CN106326610A - Design network based design alterationpropagation predicting method and system - Google Patents

Abstract

The invention discloses a design change propagation prediction method and system based on a design network, aiming at the negative impact of frequent design changes on cost and efficiency in the product design process, in order to predict the change propagation path in advance and suppress unnecessary Change, based on the established relationship network model among design attributes, uses complex network analysis techniques to analyze the optimal propagation path of design changes, so as to achieve the purpose of reducing the scope of influence of change propagation. This method can establish a fairly objective design change analysis network model to fundamentally ensure the accuracy of change propagation prediction; at the same time, a mathematical model of change propagation intensity is quantitatively established based on complex network analysis techniques.

Description

A Design Change Propagation Prediction Method and System Based on Design Network

technical field

The invention relates to a design change propagation prediction method and system based on a design network.

Background technique

In the process of product development, design requirements may change, the availability of resources and components may change, design failures and wrong decisions will often occur, and design information may be leaked during the exchange process, resulting in inevitable design changes. Product design change management is of great significance to manufacturing enterprises. Product design changes are changes to the structure (including fit, shape, size, surface, material, etc.), performance (including stability, strength, corrosion resistance, etc.), function (including speed, efficiency, etc.) or performance-function of technical products. (i.e., design principles) relationships, and performance-structure (i.e., physical laws) relationships. Design changes often occur in the continuous design improvement process of products, especially complex products, and determine 70-80% of the final cost of the product.

Design change management can help enterprises predict the possibility of change propagation, reduce unnecessary changes, choose effective change propagation paths, and implement changes as soon as possible. At the heart of design change management is the change propagation prediction and control methodology. Design change propagation is a process in which a change in a component or unit in the current system configuration or design results in further or more changes in the system. Prediction of change propagation can support effective design decisions and engineering planning, avoid subsystems with high change costs, and quantitatively estimate changes in workload and product costs in the early stages of design.

Existing studies involving design change propagation prediction still have the following two limitations and deficiencies:

(1) When modeling, it is necessary to manually specify the sensitivity and influence degree of change propagation among design attributes, which is highly subjective, thus reducing the prediction accuracy of change propagation. For example, the most commonly used change propagation method predicts only about 30% accuracy.

(2) There is a lack of research on design change propagation at the attribute granularity.

Contents of the invention

The purpose of the present invention is to solve the above problems, provide a design change propagation prediction method and system based on design network, establish a mathematical model of change propagation intensity through complex network analysis technology, improve the accuracy of design change propagation prediction, the method can objectively The probability of change propagation between design attributes is defined in a precise manner, and the margin in design parameters is considered at the same time, and the influence range of change propagation is narrowed while analyzing the optimal change propagation path.

In order to achieve the above object, the present invention adopts the following technical solutions:

A design change propagation prediction method based on a design network, comprising the following steps:

Step (1): Establish a design change analysis network model;

Step (2): Calculate the change propagation probability between the design attributes stored in the product design change database; use the change propagation probability between the design attributes as the weight of the connection in the design change analysis network model; thus establish a more objective design change analysis network model;

Step (3): Establish a design change propagation intensity model by using the propagation probability, degree of design attribute, design attribute change margin, weight, and penalty coefficient of long-chain connections, and calculate the design attribute change propagation intensity;

Step (4): Optimize the change propagation path: take the change propagation intensity as the objective function, and obtain the change propagation path corresponding to the minimum cumulative change propagation intensity on the design change analysis network through the ant colony optimization algorithm. The change propagation path is optimal propagation path.

The greater the change propagation intensity in step (3), the greater the amount of change required for subsequent attributes connected to it.

The step of described step (1) is:

Design attributes are used as network nodes, and the connection relationship between parent and child design attributes is expressed as the connection between nodes. According to the design attributes, it is gradually refined from top to bottom, and finally a design change analysis network model is formed; the relationship between design attributes belonging to the same part The connection relationship between parts is defined as a short chain, and the connection relationship between design attributes across parts is defined as a long chain.

Said step (1):

Use the set V={v ₁ ,v ₂ ,...,v _n } to correspond to the design attribute set, and define E as the connection set between nodes, and the design change analysis network model is recorded as G={V,E};

The design attribute is represented as a circle, and the connection relationship is represented as a line with an arrow. The design attribute is clustered according to the part to which it belongs, and each category is distinguished by a dotted line; the connection relationship includes two types of parameter connection relationship and constraint connection relationship.

The parameter connection relationship is expressed as a function form y=f(x ₁ , x ₂ , . . . , x _n ), where y is the parent design attribute, and x _i is the i-th child design attribute. Arrows point from child design attributes to parent design attributes in the design change analysis network model. A parameter connection relationship is composed of a parent design attribute and multiple sub-design attributes. The value of the parent design attribute is determined by the values of the independent sub-design attributes.

