CN109977572B

CN109977572B - Product part knowledge unit relation network construction method

Info

Publication number: CN109977572B
Application number: CN201910260138.8A
Authority: CN
Inventors: 张雷; 金志峰; 阚欢迎
Original assignee: Hefei University of Technology
Current assignee: Hefei University of Technology
Priority date: 2019-04-02
Filing date: 2019-04-02
Publication date: 2022-11-08
Anticipated expiration: 2039-04-02
Also published as: CN109977572A

Abstract

The invention discloses a method for constructing a product part knowledge unit relation network, which takes different knowledge units in a product as nodes of the network, takes the subordination relation of the knowledge units as edges of the network, takes all the edges as undirected edges, and takes the weight of the edges as an association coefficient between the knowledge units, thereby forming a weighted undirected network. The invention constructs a green design knowledge network model on the basis of the weighted complex network and provides a preprocessing method for subsequent knowledge processing work.

Description

Product part knowledge unit relation network construction method

Technical Field

The invention relates to the field of knowledge network model construction, in particular to a method for constructing a product part knowledge unit relation network.

Background

Product green design uses a great deal of design knowledge at each stage of the design activity, however, engineers often spend more than 50% of the time searching for the correct information in highly diverse and unstructured knowledge resources during the design process. Therefore, the effective storage, management and retrieval of design knowledge are the main measures for enterprises and industries to reduce the life cycle time and cost of product development and improve the quality of product development. Aiming at the problems that fragmented knowledge points are difficult to understand and lack of connection and the like, the problems of how to mine the relation among knowledge units and how to construct a product design knowledge resource network and the like also become hot spots for research of a plurality of scholars.

The invention constructs the incidence matrix between knowledge units based on two dimensions of a space domain and a time domain on the basis of analyzing the relevance of green design knowledge units of an electromechanical product, researches the topological characteristic structure of the incidence matrix, and provides a method for establishing a knowledge network model.

Disclosure of Invention

The invention aims to solve the defects in the prior art and provides a method for processing a multi-functional chip.

In order to achieve the purpose, the invention adopts the following technical scheme:

a method for building a knowledge unit relation network of product parts is characterized in that different knowledge units in a product are used as nodes of a network, the membership of the knowledge units is used as edges of the network, all the edges are taken as undirected edges, and the weight of the edges is an association coefficient between the knowledge units, so that a weighted undirected network is formed.

The model of the knowledge unit relationship network can be represented by equation (12):

G＝(K,E,W)(12)

wherein K represents the set of knowledge units in the network, i.e., K = (K1, K2, \8230;, kn); e represents the set of edges between knowledge units in the network, i.e., E = (E1, E2, \8230;, en); w represents the set of weights between knowledge units, i.e., W = (W1, W2, \8230;, wn), the magnitude of wn reflects the strength of association between two knowledge units, and the side weights also represent the probability of one knowledge unit causing a change in an adjacent knowledge unit after the change.

The invention has the advantages that:

the green design knowledge unit is expressed by adopting the matter element theory, the incidence relation between the knowledge units is quantitatively analyzed based on two dimensions of a time domain and a space domain, the incidence matrix between the knowledge units is formed, the green design knowledge network model is constructed on the basis of a weighted complex network, and a preprocessing method is provided for subsequent knowledge processing work.

Drawings

FIG. 1 is a network model based on improved complex network knowledge;

FIG. 2 is a correlation matrix;

fig. 3 is an engine model.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments.

Automobile engine design is a typical complex product development process in the electromechanical field, and although the engine design technology is mature, intensive design knowledge is still required in the design process as support due to the numerous engine parts and the relatively complex design process. With the increasing demand of the market for environment-friendly products, enterprises need to develop lower-carbon and higher-efficiency products to meet the intense market competition. The method takes the construction of a complex network of green design knowledge of the engine and the analysis of change influence as an example, and verifies the effectiveness of the method.

