CN110348954B - Complex process module dividing method for large-scale customization - Google Patents

Abstract

The invention discloses a method for dividing a complex process module oriented to large-scale customization, which provides a method for dividing a model on a process layer by means of a characteristic relation between processes corresponding to workpieces, constructs a fuzzy incidence matrix between process modules, utilizes a fuzzy clustering analysis method to perform conversion solution on the fuzzy incidence matrix to obtain a transfer closure matrix, forms a module clustering diagram according to different partition threshold sequences, obtains different module division schemes by selecting different lambda values, has compact logic and accurate and efficient control, and aims to realize the quick and efficient production of personalized finished products adapting to market and technical change and improve the high-level requirements of the process.

Description

Complex process module dividing method for large-scale customization

Technical Field

The invention belongs to the technical field of complex process planning, and relates to a complex process module dividing method for large-scale customization.

Background

At present, rapid economic development and continuous technical progress promote the personalized customization process to gradually gain the favor of the economic market, provide customized service for customers, comprehensively improve the satisfaction degree of the customers, and become a competitive trend among modern enterprises.

The advantages of the large-scale customized production mode and the mass production are combined, the personalized requirements of customers are met, meanwhile, the production task can be guaranteed to be completed within a short production period at low cost, and the unique advantages of the large-scale customized production mode attract wide attention of manufacturing enterprises. One of the large-scale customization core technologies is a modularization method, which is used for partitioning a general module, a customization module and a personalized module which have relatively independent functions, structures and performances on the basis of structural analysis of different functional characteristics of a product family or different geometric characteristics and performance characteristics of the same function, and can quickly and efficiently partition the modules of parts of a complex product through matching combination among the modules to produce a process product and a service which meet the personalized requirements of customers.

At present, a modular design method is widely applied to system module division of a product family, but the conventional performance division of the product is considered more in the existing division method, the idea of individual requirements is not integrated, and the method is not suitable for creating a process module with complex product. Compared with the establishment of a product component module, the establishment of a process module needs to incorporate process information of a product, and in order to solve the problem, the invention designs a complex process module dividing method oriented to large-scale customization, and intelligently divides the processing process of the product by taking the characteristic attribute of the process as a criterion, thereby realizing the personalized requirements of customers.

Disclosure of Invention

In view of the above, to solve the above-mentioned deficiencies of the prior art, the present invention aims to provide a method for partitioning a complex process module oriented to large-scale customization, which has compact logic, precise control and high efficiency, and provides a method for partitioning a model at a process level by means of a characteristic relationship between processes corresponding to a workpiece.

In order to achieve the purpose, the technical scheme adopted by the invention is as follows:

a large-scale customization-oriented complex process module dividing method comprises the following steps:

s1: establishing module characteristic attributes by a modular mapping mechanism, establishing the relevance among process modules by the attributes, and describing the relevance as replaceable relevance and influence degree relevance;

s2: the correlation degree between the process modules is related to the substitutability, the influence degree and the like, and is represented by a mathematical model of a fuzzy correlation matrix R: r ═ ω_fF+ω_pP, where F, P denotes the defined influence correlation matrix and the alternative correlation matrix, ω, respectively_f、ω_pRepresenting the weight coefficient, ω, to which the characteristic corresponds_f、ω_p∈[0，1]And satisfies the following conditions: omega_f+ω_p＝1；

S3: converting and solving the fuzzy incidence matrix model by using a fuzzy clustering analysis method to obtain a transfer closure matrix;

s31: determining the correlation among all the working procedures and determining the correlation degree among all the process modules;

s32: and (3) data standardization treatment: according to different original formats, expression formats and level quantization modes, translation and range transformation are adopted to carry out data standardization processing, and each data of the fuzzy association matrix is compressed to [0, 1 ];

s33: establishing a fuzzy similarity matrix R': the fuzzy incidence matrix R and the fuzzy similarity matrix R' are in equivalent relation, and a direct Euclidean distance method is applied, wherein R is_(i,j)＝1-c×d(x_i,x_j) Wherein C is a parameter of any selected region, so that r is more than or equal to 0_(i,j)≤1，d(x_i,x_j) Denotes x_iAnd x_jThe distance between:

s34: solving a transitive closure matrix

: according to the theorem, the fuzzy equivalent matrix is obtained by using the fuzzy incidence matrix R by a quadratic method

