CN103310279A - Multi-objective optimization product configuration method based on hybrid ant colony algorithm - Google Patents

Abstract

The invention provides a multi-objective optimization product configuration method based on a hybrid ant colony algorithm. The method comprises the steps that (1) according to different situations of customer product orders, each order is divided into a number of production stages; (2) production nodes of each order is acquired, the production time of each order is determined, and the production cost of the order product at the order production time is calculated; (3) whether the production of each order is arranged is determined; (4) the scheduling scheme of the order requiring production is determined; (5) and the scheduling scheme is output and the production is arranged. Through the method, production scheduling can be optimized, and the production time can be rationally allocated.

Description

Multiple-objection optimization product configuration method based on hybrid ant colony

Technical field

The present invention relates to a kind of multiple-objection optimization product configuration method based on hybrid ant colony, belong to the flexible programming dispatching of production schedule field under the large-scale customization pattern, be primarily aimed at multi-objective optimization question.

Background technology

At present, mainly be the expansion of finding the solution at the expression of model and algorithm to the research direction of product configuration, its algorithms most in use mainly adopts intelligent optimization algorithm to find the solution, and representative wherein has artificial fish-swarm algorithm, genetic algorithm, Artificial Immune Algorithm, utilization decision tree and minimum conflict to revise algorithm etc.

Above method rapid solving comparatively satisfactorily on general allocation problem, but still exist part not enough:

Broad sense: define from the characteristics of the product of multiple-objection optimization configuration

1, only being structural model and functional module at product, is classification foundation with the similarity of model, is the algorithm of target with the minimum production cost mostly only.Yet under the large-scale customization pattern, the product configuration of enterprise should relate to the life cycle of entire product, should include a series of new optimization purpose and constraints such as inventory cost, customer satisfaction, dynamic vacant productive capacity, the dynamic quality of production in considering scope.

2, traditional product configuration scheme is not introduced the mechanism by time benchmark scheduling, just when running into rush order, simply with rush order as finishing as the override target, and its rationality is not analyzed.

3, to be primarily aimed at be the configuration mode of the production run of single product in enterprise to the product configuration scheme of present stage, do not embody the connection problem of production phase.

Narrow sense: discuss from the structure composition of algorithm own

1, existing general intelligence optimization algorithm mainly adopts single searching mode or single pheromones algorithm of planting of planting, and function is comparatively single, has mainly simulated the part of actual information search system.And in fact, in real life any complication system all be organized, have the division of labor, different searching routes that different pheromones regulatory mechanisms is arranged.

2, the product placement algorithm being studied its important characteristic as a system is self-organization, and this is a common trait of all intelligent optimization algorithm.Feedback is defined as the present behavior of the system that influences behavior in the future on systematics.The accumulation of pheromones on the optimal path makes algorithm advance towards optimal direction as positive feedback, yet the path node of actual occurring in nature has its intrinsic saturation degree, causes the production of node congested easily.

3, when the search volume is strengthened, because being the order of magnitude, the search node rises, and the repeatability of systematic search improves greatly, and search efficiency obviously descends.

4, under the prerequisite of multi-objective optimization question, rely on the characteristics of structural model and functional module, though limited the generation of certain invalid configuration, the convergence of system is still relatively poor, product is easy to generate a large amount of possible arrangement, need manually carry out suboptimization again.

Patent of the present invention is on the basis based on above research, for finding the solution multi-objective optimization question (multi objective optimization problem under the Mass Customization Production Model, MOP), studied a kind of polymorphic ant group algorithm of mixing that contains many ants group, many pheromones (polymorphic ant colony algorithm, PACA).And the grid partition strategy is proposed, studied the hybrid ant colony under the temporal constraint condition.Set up relevant multiple goal matrix and multi-objective restriction, and proposed suitable hybrid ant colony step.

Summary of the invention

To the objective of the invention is in order solving the problems of the technologies described above, a kind of multiple-objection optimization product configuration method based on hybrid ant colony to be provided, in the hope of realizing the disposal options of more optimizing to the order product, utilize existing resource biglyyer.

The technical scheme that the present invention takes is:

A kind of multiple-objection optimization product configuration method based on hybrid ant colony comprises the steps:

(1) according to the different situations of client's product order, each order is divided into a plurality of production phases;

(2) obtain the production node of each order, determine the production of each order constantly, calculate this order product at the production production cost constantly of this order;

(3) determine whether each order arranges production;

(4) definite scheduling scheme that needs the order of production;

(5) output scheduling scheme is arranged production.

