CN105678418A - Product manufacture-oriented combined optimization method - Google Patents

Product manufacture-oriented combined optimization method Download PDF

Info

Publication number
CN105678418A
CN105678418A CN201610009554.7A CN201610009554A CN105678418A CN 105678418 A CN105678418 A CN 105678418A CN 201610009554 A CN201610009554 A CN 201610009554A CN 105678418 A CN105678418 A CN 105678418A
Authority
CN
China
Prior art keywords
product
chaos
product mix
scheme
personalized
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201610009554.7A
Other languages
Chinese (zh)
Inventor
徐小龙
戎汉中
李涛
李荣志
陈帅霖
吴晓华
张曼
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nupt Institute Of Big Data Research At Yancheng Co Ltd
Nanjing Post and Telecommunication University
Nanjing University of Posts and Telecommunications
Original Assignee
Nupt Institute Of Big Data Research At Yancheng Co Ltd
Nanjing Post and Telecommunication University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nupt Institute Of Big Data Research At Yancheng Co Ltd, Nanjing Post and Telecommunication University filed Critical Nupt Institute Of Big Data Research At Yancheng Co Ltd
Priority to CN201610009554.7A priority Critical patent/CN105678418A/en
Publication of CN105678418A publication Critical patent/CN105678418A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/04—Manufacturing
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Economics (AREA)
  • Strategic Management (AREA)
  • Theoretical Computer Science (AREA)
  • Tourism & Hospitality (AREA)
  • General Physics & Mathematics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Game Theory and Decision Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Development Economics (AREA)
  • Manufacturing & Machinery (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Primary Health Care (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The present invention discloses a product manufacture-oriented combined optimization method. The combined optimization method enables the quality-cost control function in a product manufacturing environment. On the premise that the manufacturing cost is controlled not to exceed a target cost, the product quality is ensured to be optimal. The product diversity is further improved. According to the technical scheme of the invention, the chaotic initialization and chaotic upset strategy is adopted, and the advantages and disadvantages of a product combination scheme are evaluated based on a personalized fitness function. Combined items are classified, so that the number of combination is reduced. In this way, the complete enumeration of all product combination schemes is enabled, so that the search efficiency is improved. Meanwhile, a personalized interface is provided, and a personalized combination scheme of optimal cost performance is generated according to the personalized requirements of users. During the searching process of the optimal product combination scheme, the chaotic search is conducted and all possible optimum product combination schemes are fully traversed. Therefore, the diversity of product combination schemes is ensured.