I.e. transitive closure matrix

(ii) a And, make

S35: solving the truncation matrix R_λ＝λ_(i,j): the method comprises the following steps that lambda is used as a metric value to represent a truncation matrix coefficient, a fuzzy relation matrix between modules is truncated, elements which are larger than or equal to lambda in a fuzzy association matrix R are provided with the numerical value of 1, the numerical value which is smaller than the lambda element is provided with 0, the elements with the numerical value of 1 in the same row or column are gathered into the same module, the rest elements independently become the modules, and the more the lambda value is, the more detailed the module division is; different process module clustering division results can be obtained by selecting different lambda values;

s36: with the help of Matlab tool and the application of fuzzy clustering analysis method, the standardized matrix, fuzzy similar matrix, transfer closure matrix can be calculated in turn according to the original incidence matrix, and finally the fuzzy incidence matrix R is cut by using lambda-cut matrix to form a module clustering graph:

s4: and forming an integral module clustering graph according to different partition threshold sequences, and obtaining different module division schemes by selecting different lambda values.

Further, in step S2, the method for establishing the fuzzy association matrix specifically includes the following steps:

a1: and (3) taking the attribute value domain of the output parameter as the correlation for judging each processing procedure, describing the correlation as the correlation of replaceable correlation and influence degree, and expressing a fuzzy correlation matrix R by using a mathematical model: r is_i,j＝ω_ff_i,j+ω_pp_i,jWherein r is_i,jRepresenting the overall correlation coefficient between module i and module j, f_i,j、p_i,jRespectively representing the number of subphases, ω, between module i and module j_f、ω_pRepresenting a weight coefficient corresponding to the characteristic;

a2: the fuzzy correlation matrix R is also called an original matrix R, and R satisfies the self-reflexivity: r (i, j) is not less than 0 and not more than 1, and r (i, i) is r (j, j) is 1; and R satisfies symmetry: r (i, j) ═ R (j, i), the mathematical model of the fuzzy correlation matrix R in step a1 is expressed as:

further, in the step A1, r_i,jRepresenting the overall correlation coefficient between module i and module j, f_i,jRepresenting the correlation coefficient, p, between block i and block j_i,jDenotes the correlation coefficient, ω, between block j and block i_f、ω_pRepresenting a weight coefficient corresponding to the characteristic; omega_f、ω_p∈[0，1]And satisfies the following conditions: omega_f+ω_p＝1。

Further, in the step a1, the output parameters are grain size, yield strength, elongation, residual stress, corrosion resistance, electrical conductivity, hardness, and tensile strength.

Further, in step S32, the normalization formula used in the data normalization process is:

wherein, i is 1,2₍′_i,j)Representing the original data; x is the number of₍'_i,j)man、x₍'_i,j)minRespectively representing the maximum value and the minimum value of the original data; x is the number of_(i,j)The data obtained after normalization.

Further, in step S34, a transitive closure matrix is obtained

The method specifically comprises the following steps:

a1: set up R₀Is R, wherein R₀Representing the original incidence matrix of the module;

a2: let R be_iKnowing the result, start to compare R_iAnd

the corresponding element; if it is not

Then R is_iIs that

The transitive closure matrix of (2), at which point computation may be stopped; if not satisfied

If yes, continuing to execute the next step;

a3: by algorithmic calculation

And setting the result to R_i+1Executing the step A2;

a4: repeating the above steps until finally solving the transfer closure matrix to ensure that

The invention has the beneficial effects that:

a method for dividing a complex process module oriented to large-scale customization is compact in logic, accurate in control and high-efficiency, a method for dividing a model on a process layer is provided by means of characteristic relation between processes corresponding to workpieces, a fuzzy association matrix between process modules is constructed, the fuzzy association matrix is converted and solved by using a fuzzy clustering analysis method to obtain a transfer closure matrix, a module clustering diagram is formed according to different partition threshold sequences, different module division schemes are obtained by selecting different lambda values, and therefore personalized finished products adapting to market and technical changes and high-level requirements of the process are rapidly and efficiently produced;

compared with the traditional product family module division method, the method is mainly based on the analysis of mature products of enterprises and processing technologies of products with market demands, combines the process information extraction, the process creative design and the process modularization to construct a modular process design platform, modularizes the serialized product processes, matches the similarity of the modules by combining the module configuration technology and the order requirements, and finally carries out the deformation design of the corresponding degree on the process requirements by the deformation design module to realize the functions of different product requirements.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the drawings without creative efforts.

