US-12625489-B2 - Method for constructing topology reference architecture for a production line

US12625489B2US 12625489 B2US12625489 B2US 12625489B2US-12625489-B2

Abstract

A method for constructing a topology reference structure for a production line is provided. The present disclosure is based on historical production line topology data of an enterprise to extract a commonly used topology reference structure for a production line of the enterprise by a computer through a machine learning (ML) algorithm, so as to form a typical production line topology group of the enterprise. The present disclosure can record typical production line characteristics and production habits of the enterprise, realize reuse of a production line structure and production line construction knowledge, reduce the workload of production line designers, and improve the production line construction efficiency of the enterprise. In addition, the present disclosure avoids the interference of designers' subjective decisions to a certain extent, and the reference structure extracted by the computer has high reference value, and is objective, mature, and stable.

Inventors

Jiewu Leng
Jiwei GUO
Xiaofeng Zhu
Caiyu Xu
Hongye Su
Qiang Liu

Assignees

GUANGDONG UNIVERSITY OF TECHNOLOGY

Dates

Publication Date: 20260512
Application Date: 20230704
Priority Date: 20230303

Claims (7)

1 . A method for constructing a production line, comprising the following steps: S1: calculating a comprehensive similarity between a first production line x A and a second production line x B ; S2: calculating a similarity parameter S A,B between the first production line x A and the second production line x B based on the comprehensive similarity; S3: calculating a similarity parameter s i,j between each two production lines x l and x j in n historical production lines according to steps S1 and S2, and forming a fuzzy compatibility matrix S of the n historical production lines; S4: constructing a multi-granularity quotient space based on the fuzzy compatibility matrix S; S5: selecting an optimal granular layer from the multi-granularity quotient space; and S6: constructing the topology reference structure for a production line for the production line based on the optimal granular layer; S7: constructing the production line based on the topology reference structure for the production line; wherein the calculating the comprehensive similarity between the first production line x A and the second production line x B , specifically comprises: S11: calculating a matching degree and similarity between a first device V A,j in the first production line x A , and a second device V B,j in the second production line x B , in terms of four properties, to obtain a comprehensive similarity between the first device VAI and the second device V B , j wherein, the four properties comprise material, production process, product category, and production quality; S12: calculating the comprehensive similarity S act (V A,j , V B,j ) between each pair of devices from the first production line x A and the second production line x B , according to S11; and calculating the comprehensive similarity S act (x A , x B ) between the first production line x A and the second production line x B based on the comprehensive similarity between each pair of devices and a number of devices; wherein, calculation equations are as follows: s act ( v A , i , v B , j ) = { ss type + ss fea , ss mat = 1 & ⁢ ss qua = 1 0 , otherwise ( 1 - 2 ) s act ( x A , x B ) = ∑ v A , i ⁢ ∑ v B , j ⁢ max ⁢ ( s act ⁢ ( v A , i , v B , j ) ) + ∑ v B , j ⁢ ∑ v A , i ⁢ max ⁢ ( s act ⁢ ( v A , i , v B , j ) ) V A + V B ( 1 - 3 ) wherein, SS fea represents an ontology similarity of product category between the first device V A,j and the second device V B,j ; SS type represents an ontology similarity of product quality between the first device V A,i and the second device V B,j ; SS mat represents a material matching degree between the first device V A,i and the second device V B,j , and the material matching degree is 1 in a compatible case and 0 in a non-compatible case; SS qua represents a production process matching degree between the first device V A,i and the second device V B,j , and the production process matching degree is 1 in a compatible case and 0 in a non-compatible case; V A represents a number of devices in the first production line x A ; and V B represents a number of devices in the second production line x B .
2 . The method according to claim 1 , wherein the calculating the similarity S A,B between the first production line x A and the second production line x B specifically comprises: calculating a similarity S seq (x A , x B ) between a topology of the first production line x A and a topology of the second production line x B : s seq ( x A , x B ) = 2 * M A , B E A + E B ( 1 - 4 ) calculating a similarity S A,B between the first production line x A and the second production line x B : s A , B = w act * s act ( x A , x B ) + w seq * s seq ( x A , x B ) ( 1 - 5 ) wherein, M A,B represents a number of matched relationship edges between the first production line x A and the second production line x B ; E A represents a total number of relationship edges in the first production line x A ; E B represents a total number of relational edges in the second production line x B ; W act represents a weight of the similarity between devices in the first and second production lines in determining the similarity between the first production line x A and the second production line x B ; and w seq represents a weight of the similarity between topologies of the first and second production lines in determining the similarity between the first production line x A and the second production line x B .
3 . The method according to claim 1 , wherein the fuzzy compatibility matrix S is expressed as follows: S = [ s i , j ] n × n = [ 1 ⋮ ⋱ s i , 1 … 1 ⋮ ⋮ ⋮ ⋱ s j , 1 … s j , i … 1 ⋮ ⋮ ⋮ ⋮ ⋮ ⋱ s n , 1 … s n , i … s n , j … 1 ] ( 1 - 6 ) wherein, the similarity between identical production lines is 1 and S i,j =S j,i .