$\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span><span\; class="notranslate">\; \&part;</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; /</span><span\; class="notranslate">\; \&part;</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ;</span>$

In the design change analysis network model, because the sub-design attributes are independent of each other, the change of the sub-design attributes does not directly change the sibling attributes of the sub-design attributes, but indirectly changes the sibling attributes of the sub-design attributes by changing the parent design attributes; The change of the parent design attribute is realized by changing its child design attribute, and the size of the change is set artificially; therefore, in the parameter connection relationship, the parent design attribute changes with the change of any child design attribute; when the parent design attribute changes , if the parent design attribute has a child design attribute, the child design attribute is changed, if the parent design attribute has multiple child design attributes, each child design attribute is changed according to different setting change weights.

The constraint connection relationship is an artificially formulated design rule to meet the assembly constraints, functional coordination and performance guarantee between components, expressed by a function f(x ₁ ,x ₂ ,...,x _n )≥0;

Each design attribute in the constraint connection relationship has the same dimension, which is represented by a two-way arrow in the design change analysis network model;

When any constraint attribute is changed in the constraint connection relationship, other constraint attributes must also be changed to continue to ensure that the connection conforms to the constraint relationship;

In the constraint connection relationship, the change of any attribute will cause the change of other connection attributes, and the change weight is artificially set.

Said step (2):

In the design change analysis network model, each connection has a weight, that is, the propagation probability P _ij . P _ij is the possibility that the successor attribute changes when the predecessor attribute i changes. P _ij is equal to the conditional probability of occurrence of attribute j under the premise that design attribute i appears in a record in the design change database, namely

${\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; \&cap;</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>\frac{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; i</span>}}\frac{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

In formula (1), since P(v _i ) is not equal to P(v _j ), P _ij and P _ji are not equal. If there is no connection between two design attributes, P _ij is equal to 0. And the sum of the propagation probabilities of design attribute i and all adjacent attributes of design attribute i is 1.

$\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; f</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

The parameter meaning of P(v _j |v _i ) is the conditional probability of attribute j appearing under the premise that design attribute i appears in a record in the design change database; the parameter meaning of P(v _i ∩v _j ) is the design attribute i and The probability that j appears in a record in the design change database at the same time; the parameter meaning of P(v _i ) is the probability of design attribute i appearing in the design change database; the parameter meaning of P(v _j ) is the design attribute j appearing in the design change database The parameter meaning of P(v _i |v _j ) is the conditional probability of attribute i appearing under the premise that design attribute j appears in a record in the design change database; the parameter meaning of P _ji is the change of current driving attribute j The possibility that the successor attribute i will change.

Said step (3):

At the k-th propagation, the design attribute changes the propagation intensity defined as:

${\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}\mathrm{<span\; class="notranslate">\; 0</span>}& {\mathrm{<span\; class="notranslate">\; \&Delta;\&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}\\ {\mathrm{<span\; class="notranslate">\; \&omega;</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; \&lsqb;</span>{\mathrm{<span\; class="notranslate">\; \&omega;</span>}}_{\mathrm{<span\; class="notranslate">\; p</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&omega;</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}}\frac{{\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}}{\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; f</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}}<span\; class="notranslate">\; \&rsqb;</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>\frac{{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{{\mathrm{<span\; class="notranslate">\; \&Delta;\&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}}<span\; class="notranslate">\; )</span>& {\mathrm{<span\; class="notranslate">\; \&Delta;\&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ></span>{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}\end{array}\right.<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; f</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

Among them, ρi is the change margin of the design attribute _i in the k-th propagation, is the required change value of the design attribute i in the k-th propagation; F ^k represents the set of design attributes affected by the k-th change propagation; ω _p is the weight value of the propagation probability, and ω _d is the weight value of the degree of the design attribute , and ω _p +ω _d ＝1; d _j is the degree of the design attribute, which is calculated by the adjacency matrix of the design attribute; ω _s is the penalty coefficient for long-chain connections, where ω _s ≥ ₁ , Increased change propagation strength when parts are propagated.