Further, described (2) step is determined parameter by following algorithm:

By quantizing the mathematical model of order, provide:

Objective function:

$\underset{C}{{\min Z}_1}=\underset{k=1}{\overset{K}{\mathrm{\Σ}}}\underset{r=1}{\overset{N_k}{\mathrm{\Σ}}}\underset{h=1}{\overset{G}{\mathrm{\Σ}}}\underset{j=1}{\overset{N_m}{\mathrm{\Σ}}}\{C_{\mathrm{abc}}^{\mathrm{xy}}\left(t_k\right)+k_{\mathrm{abc}}^{\mathrm{xy}}{T}_{\mathrm{abc}2}^{\mathrm{xy}}\}$

$\underset{t}{{\min Z}_2}=\left|T_{\mathrm{abcexp}}^{\mathrm{xy}}\left(t_k\right)\right[{\mathrm{<figure-callout\; id="1"\; label="T\; abc"\; filenames="120309104414.png"\; state="\{\{state\}\}">T</figure-callout>}}_{\mathrm{abc}}^1\left(t_k\right)+T_{\mathrm{abc}2}^{\mathrm{xy}}\left(t_k\right)\left]\right|+\mathrm{\&Pi;}$

Model constrained:

$\underset{h=1}{\overset{G}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{N_m}{\mathrm{\&Sigma;}}}{S}_{\mathrm{abc}.\exp }^k\left(t_k\right)\&le;\underset{r=1}{\overset{N_k}{\mathrm{\&Sigma;}}}{S}_{\mathrm{abc}.\mathrm{all}}^k\left(t_k\right)$

$\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}\underset{r=1}{\overset{N_k}{\mathrm{\&Sigma;}}}[{\mathrm{<figure-callout\; id="1"\; label="T\; abc"\; filenames="120309104414.png"\; state="\{\{state\}\}">T</figure-callout>}}_{\mathrm{abc}}^1\left(t_k\right)+T_{\mathrm{abc}2}^{\mathrm{xy}}\left(t_k\right)]\&le;(1+\mathrm{\&Pi;})\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}\underset{r=1}{\overset{N_k}{\mathrm{\&Sigma;}}}{T}_{\mathrm{abc\; exp}}^{\mathrm{xy}}\left(t_k\right)$

In the formula, K is the production phase sum of order product; K is K the index in the production phase; N _kBe the production node quantity that has in k production phase; R is the productive unit index in k production phase; G is t _kThe product order kind that constantly comprises; H is the index of order kind among the set G; N _mProduct sum for every class order; J is set N _mIn index; t _kZero-time for k production phase of order i;

Be t _kConstantly the (c) individual product is being produced node (x, production cost y) for a, b;

Be that the (c) individual product is being produced node (x, unit inventory cost y) for a, b; (c) individual product is being produced node (x, expectation production time y) for a, b to for the client;

Be that the (c) individual product is being produced node (x, actual production time y) for a, b; Be that the (c) individual product is being produced node (x, extra inventory time y) for a, b; Π is made as the irreversible extension factor of enterprise, is introduced as back order tolerance parameter, can be made as definite value;

For tk gathers N constantly _kThe productive capacity summation that can provide;

Be t _kThe needed productive capacity of moment product.

The invention has the beneficial effects as follows:

For finding the solution multi-objective optimization question (multi objective optimization problem under the Mass Customization Production Model, MOP), by a kind of polymorphic ant group algorithm of mixing that contains many ants group, many pheromones (polymorphic ant colony algorithm, PACA).And the grid partition strategy is proposed, studied the hybrid ant colony under the temporal constraint condition.Set up relevant multiple goal matrix and multi-objective restriction, and proposed suitable hybrid ant colony step, the production of order has been optimized.

Description of drawings

Accompanying drawing 1 is the present invention's product configuration structure synoptic diagram under the MC pattern;

Accompanying drawing 2 is that the present invention produces node grid division synoptic diagram;

Accompanying drawing 3 is the product configuration digraphs under the temporal constraint of the present invention;

Accompanying drawing 4 is multiple goal weight decision-making process block diagrams of all kinds of orders of the present invention;

Accompanying drawing 5 is flow charts of product configuration in the multiple-objection optimization of the present invention.