Description

The combined optimization method that a kind of used for products manufactures
Technical field
The present invention relates to the combined optimization method that a kind of used for products manufactures, belong to computer and manufacturing industry cross-application technical field.
Background technology
When enterprise carries out the market competition of fierceness, product cost number with the existence of enterprise and develop closely related. Reducing cost and be always up the target that enterprise lays siege to, the key effectively reducing Chinese Enterprises product cost is just by manufacturing cost control. Along with the development speed of new technique is beyond the imagination, the demand of people is increasingly personalized, and enterprise's facing challenges becomes increasingly complex. In recent years, manufacturing industry is faced with product price and constantly reduces, and prescription improves constantly, and date of delivery constantly shortens, the situation that kind is personalized, variation requirement is gradually increased. Therefore, market initially enters one with personalized customization to meet the age of consumer demand.
For small lot, order form, discreteness manufacturing enterprise, the problem being frequently encountered price fixing, because often there being client that product is proposed personalized requirement. So, the administration section of enterprise must according to the process data of product, actual cost, recent raw-material price, rival quotation, affect the factor direct material to product of price in conjunction with gathering mode, the tax rate, the exchange rate etc., direct labor, manufacturing expense make correct appraisal, i.e. the manufacturing cost of product. Then the appraisal according to product, is estimated controlling to the manufacturing cost of the personalized product that user requires, and when controlling manufacturing cost less than objective cost, it is necessary to ensure the optimal quality of product.
In product manufacturing, the apolegamy problem of part in a product, this is typical combinatorial optimization problem, and namely " quality-cost " in product manufacturing determines problem. Solving this problem not only consuming time, and be difficult to find best product assembled scheme, the result solved directly affects quality and the cost of product. For this problem, the quality of each part is given different weights by existing method, when manufacturing cost is less than objective cost, uses intelligent optimization algorithm that each part is realized optimized choice.So to a certain extent product mix is optimized, but still has the disadvantage that
1) in solution procedure, adopt random searching strategy, not there is good ability of searching optimum, cause that assembled scheme is precocious, it is impossible to successfully manage the diversification of varieties requirement that product manufacturing faces;
2) personalisation interface it is not provided or is not based on multi-objective restriction, all can not effectively solve discrete, multiobject, personalized combinatorial optimization problem, cannot effectively solve the difficulty that manufacturing industry faces now yet.
The situation that the requirements such as manufacturing industry is faced with product price and constantly reduces, and prescription improves constantly, and kind is personalized, diversified are gradually increased, market initially enters one with personalized customization to meet the age of consumer demand. Existing combined optimization method adopts random searching strategy in solution procedure, does not have good ability of searching optimum, causes that assembled scheme is precocious, it is impossible to successfully manage the diversification of varieties requirement that product manufacturing faces; Combinatorial optimization problem is typical NP difficult problem problem, and time complexity is the problem of polynomial time, along with the expansion of problem scale, can cause multiple shot array, so the efficiency of combined optimization method is the key factor needing to consider; Existing combined optimization method there is also a common problem, it is exactly personalisation interface is not provided or is not based on multi-objective restriction, all can not effectively solve discrete, multiobject, personalized combinatorial optimization problem, also cannot effectively solve the difficulty that manufacturing industry faces now. And the present invention can solve problem above well.
Summary of the invention
Present invention aim at solving above-mentioned the deficiencies in the prior art, propose the combined optimization method that a kind of used for products manufactures, the method is under used for products manufacturing environment, realize the combined optimization method that " quality-cost " controls, under controlling the manufacturing cost premise less than objective cost, ensure that the quality of product reaches optimum, and further increase the multiformity of personalized product, and improve the efficiency of product mix.
The technical solution adopted for the present invention to solve the technical problems is: the combined optimization method that a kind of used for products manufactures, the method adopts chaos intialization and chaos to upset, initialization and renewal to product mix scheme improve, and adopt the fitness function with personalization to evaluate the quality of product mix scheme, thus find the personalized product assembled scheme of optimum.
Method flow:
Step 1: randomly generate m Chaos Variable, k in chaotic space [0,1] based on Logic mapping equationi(i=0,1 ..., m-1), wherein m represents the sum of Chaos Variable, and i represents sequence subscript, kiRepresent i-th Chaos Variable.
Step 2: m Chaos Variable is mapped on m kind part by chaos intialization, the part of a corresponding product of Chaos Variable, and a kind of part only selects an alternate means, thus obtain one and initialize product mix scheme.
Step 3: after repetition step 1, step 2 carry out n time, obtains n and initializes product mix scheme, and wherein n represents the sum of product mix scheme.
Step 4: the fitness function according to having personalization evaluates this n product mix scheme, finds the optimum product mix scheme in this n product mix scheme, and it is theoretical optimum to judge whether the fitness function value of correspondence reaches. If it is, this product mix scheme is global optimum's product mix scheme; Otherwise, step 5 is performed.When the optimum product mix scheme or iterations that find theory reach the upper limit, then stop, optimum product mix scheme when output stops, being final global optimum's product mix scheme.
Step 5: upset by chaos and update n product mix scheme so that this n product mix scheme learns to history optimum product mix scheme own and global optimum's product mix scheme. After renewal completes, return and perform step 4.
In above-mentioned steps 1, the situation of Logic mapping equation is: Logic mapping equation can produce one group of random sequence with ergodic and pseudo-randomness by chaos iteration, and wherein Logic mapping equation formula is as shown in Equation 1.
Z:αn+1=μ αn(1-αn) formula 1