The step (4):

Based on the design change analysis network model and the change propagation intensity model, the optimization objective function of the change propagation path is established as follows:

$\left.\begin{array}{cc}\mathrm{<span\; class="notranslate">\; arg</span>}\mathrm{<span\; class="notranslate">\; min</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; max</span>}\underset{\mathrm{<span\; class="notranslate">\; k</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>& \mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; ...</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; N</span>}\\ \mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; .</span><span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; \&Delta;\&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; <</span>{\mathrm{<span\; class="notranslate">\; 10</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 5</span>}}& \mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; f</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}\end{array}\right\}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; )</span>$

Equation (4) defines the total propagation intensity when the initial design attribute is changed and propagates to other design attributes. When the design margin ρ _j and the design variable When the difference of is less than 10 ^-5 , it indicates that the change propagation convergence does not further affect the change of other design attributes. After each iteration, the pheromone left by the connected ant colony is updated according to the following rules:

${\mathrm{<span\; class="notranslate">\; \&tau;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&rho;</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; \&tau;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; +</span>\underset{\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; N</span>}}_{\mathrm{<span\; class="notranslate">\; a</span>}}}{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; \&Delta;\&tau;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; l</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; )</span>$

Among them, τ _ij is the pheromone value transmitted from design attribute i to design attribute j, ρ is the pheromone volatilization coefficient, where 0<ρ<1; N _a is the number of ant colonies, is the pheromone value released by the lth ant. τ _ij represents the degree of attraction to transition to other possible states;

Among them, Q is a constant, and D _l is the objective function value of the lth ant's walk;

change tends to select the one with the largest Adjacent design properties of the propagating, the state transition willingness η _ij is directly related to Therefore, the state transition willingness η _ij is defined as follows:

${\mathrm{<span\; class="notranslate">\; \&eta;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; )</span>$

After each iteration, the state transition willingness η _ij according to the current The value is updated. The probability that the lth ant climbs from design attribute i to design attribute j for

Among them, α is a parameter controlling the degree of influence of τ _ij , α≥0, and β is a parameter controlling the degree of influence of η _ij , β≥1; is the directly connected attribute set of the l-th ant in the design attribute i.

A design change propagation prediction method based on a design network further includes step (5): a propagation optimization result display module, which displays the design change propagation path optimization result in a user interface.

The displayed contents in step (5) include: design attributes, connections between design attributes, change propagation intensity values, and change propagation paths.

The product design change database: used to store design change records, the design change records include: design attribute serial number, forward design attribute serial number, attribute name, value before change, value after change and date of change.

The design change record is used for quantitative calculation of P _ij .

A design change propagation prediction system based on design network, including:

Design change analysis network model: design attributes are used as network nodes, and the connection relationship between parent and child design attributes is expressed as a connection between nodes; the connection relationship between design attributes belonging to the same part is defined as a short chain, and the connection relationship between parts Defined as a long chain; according to the attribute to distinguish the connection relationship, it can be divided into parameter connection relationship and constraint connection relationship;

Product design change database: responsible for storing each affected design attribute and the degree of influence after the product design change occurs, and is used to calculate the propagation impact probability between each design attribute;

Propagation probability calculation module: calculate the change propagation probability between design attributes as the weight of connections in the design change analysis network model; establish an objective design change analysis network model;

Change propagation intensity analysis module: establish a design change propagation intensity model by using the degree of each design attribute, design margin, connection weight, and penalty coefficient for long-chain connections;

Change propagation path optimization module: take the cumulative change propagation intensity as the objective function, and obtain the optimal change propagation path corresponding to the minimized maximum cumulative change propagation intensity through the ant colony optimization algorithm;

Propagation optimization result display module: display the design change propagation path optimization results in the user interface, and the display contents include: design attributes, connections between design attributes, change propagation intensity values, and change propagation paths.

Beneficial effects of the present invention:

Establish a fine-grained design change analysis network model based on design attributes and their relationships, and mine the change propagation probability from the product design change database as the weight value of the model connection. Based on this relatively objective design change analysis network model, the accuracy of propagation prediction is guaranteed. Accuracy; at the same time, with the help of complex network analysis technology to establish a mathematical model of change propagation intensity, and considering the margin of design parameters, optimal change propagation can be obtained.

Some conflicts after design changes are often not observed at large-scale levels such as systems and components, but become extremely obvious at small-scale levels such as attributes and parameters. The analysis of change propagation from the attribute granularity can objectively and quantitatively calculate the tolerance capability of components and the influence degree of change propagation, so that the change can be purposefully propagated along the path with strong tolerance capability, so as to make the design change propagation process converge as soon as possible.

