CN-116467491-B - Identification matching method for production line configuration structure diagram

CN116467491BCN 116467491 BCN116467491 BCN 116467491BCN-116467491-B

Abstract

The invention discloses a method for identifying and matching a production line configuration structure diagram, which comprises the step that an enterprise presents a personalized production line to be built by the enterprise and corresponding functions in a mode of the production line configuration structure diagram (namely, inquiring sub-diagrams) when making a selection decision of a production line equipment provider. And for the convenience of production and transportation, the supplier nodes of the production line equipment connected in the figure need to have a cooperative relationship, so that a production line configuration structure diagram with clear functions and clear supply and demand relationship can be constructed. And then, the production line configuration structure diagram is identified and matched with the existing production line equipment provider database of the manufacturing enterprise community, so that a plurality of possible result matching subgraphs are obtained for the enterprise to make the next accurate decision, most decision interference factors can be eliminated, and the decision efficiency is improved.

Inventors

GUO JIWEI
LENG JIEWU
ZHU XIAOFENG
XU CAIYU
SU HONGYE
LIU QIANG

Assignees

广东工业大学

Dates

Publication Date: 20260512
Application Date: 20230418

Claims (6)

1. The identification matching method for the production line configuration structure diagram is characterized by comprising the following steps of: The query requirement proposed by a requirement party is converted into a query subgraph, and the query subgraph represents a production line to be built by an enterprise; representing the query subgraph as a factor graph; Constructing a matching model between an enterprise production line and a production line equipment provider; Solving a matching model by utilizing the factor graph and a belief propagation algorithm, so as to find a matching result, namely a matching sub graph, from a data network containing an existing production line equipment provider database of an enterprise; converting the matching subgraph into a query result described by words; Constructing a matching model between an enterprise production line and a production line equipment provider, comprising: by maximizing edge probability To find the mapping, the LBP algorithm may find the edge probabilities to achieve maximization of the joint probability distribution of the mapping: (1-1) wherein: Mapping vector representing when joint probability distribution is maximized ; Representing mapping vectors, i.e. data networks Mid-and query subgraph Middle node A corresponding subset of nodes; Expressed in terms of a data network A joint probability distribution that is a mapping of conditions; representing a data network; wherein the probability distribution is combined I.e. the matching objective function is expressed as: (1-2) wherein: representing the normalization factor; representing mapping vectors, i.e. data networks Mid-and query subgraph Middle node A corresponding subset of nodes; representing a mapping vector One element of (a) is provided; Representing nodes Is the variable value of (a); Representing a data network Middle AND Adjacent nodes Is the variable value of (a); Representing individual matching similarity relationship functions; representing a bi-directional dependency function between nodes; Due to the structural differences of the mapping variables, the matching objective function (1-2) cannot be directly applied to the message passing of the LBP algorithm, for which purpose discrete mapping vectors are applied Substitution with 0-1 mapping matrix If (3) Then If not, the first part of the first part is connected with the second part, ; The matching objective function is converted into: , wherein: representing the normalization factor; Representing query subgraphs Middle node Is a vector of attributes of (a); Representing a data network Node in (a) Is a vector of attributes of (a); Representing query subgraphs Middle edge Is a vector of attributes of (a); Representing a data network Middle edge Is a vector of attributes of (a); representing data nodes under query node conditions A conditional probability distribution of occurrence; expressed on the query edge Mapping edges under the condition of (2) A conditional probability distribution of occurrence; representing a mapping matrix; representing a mapping matrix One element of (a) is provided; Representing query subgraphs Any one of the nodes; Representing a data network Any one of the nodes; Data network The attributes of the middle edge represent semantic relationships, time and space dependencies, interactions and influence relationships, while the attributes of the nodes represent metadata, state and semantic information; To convert the product operation of the logarithmic domain in equation (1-3) into sum operation to simplify the calculation, the matching objective function is applied Taking the logarithm, converting the logarithm into : (1-4) (1-5) (1-6) Wherein: Representing a matching degree function; Representing query nodes And data node Semantic similarity functions between; Representing query edges And data edge Semantic similarity functions between; Representing query subgraphs Any one of the edges; Representing a data network Any one of the edges of the strip.