In formula 1, Z is Chaos Variable, and μ controls parameter (when μ=4, Z is in Complete Chaos state), αnIt is a value of Chaos Variable, when composing initial value to Chaos Variable, it is necessary to meet α0≠ 0,1/4,1/2,3/4,1. One initial value is by Logic iteration, it is possible to produce one sequence Z: α0,α1,...,αm.... This sequence has ergodic in chaotic space, and each iteration can produce different values, and chaotic space is generally [0,1]. By continuous iteration, it is achieved the traversal search to chaotic space.
In above-mentioned steps 2, the situation of chaos intialization is: such as Chaos Variable kiIt is mapped on the part of product, chaotic space [0,1] is divided into niSub spaces, and successively called after 0 space, 1 space ..., ni-1 space, wherein niRepresent that i-th part has niIndividual alternate means. Judge Chaos Variable kiBelong to which chaos subspace, it is assumed that be μ, then it represents that the μ alternate means in i-th kind of part is selected, and i-th kind of part has initialized. After m Chaos Variable has mapped, a product mix scheme is generated as.
The situation of the fitness function in above-mentioned steps 4 with personalization is: in actual life, product mix optimization problem is generally all multiple constraint problem, and constraint varies with each individual. With automobile as an example, some people lies in the performance of car, and some people payes attention to the outward appearance of car, so the individual demand utilizing user in product mix optimization is necessary. The present invention adds personalized constraint in fitness function, is described in detail below:
First, by the institute's Constrained classification in product mix optimization problem, it is divided into general restriction and personalized constraint, the constraint wherein with personalization features is categorized as the personalized constraint (price such as user's expected product, product is liked degree etc. by user), other constraints are categorized as general restriction (objective cost such as product), and which is divided into general restriction or which personalized constraint to see specific product, referring in detail to Examples below; Secondly, consider that the attention degree of each constraint is likely to different (such as, some people compares the price lying in product, and some people focuses on the quality of product) by different people, so giving every one weight of personalized constraint, it is used for representing the attention degree to this personalization constraint. Personalization constraint is used for designing fitness function by the present invention, and general restriction is as the constraints of fitness function. For the convenient good and bad degree evaluating product, each personalized constraint is converted into score model, namely more high (price of such as product is equal to the price of user's expected product for the more high then mark of personalization level, then this personalization is retrained to be divided into 100 points, if below or above desired price, score is successively decreased), the score of personalized restricted model takes its weighted average and divides, and is the value of fitness function.
Assuming certain product mix optimization problem Prescribed Properties A, B, C and D, wherein A, B and C are personalized constraints, and D is general restriction, then fitness function is as shown in Equation 2:
Formula 2
Wherein S (A) is the score of personalized constraint A, and in like manner S (B) and S (C) is personalized constraint B and the score of personalized constraint C, and α, β and γ are the weight of each personalized constraint, and alpha+beta+γ=1.
The situation that in above-mentioned steps 5, chaos upsets is: namely product mix scheme is fully upset, and it upsets rule as shown in formula 3 and formula 4.
Formula 3
Formula 4
In formula 3, represent and if only if vi,j(t)==vi,j p==vi,j gTime, vi,jRemain unchanged after renewal, otherwise make vi,j(t+1)=-1, wherein t+1 represents the state after renewal, and t represents current state,pRepresent current optimum state,gRepresent overall situation optimum state.
In formula 4, i represents i-th product mix scheme, and j represents the jth dimension variable in assembled scheme. C (xi,j(t)) it is upset function, its concrete steps are divided into three steps: first, calculate variable x according to ji,j(t) which part corresponding, for instance, if j >=0 and j < n0, then it represents that xi,j(t) corresponding 0th part (from the 0th part); OrAnd(h > 0), then it represents that xi,jT () corresponding the h part, the Chaos Variable associated with the h part is ki,h. Secondly, Logic mapping equation is used, by Chaos Variable ki,hIteration once, obtains new Chaos Variable value ki,h(t+1). Finally, it is judged that new Chaos Variable value ki,h(t+1) belong to which space (see step 2) in chaos subspace, it is assumed that for pth sub-space, then make xi,p=1, and to make the variable that in this part, miscellaneous part is corresponding be 0 entirely. J (xi,j p) and J (xi,j g) for being simple conditional plan, it is divided into three kinds of situations (with J (xi,j p) illustrate, J (xi,j g) conditional plan and J (xi,j p) consistent): if the first xi,j p=1, then xi,j(t+1)=1, and to make the variable that in this part, miscellaneous part is corresponding be 0 entirely. If the second xi,j p=0 and xi,j(t)=0, then xi,j(t+1) remain unchanged. If the 3rd xi,j p=0 and xi,j(t)=1, then xi,j(t+1)=C (xi,j(t))。
Beneficial effect:
1, combinatorial optimization problem is typical NP difficult problem problem, and time complexity is the problem of polynomial time, along with the expansion of problem scale, can cause multiple shot array, and it is unpractical for traveling through all of assembled scheme completely. And the present invention is by the classification of the items by combination, making number of combinations reduce, enumerating all of assembled scheme so completely is possibly realized so that search efficiency improves.
2, existing combined optimization method is in search Optimum combinational scheme process, adopts random search and randomly updates strategy, it is impossible to ensure the multiformity of assembled scheme. And the present invention adopts Chaos Search in search procedure, fully travel through all possible best product assembled scheme, it is ensured that the multiformity of product mix scheme.
3, in order to adapt to the demand of the personalized customization in market, the present invention provides personalisation interface, it is possible to the individual requirement according to user, produces the personalized assembled scheme that cost performance is optimum.
Accompanying drawing explanation
Fig. 1 is product mix scheme optimizing simulation drawing.
Fig. 2 is for being absorbed in local optimum schematic diagram.
Fig. 3 (a), Fig. 3 (b) are for upsetting product mix scheme schematic diagram.
Fig. 4 (a), Fig. 4 (b) are each personalized constraint score model figure.
Fig. 5 is the method flow diagram of the present invention.
Detailed description of the invention
Below in conjunction with Figure of description, the invention is described in further detail.