FIG. 1 is a functional block diagram of a modular process domain mapping mechanism of the present invention;

FIG. 2 is a process flow diagram of the production of the aluminum/copper sheet strip in the example;

FIG. 3 is a block clustering diagram of the process sequence of the example.

Detailed Description

The following specific examples are given to further clarify, complete and detailed the technical solution of the present invention. The present embodiment is a preferred embodiment based on the technical solution of the present invention, but the scope of the present invention is not limited to the following embodiments.

A large-scale customization-oriented complex process module dividing method comprises the following steps:

s1: establishing module characteristic attributes by a modular mapping mechanism, establishing the relevance among process modules by the attributes, and describing the relevance as replaceable relevance and influence degree relevance;

s2: the correlation degree between the process modules is related to the substitutability, the influence degree and the like, and is represented by a mathematical model of a fuzzy correlation matrix R: r ═ ω_fF+ω_pP；

S3: converting and solving the fuzzy incidence matrix model by using a fuzzy clustering analysis method to obtain a transfer closure matrix;

s31: determining the correlation among all the working procedures and determining the correlation degree among all the process modules;

s32: and (3) data standardization treatment: according to different original formats, expression formats and level quantization modes, translation and range transformation are adopted to carry out data standardization processing, and each data of the fuzzy association matrix is compressed to [0, 1 ];

s33: establishing a fuzzy similarity matrix R': the fuzzy incidence matrix R and the fuzzy similarity matrix R' are in equivalent relation, and a direct Euclidean distance method is applied, wherein R is_(i,j)＝1-c×d(x_i,x_j) Wherein C is a parameter of any selected region, so that r is more than or equal to 0_(i,j)≤1，d(x_i,x_j) Denotes x_iAnd x_jThe distance between:

s34: solving a transitive closure matrix

: according to the theorem, the fuzzy equivalent matrix is obtained by using the fuzzy incidence matrix R by a quadratic method

I.e. transitive closure matrix

(ii) a And, make

S35: solving the truncation matrix R_λ＝λ_(i,j): the method comprises the following steps that lambda is used as a metric value to represent a truncation matrix coefficient, a fuzzy relation matrix between modules is truncated, elements which are larger than or equal to lambda in a fuzzy association matrix R are provided with the numerical value of 1, the numerical value which is smaller than the lambda element is provided with 0, the elements with the numerical value of 1 in the same row or column are gathered into the same module, the rest elements independently become the modules, and the more the lambda value is, the more detailed the module division is; by selectingObtaining different process module clustering division results by taking different lambda values;

s36: with the help of Matlab tool and the application of fuzzy clustering analysis method, the standardized matrix, fuzzy similar matrix, transfer closure matrix can be calculated in turn according to the original incidence matrix, and finally the fuzzy incidence matrix R is cut by using lambda-cut matrix to form a module clustering graph:

s4: and forming an integral module clustering graph according to different partition threshold sequences, and obtaining different module division schemes by selecting different lambda values.

Further, in step S2, the method for establishing the fuzzy association matrix specifically includes the following steps:

a1: and (3) taking the attribute value domain of the output parameter as the correlation for judging each processing procedure, describing the correlation as the correlation of replaceable correlation and influence degree, and expressing a fuzzy correlation matrix R by using a mathematical model: r is_i,j＝ω_ff_i,j+ω_pp_i,jWherein r is_i,jRepresenting the overall correlation coefficient between module i and module j, f_i,j、p_i,jRespectively representing the number of subphases, ω, between module i and module j_f、ω_pRepresenting a weight coefficient corresponding to the characteristic;

a2: the fuzzy correlation matrix R is also called an original matrix R, and R satisfies the self-reflexivity: r (i, j) is not less than 0 and not more than 1, and r (i, i) is r (j, j) is 1; and R satisfies symmetry: r (i, j) ═ R (j, i), the mathematical model of the fuzzy correlation matrix R in step a1 is expressed as:

further, in the step A1, r_i,jRepresenting the overall correlation coefficient between module i and module j, f_i,jRepresenting the correlation coefficient, p, between block i and block j_i,jDenotes the correlation coefficient, ω, between block j and block i_f、ω_pRepresenting a weight coefficient corresponding to the characteristic; omega_f、ω_p∈[0，1]And satisfies the following conditions: omega_f+ω_p＝1。