4 . The method according to claim 1 , wherein step S4 comprises: inputting the fuzzy compatibility matrix S; and outputting a series of granular layers {x(λ)10≤λ≤1} with different granularities and mutual transformability according to a granular computing algorithm of a fuzzy compatibility quotient space, wherein X represents the granular layers, and λ represents the granularities; the series of granular layers form the multi-granularity quotient space; and a specific calculation process is as follows: step 1: performing 1 st to m th loops to calculate m granularities λ to acquire m granular layers, wherein the following steps are executed in each loop: step 1.1: initializing historical production line sets A={x 1 ,x 2 . . . , x n }, B=Ø, and C=Ø; step 1.2: traversing the production lines in the set A, wherein the following steps are executed in each traversal: step 1.2.1 transferring initial production lines x j from the set A to the set B during a first loop: step 1.2.2: traversing current production lines x k in the set A during a second loop; step 1.2.2.1: determining if a similarity S(x k , x s ) between the initial production lines x j and the current production lines x k is greater than or equal to the granularity; if not, return to step 1.2.2; and if yes, executing the following steps: step 1): transferring the current production lines x k from the set A to the set B; step 2): traversing production lines x, in the set A during a third loop; step 3): determining if a similarity S(x k , x s ) between the production lines x k and the production lines x, is greater than or equal to the granularity; if yes, transferring the production lines x, from the set A to the set B; and if not, moving on to step 2); step 1.2.3: incorporating the set B into the set C to serve as a subset of the set C; and step 1.2.4: determining if the set A is an empty set; if yes, returning to the granular layer X(λ)=C corresponding to the granularity λ, and ending the first loop; and if not, letting the set B=└, skipping a current loop, and moving on to a next loop; and step 1.3: determining if i is equal to m; if yes, ending the loop, and exiting; and if not, continuing the loop; and outputting m granular layers X(λ)=C corresponding to the granularities λ i .
5 . The method according to claim 1 , wherein the selecting an optimal granular layer from the multi-granularity quotient space comprises: evaluating a granularity of a quotient space X(λ k ) based on a Shannon information entropy concept, wherein the granularity of the quotient space is defined as an average amount of information required to completely distinguish all production lines in the granular layer; E [ X ⁡ ( λ k ) ] = ∑ i = 1 g ❘ "\[LeftBracketingBar]" G i ❘ "\[RightBracketingBar]" n * log 2 ( ❘ "\[LeftBracketingBar]" G i ❘ "\[RightBracketingBar]" ) ( 1 - 7 ) wherein, g represents a number of production line granules in the quotient space X(λ k ); G i presents an i th production line granule in the quotient space X(λ k ): [G i ] represents a number of production lines in the i th production line granule in the quotient space X(λ k ); and log 2 (|G i |) represents an amount of information required to completely distinguish all the production lines in the production line granule G i ; calculating an information gain generated during a refinement process from a coarse-grained quotient space X(λ k-1 ) with a large information entropy to a fine-grained quotient space X(λ k ) as follows: IG [ X ⁡ ( λ k ) ] = E [ X ⁡ ( λ k - 1 ) ] - E [ X ⁡ ( λ k ) ] ( 1 - 8 ) finding the quotient space with the optimal granularity based on the information gain and the comprehensive similarity.
6 . The method according to claim 1 , wherein the constructing the topology reference structure for a production line based on the optimal granular layer comprises: extracting a typical production line topology sequence of production lines in each production line granule in the quotient space with the optimal granularity; calculating, by a dynamic programming method improved based on a longest common subsequence (LCS) algorithm, an LCS of all the production lines in each production line granule in the quotient space with the optimal granularity; performing, by an ontology-based computing method, property abstraction on each LCS to acquire a lowest superclass of all device properties in a domain ontology, so as to improve versatility and representativeness: vtr j , j ∈ LCS = C super ( vt 1 , j , vt 2 , j , … , vt i , j ) ( 1 - 10 ) wherein, C super (vt i,j , vt 2,j , . . . vt i,j ) represents abstract properties of a j th matched production line device node of all i production lines; further assembling an abstract set of all production line device nodes and production line topology relationship edges into a new topology reference structure for a production line, wherein the topology reference structure for a production line corresponding to each production line granule is expressed as: PRM i = ( Vr i , Er i , vtr i ) ( 1 - 11 ) wherein, Vr i ={vr i,2 vr i,2 . . . vr i,n } represents a set of matched production line device nodes; vtr i represents a lowest superclass of abstract device properties; and Er i ={er i,j,A 1er i,j,k =vr i,j *vr i,k 1≤j,k≤n,} represents an abstract set of production line topology relationship edges; and extracting a topology reference structure for a production line from each production line granule in the quotient space with the optimal granularity, wherein each topology reference structure for a production line is manifested by the set of matched production line device nodes and the abstract set of production line topology relationship edges.
7 . The method according to claim 6 , wherein the calculating, by a dynamic programming method improved based on the LCS algorithm, an LCS of all the production lines in each production line granule in the quotient space with the optimal granularity comprises: 1) starting matching from a first production line device node in a production line topology; 2) matching, based on a recursive equation, production line device nodes backwards one by one, and stacking a successfully matched production line device node into LCS (i,j) : LCS ( i , j ) = { max ⁢ { LCS ( i - 1 , j ) , LCS ( i , j - 1 ) } s act ( v A , i , v B , j ) < s t LCS ( i - 1 , j - 1 ) ⋃ { C super ( v A , i , v B , j ) } s act ( v A , i , v B , j ) ≥ s t 0 i = 0  ⁢ j = 0 . ( 1 - 9 ) wherein, s t represents a user preset similarity threshold for distinguishing a similar device from a non-similar device; C super (V A,i , V B,j ) represents abstract properties between two matched production line device nodes, that is, a superclass; and S act (V Aj , V B,j ) represents the comprehensive similarity between each pair of devices from the production line x A and the production line x B ; 3) repeating steps 1) and 2) to acquire a final LCS(s); 4) matching, if a production line granule comprises more than two production lines, the LCS (i,j) acquired by steps 1) to 3) and remaining production lines one by one, that is, repeating steps 1) to 3) until all production lines are matched, thus acquiring a final LCS corresponding to the production line granule.