Description of drawings

Figure 1 is a frame structure diagram of the design change propagation prediction method based on the design attribute network;

Figure 2 is an example diagram of the design change analysis network model;

Figure 3 An example diagram of the change propagation optimization path.

detailed description

The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

A design change propagation prediction system based on design network, including:

Design change analysis network model: This model takes design attributes as network nodes, and expresses the connection relationship between parent and child design attributes as connections between nodes; the connection relationship between design attributes belonging to the same part is defined as a short chain, and the connection between parts The connection relationship is defined as a long chain; according to the attribute, the connection relationship can be divided into parameter connection relationship and constraint connection relationship; this network model is convenient for rapid quantitative analysis of the influence range of design change propagation with the help of complex network analysis technology;

Product design change database: responsible for storing each affected design attribute and the degree of influence after the product design change occurs, and is used to evaluate the propagation impact probability between each design attribute;

Propagation probability evaluation module: This module analyzes the records in the design change database through data mining technology, and evaluates the change propagation probability between design attributes, which is used as the weight of the connection in the design change analysis network model; based on this, a relatively objective Design change analysis network model;

Change propagation intensity analysis module: Consider the degree of each design attribute, design margin, connection weight and penalty coefficient of long-chain connections to establish a design change propagation intensity model to provide a quantitative evaluation index for design change propagation optimization;

Change propagation path optimization module: take the cumulative change propagation intensity as the objective function, and obtain the optimal change propagation path corresponding to the minimized maximum cumulative change propagation intensity through a heuristic optimization algorithm;

Propagation optimization result display module: display the design change propagation path optimization results in the user interface, the displayed content includes: design attributes, connections between design attributes, change propagation intensity values, and change propagation paths;

The main claim modules in the above description are: design change analysis network model, propagation probability evaluation module, change propagation intensity analysis module and change propagation path optimization module.

The present invention mainly utilizes Java language, MySQL database, Ontology editor implementation. As shown in Figure 1, the present invention mainly includes: a design change analysis network model, a product design change database, a propagation probability evaluation module, a change propagation intensity analysis module, and a change propagation path optimization module.

1. Design change analysis The network model is based on the design theory, gradually refines the design attributes from top to bottom, and finally forms a large attribute association network. The model takes design attributes as network nodes, and expresses the connection relationship between parent and child design attributes as connections between nodes, as shown in Figure 2. Use the set V={v ₁ ,v ₂ ,...,v _n } to correspond to the design attribute set, and define E as the connection set between nodes. The design change analysis network model can be written as G={V,E}. The connection relationship between design attributes belonging to the same part is defined as a short chain, and the connection relationship between parts is defined as a long chain. The network model can be divided into the network model according to the attribute to distinguish the connection relationship, which is convenient for the rapid quantitative analysis of the influence range of design change propagation with the help of complex network analysis technology.

In Figure 2, design attributes are represented as circles, and connection relationships are represented as lines with arrows. Design attributes can be clustered according to the part belonging, and each category is distinguished by a dotted line. The connection relationship includes two types of parameter connection relationship and constraint connection relationship. The parameter connection relationship is established according to the physical laws in the design theory, and this type of connection can be uniformly expressed as a function form y=f(x ₁ ,x ₂ ,...,x _n ), where y is the parent design attribute, and _xi represents the first i subdesign attributes. In the design change analysis network model diagram, the arrow points from the child design attribute to the parent design attribute. A parameter connection relationship is composed of a parent design attribute and multiple sub-design attributes. The value of the parent design attribute is determined by the values of the independent sub-design attributes, that is, the value of the parent design attribute changes with the value of the sub-design attribute according to physical laws. and change, In the design change analysis network model, due to the mutual independence among sub-design attributes, the change of sub-design attributes does not directly affect its sibling attributes, but indirectly affects other sibling attributes by affecting its parent design attributes. Changing the parent design property cannot happen directly, the change needs to be passed to the child design property, and the size of the change needs to be weighed by the designer. Therefore, in the parametric connection relationship, the parent design attribute changes with the change of any child design attribute; when the parent design attribute changes, its only child design attribute must change, and when it has multiple child design attributes, each child design attribute Attributes are changed to varying degrees.

Constrained connections are design rules made artificially due to assembly, function fitting, and performance assurance, and are expressed by the function f(x ₁ ,x ₂ ,...,x _n )≥0. Each design attribute in the constraint connection relationship has the same dimension, which is represented by a double-headed arrow in the design change analysis network model diagram. The constraint connection relationship may be difficult to satisfy when a certain constraint attribute is changed, so other constraint attributes need to be changed to continue to ensure that the connection conforms to the constraint relationship. In the constraint connection relationship, the change of any attribute will cause the change of other connection attributes, and the degree of change needs to be specified manually.

2. The propagation probability evaluation module analyzes the records in the design change database through data mining technology, and evaluates the change propagation probability between design attributes, which is used as the weight of the connection in the design change analysis network model; based on this, a relatively objective design is established Change analysis network model. In the design change analysis network model, each connection has a weight, that is, the propagation probability P _ij . P _ij is the possibility that the successor attribute changes when the predecessor attribute i changes. The probability can be obtained by data mining the change records in the design change database. P _ij is the conditional probability of occurrence of attribute j under the premise that design attribute i appears in a record in the design change database, namely

${\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; \&cap;</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>\frac{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; i</span>}}\frac{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

In formula (4), since P(v _i ) and P(v _j ) are generally not equal, P _ij and P _ji are not equal. If there is no connection between the two design attributes, P _ij is equal to 0. And the sum of the propagation probabilities of design attribute i and all its adjacent attributes is 1.