Embodiment

Elaborate below in conjunction with the embodiment of accompanying drawing to the multiple-objection optimization product configuration method that the present invention is based on hybrid ant colony.

1, referring to accompanying drawing 1, sets up the quantitative mathematic model of product order, for ant group optimizing algorithm provides objective function and constraint condition.

Objective function:

$\underset{C}{{\min Z}_1}=\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}\underset{r=1}{\overset{N_k}{\mathrm{\&Sigma;}}}\underset{h=1}{\overset{G}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{N_m}{\mathrm{\&Sigma;}}}\{C_{\mathrm{abc}}^{\mathrm{xy}}\left(t_k\right)+k_{\mathrm{abc}}^{\mathrm{xy}}{T}_{\mathrm{abc}2}^{\mathrm{xy}}\}---\left(1\right)$

$\underset{t}{{\min Z}_2}=\left|T_{\mathrm{abc\; exp}}^{\mathrm{xy}}\left(t_k\right)\right[{\mathrm{<figure-callout\; id="1"\; label="T\; abc"\; filenames="120309104414.png"\; state="\{\{state\}\}">T</figure-callout>}}_{\mathrm{abc}}^1\left(t_k\right)+T_{\mathrm{abc}2}^{\mathrm{xy}}\left(t_k\right)\left]\right|+\mathrm{\&Pi;}---\left(2\right)$

Model constrained:

$\underset{h=1}{\overset{G}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{N_m}{\mathrm{\&Sigma;}}}{S}_{\mathrm{abc}.\exp }^k\left(t_k\right)\&le;\underset{r=1}{\overset{N_k}{\mathrm{\&Sigma;}}}{S}_{\mathrm{abc}.\mathrm{all}}^k\left(t_k\right)---\left(3\right)$

$\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}\underset{r=1}{\overset{N_k}{\mathrm{\&Sigma;}}}[{\mathrm{<figure-callout\; id="1"\; label="T\; abc"\; filenames="120309104414.png"\; state="\{\{state\}\}">T</figure-callout>}}_{\mathrm{abc}}^1\left(t_k\right)+T_{\mathrm{abc}2}^{\mathrm{xy}}\left(t_k\right)]\&le;(1+\mathrm{\&Pi;})\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}\underset{r=1}{\overset{N_k}{\mathrm{\&Sigma;}}}{T}_{\mathrm{abc\; exp}}^{\mathrm{xy}}\left(t_k\right)---\left(4\right)$

In the formula, K is to the production phase sum of tailor-made product under the MC pattern; K is K the index in the production phase; N _kBe the production node quantity that has in k production phase; R is the productive unit index in k production phase; G is t _kConstantly, the product order kind that comprises; H is the index of order kind among the set G; N _mProduct sum for every class order; J is set N _mIn index; t _kZero-time for k production phase of order i;

Be t _kConstantly the (c) individual product is being produced node (x, production cost y) for a, b;

Be that the (c) individual product is being produced node (x, unit inventory cost y) for a, b;

(c) individual product is being produced node (x, expectation production time y) for a, b to for the client;

Be that the (c) individual product is being produced node (x, actual production time y) for a, b;

Be that the (c) individual product is being produced node (x, extra inventory time y) for a, b; Π is made as the irreversible extension factor of enterprise, is introduced as back order tolerance parameter, can be made as definite value and not discuss;

For tk gathers N constantly _kThe productive capacity summation that can provide;

Be t _kThe needed productive capacity of moment product.

In multi-objective optimization question, the layoutprocedure of product often relates to n factor, and its weight relationship is mainly decided by the client.Patent of the present invention quantizes for the qualitative language between any two factors, thereby sets up hierarchical structure for numerous factors by improved analytical hierarchy process, provides foundation for system sets up product model.

(x y) is illustrated in the importance scale of factor x contrast factor y under the system synthesis evaluation to introduce function g.