2. The method for identifying and matching a structure diagram of a production line configuration according to claim 1, wherein the step of representing the query subgraph as a factor graph comprises: representing a query sub-graph as Representing a data network as = A matching representation of a set of mappings from query subgraphs to nodes and edges of the data network, the mappings being defined as a vector Wherein each query node Are all in the data network Has a mapping node Each query edge Are mapped to edges in the data network ; Wherein, the Representing query subgraphs Is provided with a set of nodes in the network, Representing a data network Node sets in (a); Representing query subgraphs Is provided with a set of edges in the middle, Representing a data network A set of edges in (a); Representing query subgraphs A set of attribute vectors for the intermediate node, Representing a data network Attribute vector sets of the intermediate nodes; Representing query subgraphs A set of attribute vectors for the mid-edge, Representing a data network Attribute vector set of middle edge; Representing a data network A set of all mapping nodes in the hierarchy; Representing a data network A set of all mapping edges in (a); Representing the query sub-graph as a factor graph, a factor graph comprising two types of nodes, one representing edges in the query sub-graph Factor node representing a node in a query sub-graph Is a variable node of (a).
3. The method of claim 1, wherein solving the matching model using the factor graph and the belief propagation algorithm to find the matching result from the data network containing the enterprise's existing line equipment provider database comprises finding the best match by passing belief messages of the mapping probabilities of the query node/edge and the data node/edge between the factor node and the variable node, resulting in an edge probability distribution of the query node For generating mapping vectors 。
4. A method for identifying and matching a production line configuration structure according to claim 3, wherein the best match is found by passing belief messages of mapping probabilities of query nodes/edges and data nodes/edges between factor nodes and variable nodes, resulting in an edge probability distribution of the query nodes For generating mapping vectors Comprising: A1, initializing a factor graph, and constructing an initial potential function, namely an individual matching similarity relation function, according to attribute features in the factor graph And a bi-directional dependency function between nodes Initializing belief values of all nodes; a2, carrying out 1 st to nth iterations, namely message transfer loop, traversing all factor nodes and variable nodes, and slave factor nodes Directional variable node Delivering messages from variable nodes Delivering messages to all adjacent factor nodes; Wherein the belief value of each node is updated by a maximum sum operation, the passing and updating of which includes two directions: (1) By variable nodes Directional factor node Delivering the message; (2) From factor nodes Transmitting a message to a variable node connected with the variable node; Message passing that simplifies the belief propagation algorithm is in two steps: In a first step, two messages passed from each factor node to the connected variable nodes are updated: (1-7) (1-8) wherein: Representing query edges Mapping to data edges Edge log probability of (2), data edges Ending at node ; Representing query edges Mapping to data edges Edge log probability of (2), data edges Starting at a node ; Representing query edges Mapping to data edges Edge log probability of (2), data edges Starting at a node ; Representing query edges Mapping to data edges Edge log probability of (2), data edges Ending at node ; Representing query nodes Mapping to data nodes Is a logarithmic edge probability of (1); Representing query nodes Mapping to data nodes Is a logarithmic edge probability of (1); Representing query edges And data edge Semantic similarity functions between; if a query edge corresponds to an empty data edge, i.e., no mapping, then Value substitution to predetermined A value; second, update the belief value of the factor node using the message received by the variable node: (1-9) wherein: Representing query nodes Mapping to data nodes Is a logarithmic edge probability of (1); Representing query edges Edge log probability mapped to data edge, data edge ending at node ; Representing query edges Edge log probability mapped to data edges, the data edges starting at nodes ; Representing a set of edges in a query sub-graph; Representing query subgraphs Intermediate and node An adjacent node; Representing query subgraphs Intermediate and node One edge connected; A3, judging the belief values of all the nodes in each iteration, wherein the judgment condition is that whether all the nodes exist or not I.e. if the difference between the belief values of the last two iterations is less than a set threshold, thus being approximately regarded as unchanged, if it is satisfied that the belief values of the nodes tend to be stable, then = Obtaining edge probability distribution, calculating maximum edge probability Can obtain and inquire nodes Mapping nodes in corresponding data networks If the judging condition is not met, continuing iteration until the judging condition is met; Wherein, the A belief value representing each iteration; Representing a convergence threshold; Representing nodes to be queried Belief value of (2) Query node for re-indexing Mapping to data nodes Is a boundary probability distribution of (1); and the mapping nodes of all the query nodes in the query subgraph can be obtained by the similar method, so that the matching subgraph is formed.
5. The method of identifying and matching a production line configuration structure according to claim 4, wherein when the belief propagation algorithm starts message delivery, there are: (1-10)。
6. The method for identifying and matching a production line configuration structure according to claim 4, wherein in order to further improve the performance of the algorithm and obtain a larger message change in each transmission iteration, the original message update mode is changed into an incremental mode: (1-11) wherein: Representing belief values, i.e. 、 And ; A convergence rate control coefficient indicating a probability of avoiding falling into a local optimum state; the belief value representing the last iteration.