Further, in the step a1, the output parameters are grain size, yield strength, elongation, residual stress, corrosion resistance, electrical conductivity, hardness, and tensile strength.

Further, in step S32, the normalization formula used in the data normalization process is:

wherein, i is 1,2₍′_i,j)Representing the original data; x is the number of₍'_i,j)man、x₍'_i,j)minRespectively representing the maximum value and the minimum value of the original data; x is the number of_(i,j)The data obtained after normalization.

Further, in step S34, a transitive closure matrix is obtained

The method specifically comprises the following steps:

a1: set up R₀Is R, wherein R₀Representing the original incidence matrix of the module;

a2: let R be_iKnowing the result, start to compare R_iAnd

the corresponding element; if it is not

Then R is_iIs that

The transitive closure matrix of (2), at which point computation may be stopped; if not satisfied

Continuing to execute the next step;

a3: by algorithmic calculation

And setting the result to R_i+1Executing the step A2;

a4: repeating the above steps until finally solving the transfer closure matrix to ensure that

The specific working process of this embodiment is as follows:

1. as shown in fig. 1, each process module has a model representing its own characteristics, each characteristic model may be represented by a plurality of attributes, when the attributes of the module have a plurality of attribute values, the attribute values may be collectively referred to as variables, a common attribute set capable of distinguishing instance modules is defined as a variable set p (x) of process-like modules, and the variable set is used to distinguish differences between the models of the same type, as shown in table 1. The reason for personalized customization is that module models are predefined in a process model library in the process of configuring process modules according to customer requirements, the models are independent of each other and have a plurality of attributes and attribute values, and different modules are called to be combined according to different requirements of customers. P (x) ═ part_i|1≤i≤n}part_i＝{val_i,j|1≤i≤n,1≤j≤m}，Part_iRepresenting configuration constraints between the inner classes;

TABLE 1 Process Module Attribute variables and value ranges

2. Modular design and fuzzy incidence matrix model establishment

(1) Establishing a fuzzy incidence matrix R

The correlation among the complex product process procedures refers to the degree of correlation in the aspects of researching the characteristics, performance, structure and the like among the configuration modules. Combined upperThe value ranges of the attribute variables are described, and output parameters such as grain size, yield strength, elongation, residual stress, corrosion resistance, electrical conductivity, hardness, tensile strength and the like are used as the correlation between the various processing procedures, and the correlation is described as a replaceable correlation and an influence degree correlation, as shown in tables 2 and 3. Expressed by a mathematical model: r is_i,j＝ω_ff_i,j+ω_pp_i,jWherein r is_i,_jRepresenting the correlation coefficient between block i and block j, f_i,j、p_i,jRespectively representing the correlation coefficient, ω, between module i and module j_f、ω_pRepresenting the weight coefficient, ω, to which the characteristic corresponds_f、ω_p∈[0，1]And satisfies the following conditions: omega_f+ω _p1, thereby forming a fuzzy incidence matrix R of the relation between the process modules;

TABLE 2 alternative correlations

TABLE 3 correlation of influence

Taking an aluminum/copper plate strip series product produced by a certain enterprise as an example, the production process of the plate strip can be divided into the following steps by carrying out on-site research on the enterprise and integrating the related production processes provided by personnel in the production department: casting, hot rolling, cold rolling and heat treatment. The specific process flow of casting is as follows: batching, feeding, melting, stirring, slagging off, analyzing and sampling, adding alloy to adjust components, stirring, refining, standing, guiding to a furnace, casting and the like; the hot rolling process flow comprises the following steps: ingot casting, surface milling and edge milling, heating, hot rolling (cogging rolling), straightening and surface milling; the cold rolling process flow comprises the following steps: rough rolling, annealing, pickling, annealing, finish rolling, and the final heat treatment is finished product annealing, which is divided into complete annealing and low-temperature annealing, as shown in table 4. Soft, hard or semi-hard products in different states are obtained by controlling the annealing temperature and the heat preservation time, so as to stabilize the size, shape and performance of the material. Analyzing the correlation degree between the whole production process and the whole production process by combining the whole production process and the whole production process, and constructing a correlation matrix between the whole production process and the whole production process;