If the property set of K client's order is W=[w ₁, w ₂, w ₃... w _n], W _NmThe comment collection Q=[L of product configuration is set up in the specific requirement of order attributes collection W ₁, L ₂, L ₃, L ₄], definition α: W → L makes the judge function, then can obtain the client for the multifactorial evaluation collection W of W _Nm

W _nm＝f(α(α ₁)，α，α ₂)，…，αα( _n))

In the formula, i=1,2 ..., N; L ₁, L ₂, L ₃, L ₄Be different scale parameters, and L ₁＜L ₂＜ _L, " L ₄Definition L ₁=1, L ₂=2, L ₃=3, L ₄=6.(institute's setting parameter might not be the regulation parameter of patent of the present invention for used for example herein, can determine this coefficient by the client)

The weight sets of definition order demand collection is combined into R=[r ₁, r ₂, r ₃... r _n], adopt the normalization weight vector:

Set up weighted mean type objective function for realizing product configuration global optimum:

$f\left(W\right)=\underset{n=1}{\overset{K}{\mathrm{\&Sigma;}}}{R}_n{W}_{\mathrm{nm}}---\left(6\right)$

The judgement rate matrix of structure evaluation index

$Q=\stackrel{\begin{array}{cccc}{q}_1& q_2& \&CenterDot;\&CenterDot;\&CenterDot;& q_n\end{array}}{\left[\begin{array}{cccc}{q}_{11}& q_{12}& \&CenterDot;\&CenterDot;\&CenterDot;& q_{1n}\\ & & & \&CenterDot;\\ q_{21}& q_{22}& \&CenterDot;\&CenterDot;\&CenterDot;& \&CenterDot;\\ & & & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ {\mathrm{<figure-callout\; id="1"\; label="q\; n"\; filenames="120309104414.png"\; state="\{\{state\}\}">q</figure-callout>}}_n& \&CenterDot;\&CenterDot;\&CenterDot;& \&CenterDot;\&CenterDot;\&CenterDot;& q_{\mathrm{nn}}\end{array}\right]}\begin{array}{c}{q}_1\\ q_2\\ q_3\\ q_4\end{array},Q\&RightArrow;R---\left(7\right)$

If a _Ij=f (q _i, q _j), find the solution and judge the rate matrix proper vector, the structure fuzzy relation matrix:

${W_{\mathrm{nm}}}^{*}={\left(\frac{r_i}{r_j}\right)}_{n\&times;n}={\left(a_{\mathrm{ij}}^{*}\right)}_{n\&times;n}---\left(8\right)$

$\left[\begin{array}{cccc}\frac{r_1}{r_1}& \frac{r_1}{r_2}& \&CenterDot;\&CenterDot;\&CenterDot;& \frac{r_1}{r_n}\\ & & & \&CenterDot;\\ \frac{r_2}{r_1}& \frac{r_2}{r_2}& \&CenterDot;\&CenterDot;\&CenterDot;& \&CenterDot;\\ & & & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ \&CenterDot;& \&CenterDot;& \&CenterDot;\&CenterDot;\&CenterDot;& \&CenterDot;\\ \&CenterDot;& \&CenterDot;& & \&CenterDot;\\ \frac{{\mathrm{<figure-callout\; id="1"\; label="r\; n\; r"\; filenames="120309104414.png"\; state="\{\{state\}\}">r</figure-callout>}}_n}{r_{}}& \&CenterDot;\&CenterDot;\&CenterDot;& \&CenterDot;\&CenterDot;\&CenterDot;& \frac{r_n}{r_n}\end{array}\right]\left[\begin{array}{c}{r}_1\\ r_2\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ r_n\end{array}\right]=n\left[\begin{array}{c}{r}_1\\ r_2\\ \&CenterDot;\\ \&CenterDot;\\ \&CenterDot;\\ r_n\end{array}\right]---\left(9\right)$

According to the adjustment flow process of Fig. 1, at the optimization operation main line of rush order and non-emergent order, bring formula (5) into objective function and model constrained, set up judgment matrix.Set up the optimization aim function of non-emergent order and rush order on this basis respectively.

If integrated cost Z ₁, comprehensive production time Z ₂Respectively with q ₁, q ₂Be mapping relations, client's order is at t _kConstantly, all the other manufacturing parameters in k production phase respectively and q ₃Q _nBe mapping relations.

In sum, first kind product order (non-emergent order) is established n=5, and q ₃The extra stock of representative products, q ₄Representative products overall quality, q ₅Represent the productive capacity saturation degree, calculate and to get by formula (6)-(9).