$\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; f</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$

3. The change propagation intensity analysis module considers the degree of each design attribute, the design margin, the weight of the connection, and the penalty coefficient of the long-chain connection to establish a design change propagation intensity model to provide a quantitative evaluation index for the design change propagation optimization algorithm.

Due to the coupling relationship between design attributes, once the value of a design attribute changes, the change will be gradually propagated to other design attributes. In the design change analysis network model, the propagation strength among design attributes is directly related to the propagation probability. During change propagation, changes are propagated preferentially to connections with a greater probability of propagation. The cumulative transmission probability decreases exponentially as transmission continues. Since the design change analysis network model often has small-world network properties, in addition to the propagation probability, the degree of nodes and long-chain connections are important influencing factors for change propagation. The larger the degree of a node, the more adjacent attributes are affected by the change propagation. In order to avoid continuous propagation of changes, design attributes with large degrees should be avoided as much as possible during the selection of change propagation paths. At the same time, considering the topological properties of the small-world network, there are a small number of long-chain connections between different design parts. Different from the short-chain connections that account for the vast majority in the design change analysis network model, the long-chain connections greatly affect the change propagation between design components, and a large propagation penalty coefficient needs to be assigned. In addition, the margin existing in the existing design parameters can absorb all or part of the amount of changes, and play a role in absorbing or buffering the propagation of changes. The design margin is the difference between the limit allowable value and the current value. From the perspective of design margin, design attributes can be divided into absorption nodes, buffer nodes and transfer nodes. Therefore, change propagation intensity is introduced to quantify the impact of change propagation, including propagation probability, node degree, long chain connection and design margin.

At the k-th propagation, the change propagation intensity is defined as:

${\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}\mathrm{<span\; class="notranslate">\; 0</span>}& {\mathrm{<span\; class="notranslate">\; \&Delta;\&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}\\ {\mathrm{<span\; class="notranslate">\; \&omega;</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; \&lsqb;</span>{\mathrm{<span\; class="notranslate">\; \&omega;</span>}}_{\mathrm{<span\; class="notranslate">\; p</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&omega;</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}}\frac{{\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}}{\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; f</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}}<span\; class="notranslate">\; \&rsqb;</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>\frac{{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{{\mathrm{<span\; class="notranslate">\; \&Delta;\&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}}<span\; class="notranslate">\; )</span>& {\mathrm{<span\; class="notranslate">\; \&Delta;\&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; ></span>{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}\end{array}\right.<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; f</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$

where ρ _i is the change margin of the design attribute i at the k-th propagation, Δρ ^k _i is the required change value; F ^k represents the set of design attributes affected by the k-th change propagation; ω _p and ω _d are respectively is the weight value of propagation probability and node degree and ω _p +ω _d = 1; d _j is the node degree which can be calculated through the adjacency matrix of the design attribute; ω _s (ω _s ≥ 1) is the penalty coefficient of long chain connection, using Increases change propagation strength when changes are propagated across parts.

The greater the intensity of change propagation, the greater the amount of change required for subsequent attributes connected to it. In other words, change propagation along connections with high propagation strength affects subsequent properties to a greater extent.

4. The change propagation path optimization module takes the cumulative change propagation strength as the objective function, and obtains the optimal change propagation path corresponding to the minimized maximum cumulative change propagation strength through a heuristic optimization algorithm. Changes to design properties can be propagated through different paths, even if some design properties are directly connected to different changed properties. In order to prevent changes in key design attributes or avalanche-style change propagation, it is necessary to properly plan the change propagation path before implementing the change. Due to the complexity of the design change analysis network model and the variability of the change propagation intensity model, it is necessary to use heuristic optimization algorithms to optimize the change propagation path. In the present invention, an optimal path for change propagation is obtained by using an ant colony optimization algorithm. The propagation path has the minimized maximum cumulative change propagation strength. Based on the above design change analysis network model and change propagation intensity model, the optimization objective function of the change propagation path is established as follows:

$\left.\begin{array}{cc}\mathrm{<span\; class="notranslate">\; arg</span>}\mathrm{<span\; class="notranslate">\; min</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; max</span>}\underset{\mathrm{<span\; class="notranslate">\; k</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>& \mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; ...</span><span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; N</span>}\\ \mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; .</span><span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; \&Delta;\&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; <</span>{\mathrm{<span\; class="notranslate">\; 10</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 5</span>}}& \mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; f</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}\end{array}\right\}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; )</span>$