table 4 full process flow module based on process route

Serial number	Name of procedure	Serial number	Name of procedure
				1	Casting blank	11	Intermediate annealing
2	Solution heat treatment	12	Acid pickling
				3	Hot rolling	13	Finish rolling
4	Shearing	14	Annealing the finished product
				5	Straightening milled surface	15	Cleaning finished product
6	Homogenizing annealing	16	Cutting and slitting
				7	Rough rolling	17	Acceptance inspection
8	Intermediate annealing	18	Package (I)
				9	Acid pickling	19	Put in storage
10	Pre-finish rolling

(2) Module division method based on fuzzy clustering analysis method

1) And (3) data standardization treatment: the original formats of the data information are different greatly, wherein the data information comprises description type, numerical type, selection type and other information expression formats, and the data grade quantization modes are different, so that the data information cannot be directly applied to operation, the data information needs to be subjected to standardization processing, and each data of the fuzzy matrix is compressed to [0, 1], so that the subsequent algorithm operation is facilitated;

2) establishing a fuzzy similarity matrix R': the fuzzy matrix can be obtained from the characteristic table, the fuzzy matrix R and the fuzzy similar matrix R' are in equivalent relation, and a direct Euclidean distance method is applied: r is_(i，j)＝1-c×d(x_i，x_j) Wherein C is a parameter of any selected region, so that r is more than or equal to 0_(i，j)≤1，d(x_i，x_j) Denotes x_iAnd x_jThe distance between

By using the module division method, an expert evaluates and scores to determine the correlation among all the processes, and if the casting blank and the solution heat treatment process are closely connected two processes and the time interval of the action is small, the influence degree correlation is 0.8, and the alternative correlation is 0.5. The weighting coefficients are taken to be 0.55 and 0.45 respectively according to the formula. The correlation between the two was calculated to be 0.665, and similarly, the influence degree correlation and the alternative correlation between the solution heat treatment and the hot rolling were taken to be 0.4 and 0.1, respectively, and the total correlation was 0.265. The correlation between the other steps was calculated in sequence, as shown in table 5. Forming a fuzzy incidence matrix R after correlation analysis and correlation calculation;

TABLE 5 fuzzy association matrix R of major process blocks

3) Solving a transitive closure matrix

: according to the fuzzy incidence matrix R obtained by calibration, only one fuzzy similar matrix is obtained

As shown in Table 6, has reflexibility and contraindicationSymmetry, but not necessarily transitivity, i.e. R is not necessarily a fuzzy equivalent matrix, and needs to be transformed into a fuzzy equivalent matrix R^kAccording to the theorem, the transitive closure matrix is solved by a quadratic method

Is the fuzzy equivalent matrix R sought^kAs shown in table 7;

TABLE 6 fuzzy similarity matrix R 'for relationships between process modules'