$W_{\mathrm{nm}11}=\left[\begin{array}{ccccc}1& 6& 6& 2& 2\\ \frac{1}{6}& 1& \frac{1}{2}& \frac{1}{3}& \frac{1}{2}\\ \frac{1}{6}& 2& 1& \frac{1}{3}& 1\\ \frac{1}{2}& 3& 3& 1& 2\\ \frac{1}{2}& 2& 1& \frac{1}{2}& 1\end{array}\right]$

Adopt and method is found the solution it, the weight vector that obtains after the normalization is

B ₁₁＝(0.4333，0.0687，0.1070，0.2474，0.1436)

So set up objective optimization generic function.

minf(W) ₁₁＝0.4333w ₁+0.0687w ₂+0.1070w ₃+0.2474w ₄+0.1436w ₅(10)

The second series products order (rush order)

At first, adopt formula (10) to set up mathematical model:

minf(W) ₂₁＝0.4333w ₁+0.0687w ₂+0.1070w ₃+0.2474w ₄+0.1436w ₅(11)

Judge and whether satisfy constraint condition formula (4) this moment.

Satisfy output, adjust the target component collection as not satisfying, set up following matrix:

$W_{\mathrm{nm}22}=\left[\begin{array}{ccccc}1& \frac{1}{6}& \frac{1}{2}& 2& 1\\ 6& 1& 6& 3& 2\\ 2& \frac{1}{6}& 1& \frac{1}{2}& 1\\ \frac{1}{2}& \frac{1}{3}& 2& 1& 2\\ 1& \frac{1}{2}& 1& \frac{1}{2}& 1\end{array}\right]$

Adopt and method is found the solution it, the weight vector that obtains after the normalization is

B ₂₂＝(0.1297，0.4637，0.1154，0.1641，0.1271)

minf(W) ₂₂＝0.1297w ₁+0.4637w ₂+0.1154w ₃+0.1641w ₄+0.1271w ₅(12)

3, the present invention adopts the grid partition strategy, and establishing the nodal point number variable is n, and it is divided into the K equal portions, thereby changes the decision-making of n variable into K level decision problem.Behind its importing actual production node set, adopt the production phase for dividing according to the actual production node set being carried out elementary grid.Under can not the prerequisite of five equilibrium, introduce virtual production node concept, and it is made as does not definitely enter node, thereby the space complexity of ant group algorithm can be reduced an order of magnitude.The grid dividing mode as shown in Figure 2.

4, be constraint condition with product order sequential, carry out the classification second time to producing node set.At the complex product order, divide for the first time and to obtain on the basis of each stage relations of production, with initial sum closing time in production stage of customer requirement be constraint condition, finish the collaborative digraph customization of making chain of various product and divide.The structure of its manufacturing operation can adopt mode shown in Figure 3 to represent.

In sum, according to the first time and the division result second time, can determine the corresponding production node of each class ant in tk moment ant group algorithm, thereby determine the feasible zone of algorithm, further reduce the space complexity of this algorithm.

6, patent of the present invention adopts hybrid ant colony to carry out the product configuration

In this algorithm, at t moment ant k by the state transition probability that element i transfers to element j be:

Wherein, α is the heuristic factor of information, the relative importance of expression track, and its value is more big, and cooperation is more strong between ant; β is the heuristic factor of expectation, the relative importance of expression visibility, and its value is more big, the more approaching greed rule of transition probability.η _Ij(t) be heuristic function, its expression formula is defined as:

τ _Ij(t) be the quantity of information function according to pheromones update strategy difference, adopt model:

The pheromones update rule is as follows:

${\mathrm{\&tau;}}_{\mathrm{ij}}(t+n)=(1-\mathrm{\&rho;})\&times;{\mathrm{\&tau;}}_{\mathrm{ij}}\left(t\right)+{\mathrm{\&Delta;\&tau;}}_{\mathrm{ij}}\left(t\right),{\mathrm{\&Delta;\&tau;}}_{\mathrm{ij}}\left(t\right)=\underset{k=1}{\overset{m}{\mathrm{\&Sigma;}}}{\mathrm{\&Delta;\&tau;}}_{\mathrm{ij}}^k\left(t\right),$ Volatility coefficient $\mathrm{\&rho;}\&Subset;\left[\mathrm{0,1}\right).$