The above formula defines the total propagation strength when a change to an initial design property is propagated to other design properties. When the difference between the design margin and the design variable is less than 10 ^-5 , the change propagation converges and no further changes affecting other design properties are produced. After each iteration, the pheromone on the connection is updated according to the following rules:

τ _ij ←(1-ρ)τ _ij +Δτ _ij (5)

${\mathrm{<span\; class="notranslate">\; \&Delta;\&tau;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; N</span>}}_{\mathrm{<span\; class="notranslate">\; a</span>}}}{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; \&Delta;\&tau;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; l</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 6</span>}<span\; class="notranslate">\; )</span>$

Among them, τ _ij is the pheromone value released from i to j; ρ(0<ρ<1) is the pheromone volatilization coefficient; N _a is the number of ant colonies, Δτ ^l _ij is the pheromone value released by the l ant .

Among them, Q is a constant, and D _l is the objective function value of the lth ant walking.

change tends to select the one with the largest Adjacent design properties of the propagating, the state transition willingness η _ij is directly related to Therefore, η _ij is defined as follows:

${\mathrm{<span\; class="notranslate">\; \&eta;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; )</span>$

After each iteration, η _ij according to the current The value is updated. The probability of the l-th ant climbing from design attribute i to design attribute j is

Among them, α(0≤α) is a parameter controlling the degree of influence of τ _ij , and β(β≥1) is a parameter controlling the degree of influence of η _ij . N ^l _i is the allowable adjacent node set of the l-th ant in the design attribute i. τ _ij and η _ij represent the degree of attraction and deviation to other possible states.

5. Product design change database: According to the designer’s change record of the product, on the one hand, through the establishment of a database, the change records are recorded according to the design attribute serial number, forward design attribute serial number, attribute name, value before change, value after change, date of change, etc. Organization, which facilitates the organization and evaluation of design changes; on the one hand, the establishment of a change impact forward index, that is, the number of occurrences of other design attributes corresponding to each design attribute, facilitates the rapid calculation of the propagation probability.

6. Propagation optimization result display module: display the optimization result of the design change propagation path in the user interface, and use a certain design attribute as the initial change node. An example of the propagation path optimization result obtained according to the present invention is shown in Figure 3. The displayed content includes: design attributes, connections between design attributes, change propagation intensity values, and change propagation paths.

Although the specific implementation of the present invention has been described above in conjunction with the accompanying drawings, it does not limit the protection scope of the present invention. Those skilled in the art should understand that on the basis of the technical solution of the present invention, those skilled in the art do not need to pay creative work Various modifications or variations that can be made are still within the protection scope of the present invention.

Claims (10)