Description

Identification matching method for production line configuration structure diagram Technical Field The invention relates to the technical field of supply and demand matching, in particular to an identification matching method of a production line configuration structure diagram. Background When a customer is looking for a product line that can achieve its own personalized needs, the description of the functional needs is often ambiguous or there are multiple possibilities at the beginning, which results in a very large number of product line design history case data that can be selected, and how to select a series of product line equipment suppliers (and require cooperation relationships between the suppliers) that meet the required functions from these massive data, and to screen out most of the non-compliant suppliers is certainly a problem. At present, most manufacturing enterprises also follow the traditional supply and demand matching thought when searching for the product line equipment suppliers, namely, discretely searching for single product line equipment suppliers according to the functional requirements of the product line, and then combining all suppliers together to build the personalized product line of the manufacturers. By adopting the traditional supply and demand matching scheme for a long time, the problem is solved for small enterprises with small production line requirements and simple production line function structures, but for large enterprises with huge production line requirements and complex production line function structures, decision-making workload is doubtless multiplied, the production line construction efficiency is obviously reduced, and the enterprise research and development cost is increased. Meanwhile, the discrete supply and demand matching also can cause problems of global consideration of the production line, strong experience dependence on designers and the like. Disclosure of Invention The invention aims to overcome the defects of the prior art, and provides an identification matching method of a production line configuration structure diagram, which is used for solving the problem of searching a production line equipment provider for building a personalized production line aiming at a manufacturing enterprise. In order to achieve the above purpose, the technical scheme provided by the invention is as follows: A method for identifying and matching a production line configuration structure diagram comprises the following steps: The query requirement proposed by a requirement party is converted into a query subgraph, and the query subgraph represents a production line to be built by an enterprise; representing the query subgraph as a factor graph; Constructing a matching model between an enterprise production line and a production line equipment provider; Solving a matching model by utilizing the factor graph and a belief propagation algorithm, so as to find a matching result, namely a matching sub graph, from a data network containing an existing production line equipment provider database of an enterprise; and converting the matching subgraph into a query result described in text. Further, representing the query subgraph as a factor graph includes: Representing the query sub-graph as Q= (V Q,EQ,vtQ,etQ), the data network as D= (V D,ED,vtD,etD), a match representing a set of mappings from the query sub-graph to nodes and edges of the data network, defining the mappings as a vector X= [ X m ], wherein each query node mεV Q has a mapping node X m∈XD in the data network D, and each query edge e= (m, n) εE Q is mapped to an edge X m、Xn)∈ED in the data network; Wherein V Q represents the node set in query sub-graph Q, V D represents the node set in data network D, E Q represents the edge set in query sub-graph Q, E D represents the edge set in data network D, vt Q represents the attribute vector set of nodes in query sub-graph Q, vt D represents the attribute vector set of nodes in data network D, et Q represents the attribute vector set of edges in query sub-graph Q, et D represents the attribute vector set of edges in data network D, X D represents the set of all mapping nodes in data network D, E D represents the set of all mapping edges in data network D; the query subgraph is represented as a factor graph, and one factor graph comprises two types of nodes, namely a factor node representing an edge e= (m, n) epsilon E Q in the query subgraph and a variable node representing a node m epsilon V Q in the query subgraph. Further, constructing a matching model between the enterprise production line and the production line equipment provider, including: the map is found by maximizing the edge probability b m(i)＝P(Xm = i|d), the LBP algorithm can find the edge probability to achieve the maximization of the joint probability distribution of the map: wherein: x represents a mapping vector, namely a node subset corresponding to a node m in a query sub-graph Q in a data netwo