The present invention is under product manufacturing environment, it is achieved that the combined optimization method that " quality-cost " controls.The present invention is upset by chaos intialization and chaos, improvement is made in the initialization of product mix scheme and the renewal of product mix scheme, and adopt the fitness function with personalization to evaluate the quality of product mix scheme, thus find the personalized product assembled scheme of optimum. The present invention not only takes full advantage of the features such as the pseudo-randomness of chaos searching method, ergodic and regularity, and can produce the product mix scheme of personalization, it is ensured that the multiformity of product mix scheme, more improves search efficiency. Below in conjunction with Figure of description, the invention is described in further detail.
As it is shown in figure 1, at X0,X1,...,Xi,...,Xn-1In, each vector all represents a kind of product mix scheme, all product mix schemes is compared, selects the product mix scheme of global optimum. Product mix scheme XiThe optimum product mix scheme of distance is nearest, so setting XiFor current global optimum product mix scheme, and set this global optimum's product mix scheme as Xg. Each product mix scheme, in searching process, all can retain oneself history optimum product mix scheme and Xi p. By constantly updating so that current production assembled scheme learns to history optimum product mix scheme and global optimum's product mix scheme, more shown in new formula such as formula 5 and formula 6.
Vi(t+1)=w Vi(t)+c1·r1·(Xi p-Xi)+c2·r2·(Xg-Xi) formula 5
Xi(t+1)=Xi+Vi(t+1) formula 6
Above-mentioned optimizing model is it is considered that the situation of only one of which global optimum product mix scheme, and optimization problem under normal circumstances is the optimization problem of multi-peak, namely has multiple extreme value and multiple optimal value. So traditional combinatorial optimization algorithm is when solving the optimization problem of multi-peak, due to fast convergence rate, in calculating process, it is easily trapped into local optimum, causes that colony is precocious. It is absorbed in local optimum situation as in figure 2 it is shown, scheme Zhong Youliangge global optimum product mix scheme, two local optimum product mix schemes. Current production assembled scheme XiDistance local optimum product mix option A is nearest, and then all of product mix scheme is all to XiClose, it is absorbed in local optimum product mix scheme.
The present invention adopts chaos to upset, and the renewal of product mix scheme is improved, and namely chaos upsets the position and the trend that make product mix scheme fully upset product mix scheme in searching process, thus greatly reducing the probability being absorbed in local optimum; As it is shown on figure 3, wherein figure (a) is the state of product mix scheme before updating, namely all product mix schemes are close to local optimum product mix option A; Figure (b) is the state after product mix scheme updates, and has fully upset position and the trend of product mix scheme, destroy product mix scheme and be absorbed in local optimum after renewal. For the optimization problem for multi-peak, the present invention may search for global optimum as much as possible product mix scheme, it is ensured that the multiformity of assembled scheme; And present invention also offers the interface of personalization, it is possible to produce to meet the product mix scheme of user individual requirement.
In order to conveniently understand technical scheme, concepts more defined below, including:
Define 1 chaos searching method. Namely it is by determining that equation obtains the random motion state point with features such as pseudo-randomness, ergodic and regularity.
In the present invention adopt be Logic mapping equation as chaos searching method, its formula as shown in Equation 7:
Z:αn+1=μ αn(1-αn) formula 7
In formula 7, Z is Chaos Variable, and μ controls parameter (when μ=4, Z is in Complete Chaos state), αnIt is a value of Chaos Variable, αn+1Be iteration once after value, when to Chaos Variable compose initial value time, it is necessary to meet α0≠ 0,1/4,1/2,3/4,1.One initial value is by Logic iteration, it is possible to produce one sequence Z: α0,α1,...,αi.... This sequence has ergodic in chaotic space, and each iteration can produce different values, and chaotic space is generally [0,1]. By continuous iteration, it is achieved the traversal search to chaotic space.
Define 2 product mix optimization. By scattered part one complete product of composition, and this product is under the constraint that meets some requirements so that final products are worth big as far as possible.
The present invention towards " quality-cost " of product manufacturing to determine that problem describes as follows:
(1) direct material a: product is made up of m part; Each part is made up of multiple alternate means; Each alternate means all has the weight representing its quality and cost.
(2) direct labor a: production and the cost of labor L that occurs.
(3) manufacturing expense: contribute to the formation of product in a process of producing product and the manufacturing expense M that occurs.
The mathematical model of its correspondence describes as shown in Equation 8:
M a x &Sigma; i = 0 m - 1 &Sigma; j = 0 n i - 1 c i , j x i , j
s . t &Sigma; i = 0 m - 1 &Sigma; j = 0 n i - 1 w i , j x i , j + L + M &le; W Formula 8
&Sigma; j = 0 n i - 1 x i , j = 1 , &ForAll; i &Element; ( 0 , 1 , ... , m )
X in formula 8i,j∈{0,1},J, and xi,j=1 represents that the jth alternate means in the i-th class part is selected, otherwise represents not selected. Wherein i represents the subscript of i-th part, and namely (from the 0th part), j is the subscript of corresponding alternate means;Represent that in each part, only one of which alternate means is selected; M represents the quantity of part, niRepresent the quantity of alternate means in i-th part; ci,jRepresent the quality weight of the i-th apoplexy due to endogenous wind jth project; wi,jRepresenting the cost of the i-th apoplexy due to endogenous wind jth project, W refers to objective cost. Only give an objective cost constraint in the model, in real product manufactures, have multiple constraint, specifically see the description in Examples below.
Define 3 chaos intialization. Namely the Chaos Variable in Logic mapping equation takes m sub-value at random, each value correspond to a kind of part, and each value may map in the alternate means of corresponding part, a kind of part only selects an alternate means, thus obtains one and initialize product mix scheme.
In order to unified with the product mix optimization mathematical model in definition 2, parameter declaration required in chaos intialization is as follows:
One product is made up of m part, has n in each partiIndividual alternate means, each alternate means represents with 0 and 1 in initialization, and 0 represents and do not choose these parts, and 1 represents and chooses this parts, so making the vector that whole alternate means of each part form beThe order of then product mix scheme is Xi=(B0,B1,...,Bm-1), its dimension is &Sigma; i = 0 m - 1 n i .