Hybrid ant colony is as a kind of multi-objective optimization question of finding the solution, mutual restriction or conflict often between this moment each target.We can say that under the prerequisite of multiple-objection optimization, the quality of solution has certain relativity, and ant group's pheromones also should have certain otherness.As ant i during in optimizing, the quantity of information θ that the companion discharges _iCorresponding difference should be arranged.According to the 2.2 multi objective fuzzy product models set up of joint, and definition B → θ _i, set ant X _IjProduce node (x, the probability of y) producing is:

$P_{\begin{array}{ccc}B& \mathrm{ij}& \mathrm{xy}\end{array}}^A=B\&CenterDot;P_{\begin{array}{cc}\mathrm{ij}& \mathrm{xy}\end{array}}^{\left(A\right)}---\left(15\right)$

Default parameters is set to α=1, β=1, and ρ=0.2, maximum cycle is 200.Determine that according to the description in 2.2 joints multiple goal parameter with the relation that all kinds of ant groups leave over the pheromones amount, determines that its new probability formula is:

A, first kind order

$P_{\begin{array}{ccc}B& \mathrm{ij}& \mathrm{xy}\end{array}}^{A1}=0.4333{\mathrm{<figure-callout\; id="1"\; label="P"\; filenames="120309104414.png"\; state="\{\{state\}\}">P</figure-callout>}}_1+0.0687P_2+0.1070P_3+0.2474P_4+0.1436\left(\begin{array}{cc}1& P_5\end{array}\right)---\left(16\right)$

B, the second class order

In the formula, (x y) produces in the node k class key element to ant A _IjThe attraction probability be $P_n=\frac{{\mathrm{\&delta;}}_{\begin{array}{cc}\mathrm{ij}& \mathrm{xy}\end{array}}^n}{\underset{j=1}{\overset{m}{\mathrm{\&Sigma;}}}{\mathrm{\&delta;}}_{\begin{array}{cc}\mathrm{ij}& \mathrm{xy}\end{array}}^n},$ I ≠ j, i, j=1,2 ..., N;

Be the pheromones amount.

7, the design process of patent of the present invention is simply described, shown in accompanying drawing 4,5:

Step 1, according to the different situations of client's product order, enterprise determines that by flow process shown in the figure two the dynamic dispatching adjustment is constantly.

Step 2, algorithm begin, and according to the information type of product order, form different production task classifications, construct corresponding ant classification correspondence with it.

Step 3, construct it by the fuzzy decision scheme for dissimilar ant groups and judge rate matrix, thereby determine the majorized function of multi-objective problem, determine every weight index.

Step 4, set for every class ant and not enter node accordingly, determine feasible zone.

Step 5, divide and the temporal constraint condition is dwindled the search volume by grid, and determine the initial moment and the planned production time of each order.

Step 6, according to Order Type, determine that according to the description in 2.2 joints multiple goal parameter leaves over the relation of pheromones amount with all kinds of ant groups, to determine to select the different probability of producing nodes.

Step 7, time t and cycle index N are set to zero, and maximum cycle is set.Make every limit that (i, j) the initialization information amount on is τ _Ij(t)=const (const is constant), and initial initial time Δ τ is set _Ij=0.

Step 8, generate N at source point _cBatch ant, every batch comprises all kinds of ants in this batch, the quantity of establishing all kinds of ants is 100.All ants adopt the roulette wheel method to select the path according to the determined new probability formula of step 6, and the lastest imformation element.

Step 9, record in this batch the ant quantity by each node, and record corresponding optimization target values.

Step 10, the renewal optimal route value of information, cycle index N _c← N _c+ 1, go back to step 8 simultaneously.

Step 11, reach maximum cycle and finish output.Calculate the optimization aim of this moment.Judge whether to satisfy the constraint condition of model.As meet the commentaries on classics step 12, otherwise change step 3.

Step 12, according to all kinds of ants of record in the distribution condition of producing on the node, carry out the despatching work of product configuration.