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
																	1	1	0.806	0.622	0.522	0.570	0.575	0.470	0.497	0.428	0.456	0.521	0.448	0.508	0.536	0.463	0.471
2	0.806	1	0.621	0.523	0.557	0.658	0.517	0.561	0.398	0.520	0.596	0.414	0.516	0.610	0.412	0.480
																	3	0.622	0.621	1	0.726	0.753	0.786	0.558	0.508	0.398	0.529	0.475	0.409	0.517	0.478	0.369	0.528
4	0.522	0.523	0.726	1	0.869	0.635	0.527	0.406	0.383	0.517	0.400	0.372	0.535	0.426	0.410	0.552
																	5	0.570	0.557	0.753	0.869	1	0.671	0.539	0.453	0.405	0.536	0.445	0.406	0.550	0.455	0.416	0.566
6	0.575	0.658	0.786	0.635	0.671	1	0.625	0.614	0.445	0.596	0.569	0.404	0.544	0.512	0.344	0.526
																	7	0.470	0.517	0.558	0.527	0.539	0.625	1	0.740	0.657	0.788	0.661	0.559	0.558	0.491	0.422	0.560
8	0.497	0.561	0.508	0.406	0.453	0.614	0.740	1	0.593	0.702	0.791	0.538	0.497	0.543	0.377	0.463
																	9	0.428	0.398	0.398	0.383	0.405	0.445	0.657	0.593	1	0.661	0.580	0.837	0.494 0	0.441	0.604	0.500
10	0.456	0.520	0.529	0.511	0.536	0.596	0.788	0.702	0.661	1	0.657	0.591	0.606	0.513	0.470	0.615
																	11	0.521	0.596	0.475	0.400	0.445	0.569	0.661	0.791	0.580	0.657	1	0.545	0.575	0.655	0.450	0.531
12	0.448	0.414	0.409	0.372	0.406	0.404	0.559	0.538	0.837	0.591	0.545	1	0.507	0.491	0.680	0.540
																	13	0.508	0.516	0.517	0.535	0.550	0.544	0.558	0.497	0.490	0.606	0.575	0.507	1	0.711	0.597	0.777
14	0.536	0.610	0.478	0.426	0.455	0.512	0.491	0.543	0.441	0.513	0.655	0.491	0.711	1	0.583	0.661
																	15	0.463	0.412	0.369	0.410	0.416	0.344	0.422	0.377	0.604	0.470	0.450	0.680	0.597	0.583	1	0.663
16	0.471	0.480	0.528	0.552	0.566	0.526	0.560	0.463	0.500	0.615	0.531	0.540	0.777	0.661	0.663	1

TABLE 7 transitive closure matrix

	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
																	1	1	0.80	0.62	0.52	0.57	0.57	0.47	0.49	0.42	0.45	0.52	0.44	0.50	0.53	0.46	0.47
2	0.80	1	0.62	0.52	0.55	0.65	0.51	0.56	0.39	0.52	0.59	0.41	0.51	0.61	0.41	0.48
																	3	0.62	0.62	1	0.72	0.75	0.78	0.55	0.50	0.39	0.52	0.47	0.40	0.51	0.47	0.36	0.52
4	0.52	0.52	0.72	1	0.86	0.63	0.52	0.40	0.38	0.51	0.40	0.37	0.53	0.42	0.41	0.55
																	5	0.57	0.55	0.75	0.86	1	0.67	0.53	0.45	0.40	0.53	0.44	0.40	0.55	0.45	0.41	0.56
6	0.57	0.65	0.78	0.63	0.67	1	0.62	0.61	0.44	0.59	0.56	0.40	0.54	0.51	0.34	0.52
																	7	0.47	0.51	0.55	0.52	0.53	0.62	1	0.74	0.65	0.78	0.66	0.55	0.55	0.49	0.42	0.56
8	0.49	0.56	0.50	0.40	0.45	0.61	0.74	1	0.59	0.70	0.79	0.53	0.49	0.54	0.37	0.46
																	9	0.42	0.39	0.39	0.38	0.40	0.44	0.65	0.59	1	0.66	0.58	0.83	0.49	0.44	0.60	0.50
10	0.45	0.52	0.52	0.51	0.53	0.59	0.78	0.70	0.66	1	0.65	0.59	0.60	0.51	0.47	0.61
																	11	0.52	0.59	0.47	0.40	0.44	0.56	0.66	0.79	0.58	0.65	1	0.54	0.57	0.65	0.45	0.53
12	0.44	0.41	0.40	0.37	0.40	0.40	0.55	0.53	0.83	0.59	0.54	1	0.50	0.49	0.68	0.54
																	13	0.50	0.51	0.51	0.53	0.55	0.54	0.55	0.49	0.49	0.60	0.57	0.50	1	0.71	0.59	0.77
14	0.53	0.61	0.47	0.42	0.45	0.51	0.49	0.54	0.44	0.51	0.65	0.49	0.71	1	0.58	0.66
																	15	0.46	0.41	0.36	0.41	0.41	0.34	0.42	0.37	0.60	0.47	0.45	0.68	0.59	0.58	1	0.66
16	0.47	0.48	0.52	0.55	0.56	0.52	0.56	0.46	0.50	0.61	0.53	0.54	0.77	0.66	0.66	1