Define 4 fitness function values. The i.e. value of fitness function, this value is quantizating index, is used for evaluating the good and bad degree of product mix scheme.
The combined optimization method that the present invention proposes is suitable in product manufacturing. " quality-cost " of product manufacturing determines that the quality in problem includes the inherent quality of product and the presentation quality of product. Wherein inherent quality mainly includes the aspects such as performance, life-span, reliability, safety. And product appearance quality mainly includes the fineness of product, moulding, color and luster, packaging etc. (such as the moulding of bicycle, color light cleanliness etc.). One product is formed (such as by multiple parts, one automobile is made up of numerous parts, including electromotor, chassis, vehicle body, tire, change speed gear box, electric equipment, electronic equipment etc.), and each part quality has quality, price has height, so each part is formed (such as by the alternate means of substantial amounts of different quality, different prices, electromotor has Diesel engine, petrol engine, electric automobile motor and hybrid power etc., and its price of different electromotors is also different).
In actual life, product mix optimization problem is generally all multiple constraint problem, and constraint varies with each individual. With automobile as an example, the color of the car that everyone likes, vehicle, variator etc. are likely to each variant, so the individual demand utilizing user in product mix optimization is necessary. The present invention adds personalized constraint in fitness function, is described in detail below:
First, by the institute's Constrained classification in product mix optimization problem, it is divided into general restriction and personalized constraint, the constraint wherein with personalization features is categorized as the personalized constraint (price such as user's expected product, product is liked degree etc. by user), other constraints are categorized as general restriction (objective cost such as product), and which is divided into general restriction or which personalized constraint to see specific product, referring in detail to Examples below; Secondly, consider that the attention degree of each constraint is likely to different (such as, some people compares the price lying in product, and some people focuses on the quality of product) by different people, so giving every one weight of personalized constraint, it is used for representing the attention degree to this personalization constraint. Personalization constraint is used for designing fitness function by the present invention, and general restriction is as the constraints of fitness function. For the convenient good and bad degree evaluating product, each personalized constraint is converted into score model, namely more high (price of such as product is equal to the price of user's expected product for the more high then mark of personalization level, then this personalization is retrained to be divided into 100 points, if below or above desired price, score is successively decreased), the score of personalized restricted model takes its weighted average and divides, and is the value of fitness function.
Assuming certain product mix optimization problem Prescribed Properties A, B, C and D, wherein A, B and C are personalized constraints, and D is general restriction, then fitness function is as shown in Equation 9:
Formula 9
Wherein S (A) is the score of personalized constraint A, and in like manner S (B) and S (C) is personalized constraint B and the score of personalized constraint C, and α, β and γ are the weight of each personalized constraint, and alpha+beta+γ=1.
With product manufacturing for background, when controlling manufacturing cost less than objective cost, it is necessary to ensure the optimal quality of product. Here quality includes the economy of the inherent quality of product, the presentation quality of product and product. So the personalization that the present invention takes is constrained to three, first, the price of user's expected product; The second, the satisfaction of product inherent quality; 3rd, the satisfaction of the presentation quality of product. General restriction is the objective cost (representing with letter O below) of product. The price of user's expected product is the rational average price drawn by market survey, for producer, if price is higher than the price of user's expected product, then sales volume may reduce; If price is lower than the price of user's expected product, then sales volume may improve, but the profit of every product reduces, and this is small profits and quick turnover strategies. The present invention is left out the price impact on sales volume, it is assumed that price is exactly equal to the price of user's expected product, and now producer's acquisition profit is the highest. Price is higher or lower than the price of user's expected product, then the profit that producer obtains can reduce. So the price score model of user's expected product is as shown in Equation 10, the score model figure of its correspondence is such as shown in Fig. 4 (a).
S c = R / S &times; 100 , i f ( R < S ) ( 1 - ( R - S ) / S ) &times; 100 , i f ( R &GreaterEqual; S ) Formula 10
Wherein R indicates that price, S are the prices of user's expected product, ScIt is the score of the price constraints of user's expected product.
The inherent quality of product and the external quality constraint of product are quality constraint, and the quality of quality directly affects sales volume, and for the identical product that price is suitable, the measured sales volume of matter is good. So by cost control in certain objective cost, it is necessary for improving product quality. Its required part of different products is different, but what measured product of matter was made up of the part of Functionality, quality and appealing design. " quality-cost " in definition 2 determines that problem is known in describing, one product is made up of m part, each part is made up of multiple alternate means, each alternate means all has the weight representing its quality and cost, so the quality weight sum of all parts of one product of composition can be converted into score model, the more big expression quality of quality weight sum is more good, then corresponding score is more high. The mode that the inherent quality of product changes into score model with the external quality of product is consistent, so the inherent quality score model of product is detailed below, as shown in Equation 11, the score model figure of its correspondence is such as shown in Fig. 4 (b).
Sq=R/S × 100 formula 11
Wherein R represent the parts affecting inherent quality in product mix scheme quality weight and,(wherein i ∈ B is the part affecting inherent quality), S represent the parts affecting inherent quality in product mix scheme biggest quality weight and, namely S = M a x &Sigma; i &Element; B &Sigma; j = 0 n i - 1 c i , j x i , j .
In like manner can obtaining presentation quality score model, as shown in Equation 12, the score model figure of its correspondence is such as shown in Fig. 4 (b).
Sp=R/S × 100 formula 12
Can be obtained from above, fitness function as shown in Equation 13:
F = &alpha; &times; S c + &beta; &times; S q + &gamma; &times; S p ( C &le; O ) F = - 1 ( C > O ) Formula 13
Wherein C represents actual fabrication cost, namelyWherein O represents objective cost.
Define 5 chaos to upset. I.e. product mix scheme introduces chaos searching method at no point in the update process, fully upsets product mix scheme, as far as possible traversal search space.
If the X in formula 5i p-Xi=Vi pAnd Xg-Xi=Vi g, then speed more new formula is as shown in Equation 14.
Vi(t+1)=w Vi(t)+c1·r1·Vi p+c2·r2·Vi gFormula 14
Wherein Vi p=(vi,0 p,vi,1 p,...,vi,j p... .), Vi g=(vi,0 g,vi,1 g,...,vi,j g... .), and vi,j pAnd vi,j gShown in corresponding rule such as formula 15 and formula 16.
v i , j p = r a n d o m ( 1 ) , i f ( x i , j p = = x i , j ) 0 , i f ( x i , j p &NotEqual; x i , j ) Formula 15
v i , j p = r a n d o m ( 1 ) , i f ( x j g = x i , j ) 0 , i f ( x j g &NotEqual; x i , j ) Formula 16
Random (1) in formula 15 and formula 16 represents the random number randomly generating 0 or 1, i.e. current production assembled scheme XiJth dimension variate-value equal to the history best product assembled scheme X of the programi pJth dimension variate-value, then vi,j pThe random number that value is 0 or 1, be otherwise 0.
The present invention is for " hardware product combination ", and its " quality-cost " determines that the quality in problem includes the inherent quality of product and the presentation quality of product; Cost includes direct material, direct labor, manufacturing expense. The product mix scheme flow sheet of the present invention is as shown in Figure 5. Its concrete operation step is as follows:
Step 1: randomly generate m Chaos Variable, k in chaotic space [0,1] based on Logic mapping equationi(i=0,1 ..., m-1). It is described in detail below:
Step 1-1: the random array k [] between using random function Math.random () to generate 0～1, array length is m.
Step 1-2: judge whether each numerical value in random array is in 0,0.25,0.5,0.75 and 1, if it has, then use random function to regenerate a random number be assigned to the numerical value of correspondence, until not having these five to be worth in array.
Step 2: m Chaos Variable is mapped on m kind part by chaos intialization, the part of a corresponding product of Chaos Variable, and a kind of part only selects an alternate means, thus obtain one and initialize product mix scheme.It specifically comprises the following steps that
Step 2-1: chaotic space [0,1] is divided into niSub spaces, and successively called after 0 space, 1 space ..., ni-1 space, wherein niRepresent that i-th part is by niIndividual alternate means.
Step 2-2: judge which chaos subspace Chaos Variable k [i] belongs to, it is assumed that be μ, then it represents that the μ alternate means in i-th kind of part is selected, and i-th kind of part has initialized, i.e. Bi=(0,0 ..., 1 ..., 0...), value corresponding for subscript μ is 1, and other are 0.
Step 2-3: traversal Chaos Variable array, both maps in the part of correspondence by each Chaos Variable, then obtain one and initialize product mix scheme, i.e. Xi=(B0,B1,...,Bm-1)。
Step 3: repeat step 1, step 2 n time altogether, obtains n and initializes product mix scheme. In this flow process, n represents the sum of product mix scheme. It specifically comprises the following steps that
Step 3-1: n step 1 of iteration, namely generates the Chaos Variable matrix of a n × m, as shown in Equation 17.
Formula 17
Step 3-2: n step 2 of iteration, namely generates n and initializes product mix scheme, as shown in Equation 18.
Formula 18
Step 4: the fitness function according to having personalization evaluates this n product mix scheme, finds the optimum product mix scheme in this n product mix scheme, and it is theoretical optimum to judge whether the fitness function value of correspondence reaches. If it is, this product mix scheme is global optimum's product mix scheme; Otherwise, step 5 is performed. When finding final optimum product mix scheme or iterations to reach the upper limit, then stop, optimum product mix scheme when output stops, being global optimum's product mix scheme. It is described in detail below:
Step 4-1: travel through n product mix scheme, calculates the fitness function value of each product mix scheme respectively, deposits in array F [n].
Step 4-2: traversal array F [n], is assigned to global optimum appropriateness value F_Best by maximum, and the subscript that maximum is corresponding is assigned to index.
Step 4-3: judge whether F [index] reaches theoretially optimum value F_Theory (this value represents under corresponding objective cost, the accessible theoretical optimum appropriateness value of product), if reaching theoretical optimum, then Xindex=(B0 (index),B1 (index),...,Bm-1 (index)) for best product assembled scheme. Otherwise, Count++ (Count value represents iteration how many times, and initial value is 0), and judge whether Count reaches maximum iteration time Iteration, if it is, output Xindex, terminate search; Otherwise perform step 5.
Step 5: upset by chaos and update n product mix scheme so that this n product mix scheme learns to history optimum product mix scheme own and global optimum's product mix scheme. After renewal completes, return and perform step 4. It specifically comprises the following steps that
Step 5-1: the trend of upgrading products assembled scheme, the more new regulation of its correspondence is as shown in Equation 19.
Formula 19
Formula 19 represents and if only if vi,j(t)==vi,j p==vi,j gTime, vi,jRemain unchanged after renewal, otherwise make vi,j(t+1)=-1, wherein t+1 represents the state after renewal, and t represents current state,pRepresent current optimum state,gRepresent overall situation optimum state.
Step 5-2: upgrading products assembled scheme, the more new regulation of its correspondence is as shown in Equation 20.
Formula 20
In formula 20, i represents i-th product mix scheme, and j represents the jth dimension variable in assembled scheme. C (xi,j(t)) it is upset function, its concrete steps are divided into three steps: first, calculate variable x according to ji,j(t) which part corresponding, for instance, if j >=0 and j < n0, then it represents that xi,j(t) corresponding 0th part (from the 0th part);OrAnd(h > 0), then it represents that xi,jT () corresponding the h part, the Chaos Variable associated with the h part is ki,h. Secondly, Logic mapping equation is used, by Chaos Variable ki,hIteration once, obtains new Chaos Variable value ki,h(t+1). Finally, it is judged that new Chaos Variable value ki,h(t+1) belong to which space (referring to step 2) in chaos subspace, it is assumed that for pth sub-space, then make xi,p=1, and to make the variable that in this part, miscellaneous part is corresponding be 0 entirely. J (xi,j p) and J (xi,j g) for being simple conditional plan, it is divided into three kinds of situations (with J (xi,j p) illustrate, J (xi,j g) conditional plan and J (xi,j p) consistent): if the first xi,j p=1, then xi,j(t+1)=1, and to make the variable that in this part, miscellaneous part is corresponding be 0 entirely. If the second xi,j p=0 and xi,j(t)=0, then xi,j(t+1) remain unchanged. If the 3rd xi,j p=0 and xi,j(t)=1, then xi,j(t+1)=C (xi,j(t))。