In sum, under Mass Customization Production Model, enterprise requires and will improve constantly for the flexible operating efficiency of production system.Based on this kind prospect, the present invention is directed to the client for the multifactorial evaluation collection of order, set up Model for Multi-Objective Optimization; And a kind of hybrid ant colony proposed.In patent of the present invention, algorithm constantly all has versatility for each of production phase, has strengthened the degradation mechanism of algorithm, introduces the productive capacity saturation degree, has eliminated the congestion problems of node; And adopt Fuzzy AHP to quantize the weight vectors of multi-objective optimization question; At last, relying on grid division and temporal constraint further to dwindle and find the solution the space, is to solve the better means that multitask complex apparatus product layoutprocedure is found the solution.

Under Mass Customization Production Model, the client improves constantly the complicacy of tailor-made product, and the unordered of bulky complex configuration data is processed in order to restrict the main constraint of enterprise's benign development.The present invention is driving with the customer demand, has improved the validity of flexible product configuration schedules planning.

The above only is preferred implementation of the present invention; should be pointed out that for those skilled in the art, under the prerequisite that does not break away from the principle of the invention; can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.

Claims (2)

1. the multiple-objection optimization product configuration method based on hybrid ant colony comprises the steps:

(1) according to the different situations of client's product order, each order is divided into a plurality of production phases;

(2) obtain the production node of each order, determine the production of each order constantly, calculate this order product at the production production cost constantly of this order;

(3) determine whether each order arranges production;

(4) definite scheduling scheme that needs the order of production;

(5) output scheduling scheme is arranged production.

2. the multiple-objection optimization product configuration method based on hybrid ant colony according to claim 1 is characterized in that: described (2) step is determined parameter by following algorithm:

By quantizing the mathematical model of order, provide:

Objective function:

$\underset{C}{{\min Z}_1}=\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}\underset{r=1}{\overset{N_k}{\mathrm{\&Sigma;}}}\underset{h=1}{\overset{G}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{N_m}{\mathrm{\&Sigma;}}}\{C_{\mathrm{abc}}^{\mathrm{xy}}\left(t_k\right)+k_{\mathrm{abc}}^{\mathrm{xy}}{T}_{\mathrm{abc}2}^{\mathrm{xy}}\}$

$\underset{t}{{\min Z}_2}=\left|T_{\mathrm{abc\; exp}}^{\mathrm{xy}}\left(t_k\right)\right[T_{\mathrm{abc}1}^{\mathrm{xy}}\left(t_k\right)+T_{\mathrm{abc}2}^{\mathrm{xy}}\left(t_k\right)\left]\right|+\mathrm{\&Pi;}$

Model constrained:

$\underset{h=1}{\overset{G}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{N_m}{\mathrm{\&Sigma;}}}{S}_{\mathrm{abc}.\exp }^k\left(t_k\right)\&le;\underset{r=1}{\overset{N_k}{\mathrm{\&Sigma;}}}{S}_{\mathrm{abc}.\mathrm{all}}^k\left(t_k\right)$

$\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}\underset{r=1}{\overset{N_k}{\mathrm{\&Sigma;}}}[T_{\mathrm{abc}1}^{\mathrm{xy}}\left(t_k\right)+T_{\mathrm{abc}2}^{\mathrm{xy}}\left(t_k\right)]\&le;(1+\mathrm{\&Pi;})\underset{k=1}{\overset{K}{\mathrm{\&Sigma;}}}\underset{r=1}{\overset{N_k}{\mathrm{\&Sigma;}}}{T}_{\mathrm{abc\; exp}}^{\mathrm{xy}}\left(t_k\right)$

In the formula, K is the production phase sum of order product; K is K the index in the production phase; N _kBe the production node quantity that has in k production phase; R is the productive unit index in k production phase; G is t _kThe product order kind that constantly comprises; H is the index of order kind among the set G; N _mProduct sum for every class order; J is set N _mIn index; t _kZero-time for k production phase of order i;

Be t _kConstantly the (c) individual product is being produced node (x, production cost y) for a, b;

Be that the (c) individual product is being produced node (x, unit inventory cost y) for a, b;

(c) individual product is being produced node (x, expectation production time y) for a, b to for the client;

Be that the (c) individual product is being produced node (x, actual production time y) for a, b;

Be that the (c) individual product is being produced node (x, extra inventory time y) for a, b; Π is made as the irreversible extension factor of enterprise, is introduced as back order tolerance parameter, can be made as definite value;

Be t _kConstantly gather N _kThe productive capacity summation that can provide;

Be t _kThe needed productive capacity of moment product.

Images