1. A design change propagation prediction method based on a design network, characterized in that it comprises the steps of: Step (1): Establish a design change analysis network model; Step (2): Calculate the change propagation probability between the design attributes stored in the product design change database; use the change propagation probability between the design attributes as the weight of the connection in the design change analysis network model; thus establish a more objective design change analysis network model; Step (3): Establish a design change propagation intensity model by using the propagation probability, degree of design attribute, design attribute change margin, weight, and penalty coefficient of long-chain connections, and calculate the design attribute change propagation intensity; Step (4): Optimize the change propagation path: take the change propagation intensity as the objective function, and obtain the change propagation path corresponding to the minimum cumulative change propagation intensity on the design change analysis network through the ant colony optimization algorithm. The change propagation path is optimal propagation path. 2. A design network-based design change propagation prediction method as claimed in claim 1, further comprising step (5): a propagation optimization result display module, displaying the design change propagation path optimization result in the user interface . 3. a kind of design change propagation prediction method based on design network as claimed in claim 1, is characterized in that, the step of described step (1) is: Design attributes are used as network nodes, and the connection relationship between parent and child design attributes is expressed as the connection between nodes. According to the design attributes, it is gradually refined from top to bottom, and finally a design change analysis network model is formed; the relationship between design attributes belonging to the same part The connection relationship between parts is defined as a short chain, and the connection relationship between design attributes across parts is defined as a long chain. 4. A kind of design change propagation prediction method based on design network as claimed in claim 1, it is characterized in that, Said step (1): Use the set V={v ₁ ,v ₂ ,...,v _n } to correspond to the design attribute set, and define E as the connection set between nodes, and the design change analysis network model is recorded as G={V,E}; The design attribute is represented as a circle, and the connection relationship is represented as a line with an arrow. The design attribute is clustered according to the part to which it belongs, and each category is distinguished by a dotted line; the connection relationship includes two types of parameter connection relationship and constraint connection relationship. 5. A kind of design change propagation prediction method based on design network as claimed in claim 4, it is characterized in that, The parameter connection relationship is expressed as a function form y=f(x ₁ ,x ₂ ,...,x _n ), where y is the parent design attribute, and x _i represents the i-th child design attribute; in the design change analysis network model The arrow points from the sub-design attribute to the parent design attribute; a parameter connection relationship is composed of a parent design attribute and multiple sub-design attributes, and the value of the parent design attribute is determined by the values of the independent sub-design attributes. $\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span><span\; class="notranslate">\; \&part;</span>\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; /</span><span\; class="notranslate">\; \&part;</span>{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ;</span>$ In the design change analysis network model, because the sub-design attributes are independent of each other, the change of the sub-design attributes does not directly change the sibling attributes of the sub-design attributes, but indirectly changes the sibling attributes of the sub-design attributes by changing the parent design attributes; The change of the parent design attribute is realized by changing its child design attribute, and the size of the change is set artificially; therefore, in the parameter connection relationship, the parent design attribute changes with the change of any child design attribute; when the parent design attribute changes , if the parent design attribute has a child design attribute, the child design attribute is changed, if the parent design attribute has multiple child design attributes, each child design attribute is changed according to different setting change weights. 6. A kind of design change propagation prediction method based on design network as claimed in claim 4, it is characterized in that, The constraint connection relationship is an artificially formulated design rule to meet the assembly constraints, functional coordination and performance guarantee between components, expressed by a function f(x ₁ ,x ₂ ,...,x _n )≥0; Each design attribute in the constraint connection relationship has the same dimension, which is represented by a two-way arrow in the design change analysis network model; When any constraint attribute is changed in the constraint connection relationship, other constraint attributes must also be changed to continue to ensure that the connection conforms to the constraint relationship; In the constraint connection relationship, the change of any attribute will cause the change of other connection attributes, and the change weight is artificially set. 7. A kind of design change propagation prediction method based on design network as claimed in claim 1, it is characterized in that, Said step (2): In the design change analysis network model, each connection has a weight, that is, the propagation probability P _ij ; P _ij is the possibility of the successor attribute changing when the predecessor attribute i changes; P _ij is equal to a record in the design change database The conditional probability of attribute j appearing under the premise that design attribute i appears in , that is ${\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; \&cap;</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; |</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>\frac{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}\mathrm{<span\; class="notranslate">\; i</span>}}\frac{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span>}{\mathrm{<span\; class="notranslate">\; P</span>}<span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; v</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; )</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$ In formula (1), since P(v _i ) and P(v _j ) are not equal, P _ij and P _ji are not equal; if there is no connection between the two design attributes, P _ij is equal to 0; and design attribute i and design attribute The sum of the propagation probabilities of all adjacent attributes of i is 1; $\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; f</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; )</span>$ The parameter meaning of P(v _j |v _i ) is the conditional probability of attribute j appearing under the premise that design attribute i appears in a record in the design change database; the parameter meaning of P(v _i ∩v _j ) is the design attribute i and The probability that j appears in a record in the design change database at the same time; the parameter meaning of P(v _i ) is the probability of design attribute i appearing in the design change database; the parameter meaning of P(v _j ) is the design attribute j appearing in the design change database The parameter meaning of P(v _i |v _j ) is the conditional probability of attribute i appearing under the premise that design attribute j appears in a record in the design change database; the parameter meaning of P _ji is the change of current driving attribute j The possibility that the successor attribute i will change. 