Claims (7)

1. the combined optimization method that a used for products manufactures, it is characterised in that described method comprises the steps:
Step 1: randomly generate m Chaos Variable, k in chaotic space [0,1] based on Logic mapping equationi(i=0,1 ..., m-1), wherein m represents the sum of Chaos Variable, and i represents subscript sequence number, kiRepresent i-th Chaos Variable;
Step 2: m Chaos Variable is mapped on m kind part by chaos intialization, the part of a corresponding product of Chaos Variable, and a kind of part only selects an alternate means, thus obtain one and initialize product mix scheme;
Step 3: repeat step 1, step 2 carries out n time, obtains n and initializes product mix scheme, and wherein n represents the sum of product mix scheme;
Step 4: the fitness function according to having personalization evaluates this n product mix scheme, finds the optimum product mix scheme in this n product mix scheme, and it is theoretical optimum to judge whether the fitness function value of correspondence reaches; If it is, this product mix scheme is global optimum's product mix scheme; Otherwise, step 5 is performed; When finding final optimum product mix scheme or iterations to reach the upper limit, then stop, optimum product mix scheme when output stops, being global optimum's product mix scheme;
Step 5: upset by chaos and update n product mix scheme so that this n product mix scheme learns to history optimum product mix scheme own and global optimum's product mix scheme; After renewal completes, return and perform step 4.
2. the combined optimization method that a kind of used for products according to claim 1 manufactures, it is characterized in that: described method adopts chaos intialization and chaos to upset, improvement is made in initialization and renewal to product mix scheme, and adopts the fitness function with personalization to evaluate the quality of product mix scheme.
3. the combined optimization method that a kind of used for products according to claim 1 manufactures, it is characterized in that: in described step 1, the situation of Logic mapping equation is: Logic mapping equation can produce one group of random sequence with ergodic and pseudo-randomness by chaos iteration, wherein Logic mapping equation formula is as shown in Equation 1, it may be assumed that
Z:αn+1=μ αn(1-αn) formula 1
In formula 1, Z is Chaos Variable, and μ controls parameter (when μ=4, Z is in Complete Chaos state), αnIt is a value of Chaos Variable, when composing initial value to Chaos Variable, it is necessary to meet α0≠ 0,1/4,1/2,3/4,1, an initial value is by Logic iteration, it is possible to produce one sequence Z: α0,α1,...,αm..., this sequence has ergodic in chaotic space, and each iteration can produce different values, and chaotic space is generally [0,1];By continuous iteration, it is achieved the traversal search to chaotic space.
4. the combined optimization method that a kind of used for products according to claim 1 manufactures, it is characterised in that: in described step 2, the situation of chaos intialization is: Chaos Variable kiIt is mapped on the part of product, chaotic space [0,1] is divided into niSub spaces, niRepresent that i-th part has niIndividual alternate means, and successively called after 0 space, 1 space ..., ni-1 space; Judge Chaos Variable kiBelong to which chaos subspace, it is assumed that be μ, then it represents that the μ alternate means in i-th kind of part is selected, and i-th kind of part has initialized; After m Chaos Variable has mapped, a product mix scheme is generated as.
5. the combined optimization method that a kind of used for products according to claim 1 manufactures, it is characterized in that: the situation of the fitness function in described step 4 with personalization includes: first, by the institute's Constrained classification in product mix optimization problem, it is divided into general restriction and personalized constraint, the constraint wherein with personalization features is categorized as personalized constraint, other constraints are categorized as general restriction, which is divided into general restriction or which personalized constraint to see specific product, referring in detail to Examples below; Secondly, it is contemplated that the attention degree of each constraint is likely to difference by different people, so giving every one weight of personalized constraint, it is used for representing the attention degree to this personalization constraint.
6. the combined optimization method that a kind of used for products according to claim 5 manufactures, it is characterised in that: described method is to retrain personalization for designing fitness function, and general restriction is as the constraints of fitness function; For the convenient good and bad degree evaluating product, each personalized constraint being converted into score model, namely the more high then mark of personalization level is more high, and the score of personalized restricted model takes its weighted average and divides, and is the value of fitness function;
Assuming certain product mix optimization problem Prescribed Properties A, B, C and D, wherein A, B and C are personalized constraints, and D is general restriction, then fitness function is as shown in Equation 2:
Formula 2
Wherein S (A) is the score of personalized constraint A, and in like manner S (B) and S (C) is personalized constraint B and the score of personalized constraint C, and α, β and γ are the weight of each personalized constraint, and alpha+beta+γ=1.
7. the combined optimization method that a kind of used for products according to claim 1 manufactures, in described step 5, chaos upset situation is: the fully trend of upgrading products scheme, and product mix scheme, shown in its correspondence more new regulation such as formula 3 and formula 4;
Formula 3
Formula 4
Formula 3 represents and if only if vi,j(t)==vi,j p==vi,j gTime, vi,jRemain unchanged after renewal, otherwise make vi,j(t+1)=-1, wherein t represent current state, t+1 represent renewal after state, p represents current optimum state, g represent the overall situation optimum state, lower with;
In formula 4, i represents i-th product mix scheme, and j represents the jth dimension variable in assembled scheme; C (xi,j(t)) it is upset function, its concrete steps are divided into three steps: first, calculate variable x according to ji,j(t) which part corresponding, if j >=0 and j < n0, then it represents that xi,jT () corresponding 0th part, namely (from the 0th part); OrAnd(h > 0), then it represents that xi,jT () corresponding the h part, the Chaos Variable associated with the h part is ki,h, wherein niRepresent that i-th part has niIndividual alternate means; Secondly, Logic mapping equation is used, by Chaos Variable ki,hIteration once, obtains new Chaos Variable value ki,h(t+1); Finally, it is judged that new Chaos Variable value ki,h(t+1) belong to which space in chaos subspace, it is assumed that for pth sub-space, then make xi,p=1, and to make the variable that in this part, miscellaneous part is corresponding be 0 entirely; J (xi,j p) and J (xi,j g) for being simple conditional plan, with J (xi,j p) illustrate, J (xi,j g) conditional plan and J (xi,j p) consistent, it is divided into three kinds of situations: if the first xi,j p=1, then xi,j(t+1)=1, and to make the variable that in this part, miscellaneous part is corresponding be 0 entirely; If the second xi,j p=0 and xi,j(t)=0, then xi,j(t+1) remain unchanged; If the 3rd xi,j p=0 and xi,j(t)=1, then xi,j(t+1)=C (xi,j(t))。
CN201610009554.7A 2016-01-07 2016-01-07 Product manufacture-oriented combined optimization method Pending CN105678418A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610009554.7A CN105678418A (en) 2016-01-07 2016-01-07 Product manufacture-oriented combined optimization method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610009554.7A CN105678418A (en) 2016-01-07 2016-01-07 Product manufacture-oriented combined optimization method

