Quantity decreased. Quantitative measures increase the accuracy of similarity comparisons and thus increase recommendation accuracy compared to pure topological comparisons. Therefore, the present invention is more efficient and more precise.

Although the content of the present invention has been described in detail through the above preferred embodiments, it should be understood that the above description should not be considered as limiting the present invention. Various modifications and alterations to the present invention will become apparent to those skilled in the art upon reading the above disclosure. Therefore, the protection scope of the present invention should be defined by the appended claims. In addition, any reference signs in the claims should not be construed as limiting the involved claims; the word "comprising" does not exclude other claims or means or steps not listed in the description; "first", "the first Words such as "two" are used to denote names only, and do not imply any particular order.

Claims

A knowledge-based assembly process planning method, which includes the following steps:

S1. Receive a product design bill of materials, acquire knowledge information in the product design bill, generate a topology model instance based on a topology model template, and match a historical design bill of materials similar to the product design bill of materials in the database;

S2. Retrieve the historical assembly plan corresponding to the similar historical design bill of materials from the database, replace the nodes in the historical assembly plan with nodes corresponding to the product design bill of materials, and remove the The nodes in the historical assembly plan that are not related to the historical design bill of materials and their corresponding process steps are reconstructed into a product assembly plan;

S3. Retrieve the historical assembly line corresponding to the corresponding historical assembly plan from the database, remove the workstations corresponding to the process steps removed by the step S2 in the historical assembly line, and reconstruct it into a product assembly line.
The knowledge-based assembly process planning method according to claim 1, wherein said step S1 further comprises the following steps:

S11. Receive a product design bill of materials, analyze the data structure of the product design bill of materials based on a topology model template, acquire knowledge information in the product design bill, and generate a topology model instance based on the topology model template;

S22. Retrieve the historical design material list from the database, and convert the product design material list and the historical design material list into a product CBS map and a historical CBS map respectively;

S23. Match the historical CBS graph according to the product CBS graph. When the product CBS graph matches multiple historical CBS graphs, calculate the relationship between the product CBS graph and the nodes corresponding to the multiple historical CBS graphs using the Euclidean algorithm. The similarity measurement is performed based on the index difference, so as to select a similar CBS map with the smallest index difference among the plurality of historical CBS maps.
The knowledge-based assembly process planning method according to claim 2, wherein said step S23 further comprises the following steps:

Hierarchically match the historical CBS graphs according to the product CBS graphs, and when a part of the product CBS graphs matches multiple historical CBS graphs, calculate the nodes corresponding to the product CBS graphs and the multiple historical CBS graphs respectively through the Euclidean algorithm The index difference between them is used to perform similarity measurement, deconstruct the matched and unmatched part of the product CBS map and continue to perform matching and similarity measurement on the unmatched part of the product CBS map until a complete match or a node is left unmatched,

When there is one unmatched node in the CBS map of the product, the standard CBS map node in the database is called to perform matching on the node.
According to the knowledge-based assembly process planning method according to claim 2, it is characterized in that, the nodes of the topology model include various types of entities, and the entities include assemblies, subassemblies, parts, and connectors, and the nodes of the CBS graph Nodes are connectors and the multiple assemblies, subassemblies, parts, and connectors they connect to.
According to the knowledge-based assembly process planning method according to claim 2, it is characterized in that said step S22 also includes the following steps: converting the nodes of said topology model into nodes of said CBS graph, so as to convert said product design bill of materials and The historical design bill of materials is transformed into product CBS map and historical CBS map respectively.
The knowledge-based assembly process planning method according to claim 1, wherein the database stores design material lists, standard CBS diagrams, historical assembly planning, topology model templates, and historical assembly lines.
Knowledge-based assembly process planning system, including:

processor; and

a memory coupled to the processor, the memory having stored therein instructions that when executed by the processor cause the electronic device to perform actions, the actions comprising:

S1. Receive a product design bill of materials, acquire knowledge information in the product design bill, generate a topology model instance based on a topology model template, and match a historical design bill of materials similar to the product design bill of materials in the database;

S2. Retrieve the historical assembly plan corresponding to the similar historical design bill of materials from the database, replace the nodes in the historical assembly plan with nodes corresponding to the product design bill of materials, and remove the The nodes in the historical assembly plan that are not related to the historical design bill of materials and their corresponding process steps are reconstructed into a product assembly plan;

S3. Retrieve the historical assembly line corresponding to the corresponding historical assembly plan from the database, remove the workstation corresponding to the process step removed by the action S2 in the historical assembly line, and reconstruct it as a product assembly line.
The knowledge-based assembly process planning system according to claim 7, wherein said action S1 further comprises the following steps:

S11, receiving the product design bill of materials, analyzing the data structure of the product design bill of materials based on the topology model template, obtaining the knowledge information in the product design bill and generating a topology model instance based on the topology model template;

S22. Retrieve the historical design material list from the database, and convert the product design material list and the historical design material list into a product CBS map and a historical CBS map respectively;

S23. Match the historical CBS graph according to the product CBS graph. When the product CBS graph matches multiple historical CBS graphs, calculate the relationship between the product CBS graph and the nodes corresponding to the multiple historical CBS graphs using the Euclidean algorithm. The similarity measurement is performed based on the index difference, so as to select a similar CBS map with the smallest index difference among the plurality of historical CBS maps.
The knowledge-based assembly process planning system according to claim 8, wherein said action S23 further comprises:

Hierarchically match the historical CBS graphs according to the product CBS graphs, and when a part of the product CBS graphs matches multiple historical CBS graphs, calculate the nodes corresponding to the product CBS graphs and the multiple historical CBS graphs respectively through the Euclidean algorithm The index difference between them is used to perform similarity measurement, deconstruct the matched and unmatched part of the product CBS map and continue to perform matching and similarity measurement on the unmatched part of the product CBS map until a complete match or a node is left unmatched,

When there is one unmatched node in the CBS map of the product, the standard CBS map node in the database is called to perform matching on the node.
According to the knowledge-based assembly process planning system according to claim 8, it is characterized in that the nodes of the topology model include various types of entities, and the entities include assemblies, subassemblies, parts, and connectors, and the nodes of the CBS graph Nodes are connectors and the multiple assemblies, subassemblies, parts, and connectors they connect to.
According to the knowledge-based assembly process planning system according to claim 8, it is characterized in that, said step S22 also includes the following steps: converting the nodes of said topology model into nodes of said CBS graph, so as to convert said product design bill of materials and The historical design bill of materials is transformed into product CBS map and historical CBS map respectively.
The knowledge-based assembly process planning system according to claim 7, wherein the database stores design material lists, standard CBS diagrams, historical assembly planning, topology model templates, and historical assembly lines.
The knowledge-based assembly process planning device is characterized in that the knowledge-based assembly process planning device also includes:

A similarity retrieval device, which receives the product design material list, acquires the knowledge information in the product design list and generates a topology model instance based on the topology model template, and matches the historical design material list similar to the product design material list in the database;

An assembly plan generating device, which retrieves the historical assembly plan corresponding to the similar historical design BOM from the database, and replaces the nodes in the historical assembly plan with nodes corresponding to the product design BOM , removing the nodes in the historical assembly plan that are not related to the historical design bill of materials and their corresponding process steps, and reconstructing it into a product assembly plan;

An assembly line generation device, which retrieves the historical assembly line corresponding to the corresponding historical assembly plan from the database, removes the workstations corresponding to the process steps removed by the assembly plan generation device in the historical assembly line, and restructures it into a product assembly line .
A computer program product tangibly stored on a computer-readable medium and comprising computer-executable instructions which, when executed, cause at least one processor to perform the one of the methods described.
A computer-readable medium having stored thereon computer-executable instructions which, when executed, cause at least one processor to perform the method of any one of claims 1-7.