4) Solving the truncation matrix R_λ＝λ_(i，j): the method comprises the following steps that lambda is used as a metric value to represent a truncation matrix coefficient, a fuzzy relation matrix among modules is truncated, elements which are larger than or equal to lambda in a fuzzy relation matrix R are provided with the numerical value of 1, the numerical value which is smaller than the lambda element is provided with 0, the elements with the numerical value of 1 in the same row or column are gathered into the same module, the rest elements independently become the modules, and the more the lambda value is, the more detailed the module division is; different process module clustering division results can be obtained by selecting different lambda values; and sequentially calculating a standardized matrix, a fuzzy similar matrix and a transfer closure matrix according to the original incidence matrix by using a Matlab tool and a fuzzy clustering analysis algorithm. Finally, a dynamic clustering module can be formed by cutting R by using the lambda-cut matrix, as shown in FIG. 3. Large to small threshold partition sequence: {10.84770.80140.78230.77090.74740.72740.71670.71210.68930.68190.66250.6490.64310.60920.5865}, the full process sequence is divided into different process categories. When lambda is>0.6625, the number of divided modules is excessive and deviates from the actual situation. When λ is 0.6625, all process modules can be classified into 4 major categories, which are fusion casting module, hot rolling module, cold rolling module, and heat treatment module. The cleaning module as an auxiliary process module can be divided independently and is basically consistent with the actual situation.

In conclusion, the method for dividing the complex process modules for large-scale customization is compact in logic, accurate in control and efficient, a method for dividing the models on the process level is provided by means of characteristic relations between processes corresponding to workpieces, a fuzzy association matrix between the process modules is constructed, the fuzzy association matrix is converted and solved by using a fuzzy clustering analysis method to obtain a transfer closure matrix, a module clustering diagram is formed according to different partition threshold sequences, different module division schemes are obtained by selecting different lambda values, and therefore personalized finished products adapting to market and technical changes can be produced quickly and efficiently, and high-level requirements of the process are met;

compared with the traditional product family module division method, the method is mainly based on the analysis of mature products of enterprises and processing technologies of products with market demands, combines the process information extraction, the process creative design and the process modularization to construct a modular process design platform, modularizes the serialized product processes, matches the similarity of the modules by combining the module configuration technology and the order requirements, and finally carries out the deformation design of the corresponding degree on the process requirements by the deformation design module to realize the functions of different product requirements.

The principal features, principles and advantages of the invention have been shown and described above. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are described in the specification and illustrated only to explain the principles of the invention, but that various changes and modifications may be made therein without departing from the spirit and scope of the invention as expressed in the following claims. The scope of the invention is defined by the appended claims and equivalents thereof.

Claims (6)

1. A method for dividing complex process modules for large-scale customization is characterized by comprising the following steps: the method comprises the following steps:

s1: establishing module characteristic attributes by a modular mapping mechanism, establishing the relevance among process modules by the attributes, and describing the relevance as replaceable relevance and influence degree relevance;

s2: the correlation degree between the process modules is related to the substitutability, the influence degree and the like, and is represented by a mathematical model of a fuzzy correlation matrix R: r ═ ω_fF+ω_pP, where F, P denotes the defined influence correlation matrix and the alternative correlation matrix, ω, respectively_f、ω_pRepresenting the weight coefficient, ω, to which the characteristic corresponds_f、ω_p∈[0，1]And satisfies the following conditions: omega_f+ω_p＝1；

S3: converting and solving the fuzzy incidence matrix model by using a fuzzy clustering analysis method to obtain a transfer closure matrix;

s31: determining the correlation among all the working procedures and determining the correlation degree among all the process modules;

s32: and (3) data standardization treatment: according to different original formats, expression formats and level quantization modes, translation and range transformation are adopted to carry out data standardization processing, and each data of the fuzzy association matrix is compressed to [0, 1 ];

s33: establishing a fuzzy similarity matrix R': the fuzzy incidence matrix R and the fuzzy similarity matrix R' are in equivalent relation, and a direct Euclidean distance method is applied, wherein R is_(i,j)＝1-c×d(x_i,x_j) Wherein C is a parameter arbitrarily selected so that r is not less than 0_(i,j)≤1，d(x_i,x_j) Denotes x_iAnd x_jThe distance between:

s34: solving a transitive closure matrix

According to the theorem, the fuzzy equivalent matrix is obtained by using the fuzzy incidence matrix R by a quadratic method