Description

CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority to Chinese Patent Application No. 202310198392.6 with a filing date of Mar. 3, 2023. The content of the aforementioned application, including any intervening amendments thereto, is incorporated herein by reference. TECHNICAL FIELD The present disclosure relates to the technical field of production line construction for enterprises, and in particular to a method for constructing a topology reference structure for a production line. BACKGROUND At present, production lines constructed by traditional methods are constructed from beginning to end based on specific requirements on production functions, or constructed by combining multiple existing small work stations that perform specific production tasks into a production line according to their functions. These methods basically adopt serial design, without considering the overall situation of the production line, and require a long construction cycle. In addition, they are faced with problems such as insufficient connection and integration, and heavy reliance on designers' experience, which inevitably leads to many unreasonable aspects in the design. Currently, there are many complex production lines at home and abroad that fail to meet the pre-designed targets due to unreasonable or incorrect initial planning. Traditional methods for constructing production lines may not pose major problems for small enterprises with a small number of simply structured production lines. However, they will bring huge workload and production risks to large enterprises with a large number of complexly structured production lines. Furthermore, as traditional methods for constructing production lines cannot achieve reuse of the production line structure, they have significantly low construction efficiency and increase the enterprise's research and development costs. SUMMARY In order to overcome the shortcomings in the prior art, an objective of the present disclosure is to provide a method for constructing a topology reference structure for a production line. The present disclosure is based on the historical production line topology data of an enterprise to extract a commonly used topology reference structure for the production lines of the enterprise by a computer through a machine learning (ML) algorithm, so as to form a typical production line topology group of the enterprise. The present disclosure can record typical production line characteristics and production habits of the enterprise, realize reuse of a production line structure and production line construction knowledge, reduce the workload of production line designers, and improve the production line construction efficiency of the enterprise. In addition, the present disclosure avoids the interference of designers' subjective decisions to a certain extent, and the reference structure extracted by the computer has high reference value, and is objective, mature, and stable. To achieve the above objective, the present disclosure provides the following technical solution. The method for constructing a topology reference structure for a production line includes the following steps: S1: calculating a comprehensive similarity between a first production line xA and a second production line xB;S2: calculating a similarity parameter SA,B between the first production line xA and the second production line xB based on the comprehensive similarity;S3: calculating a similarity parameter si,j between each two production lines xi and xj in n historical production lines according to steps S1 and S2, and forming a fuzzy compatibility matrix S of the n historical production lines;S4: constructing a multi-granularity quotient space based on the fuzzy compatibility matrix S;S5: selecting an optimal granular layer from the multi-granularity quotient space; andS6: constructing the topology reference structure for the production line based on the optimal granular layer. In one embodiment, the calculating the comprehensive similarity between the first production line xA and the second production line xB specifically includes: S11: calculating a matching degree and similarity between a first device vA,i in the first production line xA and a second device VB,j in the second production line xB in terms of four properties, to obtain a comprehensive similarity between the first device VA,i and the second device VB,j; wherein, the four properties include material, production process, product category, and production quality;S12: calculating the comprehensive similarity sact(VA,j, VB,j)) between each pair of devices from the first production line xA and the second production line xB according to S11; and calculating the comprehensive similarity sact(xA, xB) between the first production line xA and the second production line xB based on the comprehensive similarity between each pair of devices and a number of devices; wherein, calculation equations are as follows: sact(vA,i,vB,j)={