8. A kind of design change propagation prediction method based on design network as claimed in claim 1, it is characterized in that, Said step (3): At the k-th propagation, the design attribute changes the propagation intensity defined as: ${\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}\mathrm{<span\; class="notranslate">\; 0</span>}& {\mathrm{<span\; class="notranslate">\; \&Delta;\&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}\\ {\mathrm{<span\; class="notranslate">\; \&omega;</span>}}_{\mathrm{<span\; class="notranslate">\; the\; s</span>}}<span\; class="notranslate">\; \&lsqb;</span>{\mathrm{<span\; class="notranslate">\; \&omega;</span>}}_{\mathrm{<span\; class="notranslate">\; p</span>}}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; \&omega;</span>}}_{\mathrm{<span\; class="notranslate">\; d</span>}}\frac{{\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}}{\underset{\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; f</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}}<span\; class="notranslate">\; \&rsqb;</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>\frac{{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}}{{\mathrm{<span\; class="notranslate">\; \&Delta;\&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}}<span\; class="notranslate">\; )</span>& {\mathrm{<span\; class="notranslate">\; \&Delta;\&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; \&le;</span>{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}\end{array}\right.<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; f</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; )</span>$ Among them, ρi is the change margin of the design attribute _i in the k-th propagation, is the required change value of the design attribute i in the k-th propagation; F ^k represents the set of design attributes affected by the k-th change propagation; ω _p is the weight value of the propagation probability, and ω _d is the weight value of the degree of the design attribute , and ω _p +ω _d ＝1; d _j is the degree of the design attribute, which is calculated by the adjacency matrix of the design attribute; ω _s is the penalty coefficient for long-chain connections, where ω _s ≥ ₁ , Increased change propagation strength when parts are propagated. 9. A kind of design change propagation prediction method based on design network as claimed in claim 1, it is characterized in that, The step (4): Based on the design change analysis network model and the change propagation intensity model, the optimization objective function of the change propagation path is established as follows: $\left.\begin{array}{cc}\mathrm{<span\; class="notranslate">\; arg</span>}\mathrm{<span\; class="notranslate">\; min</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; max</span>}\underset{\mathrm{<span\; class="notranslate">\; k</span>}}{<span\; class="notranslate">\; \&Sigma;</span>}{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; )</span>& \mathrm{<span\; class="notranslate">\; k</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; ...</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; N</span>}\\ \mathrm{<span\; class="notranslate">\; the\; s</span>}<span\; class="notranslate">\; .</span>\mathrm{<span\; class="notranslate">\; t</span>}<span\; class="notranslate">\; .</span><span\; class="notranslate">\; (</span>{\mathrm{<span\; class="notranslate">\; \&Delta;\&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span>{\mathrm{<span\; class="notranslate">\; \&rho;</span>}}_{\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; <</span>{\mathrm{<span\; class="notranslate">\; 10</span>}}^{<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 5</span>}}& \mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; f</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}\end{array}\right\}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; )</span>$ Equation (4) defines the total propagation intensity when the initial design attribute changes to other design attributes; when the design margin ρ _j and the design variable When the difference between is less than 10 ^-5 , it means that the convergence of change propagation does not further affect the change of other design attributes; after each iteration, the pheromone left by the connected ant colony is updated according to the following rules: ${\mathrm{<span\; class="notranslate">\; \&tau;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; \&rho;</span>}<span\; class="notranslate">\; )</span>{\mathrm{<span\; class="notranslate">\; \&tau;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; +</span>\underset{\mathrm{<span\; class="notranslate">\; l</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\overset{{\mathrm{<span\; class="notranslate">\; N</span>}}_{\mathrm{<span\; class="notranslate">\; a</span>}}}{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; \&Delta;\&tau;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; l</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; )</span>$ Among them, τ _ij is the value of pheromone released from design attribute i to design attribute j, ρ is the pheromone volatilization coefficient, where 0<ρ<1; N _a is the number of ant colonies, is the pheromone value released by the lth ant; τ _ij represents the degree of attraction to other possible states; Among them, Q is a constant, and D _l is the objective function value of the lth ant's walk; change tends to select the one with the largest Adjacent design properties of the propagating, the state transition willingness η _ij is directly related to Therefore, the state transition willingness η _ij is defined as follows: ${\mathrm{<span\; class="notranslate">\; \&eta;</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; =</span>{\mathrm{<span\; class="notranslate">\; I</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}\mathrm{<span\; class="notranslate">\; j</span>}}^{\mathrm{<span\; class="notranslate">\; k</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; )</span>$ After each iteration, the state transition willingness η _ij according to the current The value is updated; the probability that the lth ant climbs from the design property i to the design property j for Among them, α is a parameter controlling the degree of influence of τ _ij , α≥0, and β is a parameter controlling the degree of influence of η _ij , β≥1; is the directly connected attribute set of the l-th ant in the design attribute i. 10. A design change propagation prediction system based on a design network, characterized by comprising: Design change analysis network model: use design attributes as network nodes, and express the connection relationship between parent and child design attributes as the connection between nodes; the connection relationship between design attributes belonging to the same part is defined as a short chain, and the connection relationship between parts Defined as a long chain; according to the attribute to distinguish the connection relationship, it can be divided into parameter connection relationship and constraint connection relationship; Product design change database: responsible for storing each affected design attribute and the degree of influence after the product design change occurs, and is used to calculate the propagation impact probability between each design attribute; Propagation probability calculation module: calculate the change propagation probability between design attributes as the weight of connections in the design change analysis network model; establish an objective design change analysis network model; Change propagation intensity analysis module: establish a design change propagation intensity model using the degree of each design attribute, design margin, connection weight, and penalty coefficient for long-chain connections; Change propagation path optimization module: take the cumulative change propagation intensity as the objective function, and obtain the optimal change propagation path corresponding to the minimized maximum cumulative change propagation intensity through the ant colony optimization algorithm; Propagation optimization result display module: display the design change propagation path optimization results in the user interface, the displayed content includes: design attributes, connections between design attributes, change propagation intensity values, and change propagation paths.