The specific structure of the engine is shown in FIG. 3; the method is mainly divided into 24 structural units, knowledge units corresponding to each structural unit are represented by k1 to k24, and the product green design knowledge units are described. For example, a link knowledge unit may be represented as

Taking the calculation of the correlation coefficient of the cylinder block knowledge unit and the connecting rod knowledge unit in the spatial domain as an example, the correlation coefficient of the knowledge unit in the knowledge network is calculated. Selecting 6 characteristic attributes of geometric attribute constraint, assembly attribute constraint, disassembly attribute constraint, material attribute constraint, precision attribute constraint and environmental performance constraint, wherein the corresponding importance degrees are as follows in sequence: 0.7,0.6,0.4,0.2,0,0.4. The calculation can obtain:

namely, the correlation coefficient of the cylinder block knowledge unit and the connecting rod knowledge unit in the spatial domain is 0.38, and by analogy, the correlation strength of the knowledge units in the spatial domain and the correlation strength in the time domain are respectively calculated, the correlation matrix among the knowledge units is established, and the correlation matrix among partial knowledge units is shown as follows.

And (4) association matrix of knowledge units in a spatial domain.

The correlation matrix of the knowledge unit in the time domain.

Establishing a weight set a, namely:

a matrix representation of the corresponding system of linear equations is obtained:

the solution to the system of linear equations is thus: β = (0.04, 0.007, 0.963), the coefficient is the optimal combination coefficient of the weight set, and the game theory combination weight vector for obtaining the correlation coefficient is:

the strength of association between the knowledge units Ci and Cj thus available can be expressed as:

and obtaining an incidence matrix among the engine knowledge network models.

The knowledge network model was designed by creating an engine green with 24 knowledge nodes and 177 edges in the Pajek software through the correlation matrix.

Claims

1. A product part knowledge unit relation network construction method is characterized in that different knowledge units in a product are used as nodes of a network, membership of the knowledge units are used as edges of the network, all the edges are taken as undirected edges, and the weight of the edges is an association coefficient between the knowledge units, so that a weighted undirected network is formed;

the specific process of the construction method is as follows: the product is an engine and is divided into 24 structural units, knowledge units corresponding to each structural unit are represented by k1 to k24, green design knowledge units of the product are described, and connecting rod knowledge units can be represented as

Calculating the correlation coefficient of knowledge units in a knowledge network, selecting 6 characteristic attributes of geometric attribute constraint, assembly attribute constraint, disassembly attribute constraint, material attribute constraint, precision attribute constraint and environmental performance constraint, calculating the correlation coefficient of a cylinder body knowledge unit and a connecting rod knowledge unit in a spatial domain, and so on, calculating the correlation strength of the knowledge units in the spatial domain and the correlation strength in a time domain respectively, and establishing a correlation matrix among the knowledge units;

the specific steps of establishing the incidence matrix among the knowledge units are as follows: establishing an incidence matrix of a knowledge unit in a space domain, establishing an incidence matrix of the knowledge unit in a time domain, establishing a weight set a, obtaining a matrix representation of a corresponding linear equation set so as to obtain a solution of the linear equation set, then obtaining a game theory combined weight vector of an incidence coefficient, calculating the incidence strength between the knowledge unit and the knowledge unit, and finally obtaining an incidence matrix between knowledge network models of the engine;

and designing a knowledge network model in Pajek software through the incidence matrix.

2. The method for building the knowledge unit relation network of the product part according to claim 1, wherein the model of the knowledge unit relation network can be represented by equation (12):

G＝(K,E,W) (12)

wherein K represents a set of knowledge units in the network, i.e., K = (K1, K2, \8230;, kn); e represents the set of edges between knowledge units in the network, i.e., E = (E1, E2, \8230;, en); w represents the set of weights between knowledge units, i.e., W = (W1, W2, \8230;, wn), the magnitude of wn reflects the strength of association between two knowledge units, and the side weights also represent the probability of one knowledge unit causing a change in an adjacent knowledge unit after the change.