Namely transitive closure matrix

And, make

S35: solving the truncation matrix R_λ＝λ_(i,j): wherein, lambda is used as a metric value to represent a truncation matrix coefficient, the fuzzy relation matrix between the modules is truncated, elements which are larger than or equal to lambda in the fuzzy incidence matrix R are taken as 1, the value which is smaller than the lambda element is taken as 0, and the elements with the value of 1 in the same row or column are gathered into the same row or columnThe residual elements become modules independently, and the more the lambda value is, the more detailed the module division is; different process module clustering division results can be obtained by selecting different lambda values;

s36: with the help of Matlab tool and the application of fuzzy clustering analysis method, the standardized matrix, fuzzy similar matrix, transfer closure matrix can be calculated in turn according to the original incidence matrix, and finally the fuzzy incidence matrix R is cut by using lambda-cut matrix to form a module clustering graph:

s4: and forming an integral module clustering graph according to different partition threshold sequences, and obtaining different module division schemes by selecting different lambda values.

2. The method for dividing the complex process module for large-scale customization according to claim 1, wherein: in step S2, the method for establishing the fuzzy association matrix specifically includes the following steps:

a1: and (3) taking the attribute value domain of the output parameter as the correlation for judging each processing procedure, describing the correlation as the correlation of replaceable correlation and influence degree, and expressing a fuzzy correlation matrix R by using a mathematical model: r is_i,j＝ω_ff_i,j+ω_pp_i,jWherein r is_i,jRepresenting the overall correlation coefficient between module i and module j, f_i,jRepresenting the correlation coefficient, p, between block i and block j_i,jDenotes the correlation coefficient, ω, between block j and block i_f、ω_pRepresenting a weight coefficient corresponding to the characteristic;

a2: the fuzzy correlation matrix R is also called an original matrix R, and R satisfies the self-reflexivity: r (i, j) is not less than 0 and not more than 1, and r (i, i) is r (j, j) is 1; and R satisfies symmetry: r (i, j) ═ R (j, i), the mathematical model of the fuzzy correlation matrix R in step a1 is expressed as:

3. the method for dividing the complex process module for large-scale customization according to claim 2, wherein: in the step A1, r_i,jRepresenting the overall correlation coefficient between module i and module j, f_i,jRepresenting the correlation coefficient, p, between block i and block j_i,jDenotes the correlation coefficient, ω, between block j and block i_f、ω_pRepresenting a weight coefficient corresponding to the characteristic; omega_f、ω_p∈[0，1]And satisfies the following conditions: omega_f+ω_p＝1。

4. The method for dividing the complex process module for large-scale customization according to claim 2, wherein: in the step A1, the output parameters include grain size, yield strength, elongation, residual stress, corrosion resistance, electrical conductivity, hardness, and tensile strength.

5. The method for dividing the complex process module for large-scale customization according to claim 1, wherein: in step S32, the normalization formula used in the data normalization process is:

wherein i ═ 1, 2., m, x'_(i,j)Representing the original data; x'_(i,j)man、x'_(i,j)minRespectively representing the maximum value and the minimum value of the original data; x is the number of_(i,j)The data obtained after normalization.

6. The method for dividing the complex process module for large-scale customization according to claim 1, wherein: in the step S34, a transitive closure matrix is obtained

The method specifically comprises the following steps:

a1: set up R₀Is R, wherein R₀Representing the original incidence matrix of the module;

a2: let R be_iKnowing the result, start to compare R_iAnd

the corresponding element; if it is not

Then R is_iIs that

The transitive closure matrix of (2), at which point computation may be stopped; if not satisfied

Continuing to execute the next step;

a3: by algorithmic calculation

And setting the result to R_i+1Executing the step A2;

a4: repeating the above steps until finally solving the transfer closure matrix to ensure that

Images