Publications (1)

Publication Number Publication Date
CN105678418A true CN105678418A (en) 2016-06-15

Family

ID=56299350

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610009554.7A Pending CN105678418A (en) 2016-01-07 2016-01-07 Product manufacture-oriented combined optimization method

Country Status (1)

Country Link
CN (1) CN105678418A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11551155B2 (en) 2018-11-09 2023-01-10 Industrial Technology Research Institute Ensemble learning predicting method and system

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11551155B2 (en) 2018-11-09 2023-01-10 Industrial Technology Research Institute Ensemble learning predicting method and system

Similar Documents

Publication Publication Date Title
Hafezalkotob et al. Interval MULTIMOORA method integrating interval Borda rule and interval best–worst-method-based weighting model: case study on hybrid vehicle engine selection
Syahputra APPLICATION OF NEURO-FUZZY METHOD FOR PREDICTION OF VEHICLE FUEL CONSUMPTION.
CN105512258B (en) A kind of automobile Generalized Product intelligent allocation method
CN103761138B (en) Parameter correction method for traffic simulation software
CN103077288B (en) Towards hard measurement and the formula decision-making technique thereof of the multicomponent alloy material of small sample test figure
CN109685277A (en) Electricity demand forecasting method and device
CN107516107A (en) A kind of driving cycle classification Forecasting Methodology of motor vehicle driven by mixed power
CN106910337A (en) A kind of traffic flow forecasting method based on glowworm swarm algorithm Yu RBF neural
CN104679956B (en) Consider the high speed pressure machine base Reliability-based Robust Design method of dynamic characteristic
CN104657777B (en) Electric wire for automobile extrusion process parameter optimization method based on chaos wavelet neural network
CN102750286B (en) A kind of Novel decision tree classifier method processing missing data
CN115099519B (en) Oil well yield prediction method based on multi-machine learning model fusion
CN106022385A (en) Feature selection algorithm based on dynamic programming and K-means clustering
CN114912195B (en) Aerodynamic sequence optimization method for commercial vehicle
CN109741120A (en) Product depth personalized customization method for user demand
CN108803609A (en) Based on the partially observable automatic Pilot decision-making technique and system for constraining in line gauge stroke
CN107247447A (en) A kind of mixed-model assembly dispatch control method based on hybrid algorithm framework
CN103530700B (en) Urban distribution network saturation loading Comprehensive Prediction Method
CN107400935A (en) Adjusting method based on the melt-spinning technology for improving ELM
Li et al. Short term prediction of photovoltaic power based on FCM and CG-DBN combination
CN103955580A (en) Integrated circuit parameter yield estimation method based on BRB (Belief Rule Base) ratiocination
CN111832839A (en) Energy consumption prediction method based on sufficient incremental learning
Yang et al. Research on product primitives recognition in a computer-aided brand product development system
CN105678418A (en) Product manufacture-oriented combined optimization method
Liu et al. Wheel hub customization with an interactive artificial immune algorithm

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication

Application publication date: 20160615

RJ01 Rejection of invention patent application after publication