CN104504469A - Box loading method based on three-dimensional moving mode sequence and multi-intelligent-agent genetic algorithm - Google Patents
Box loading method based on three-dimensional moving mode sequence and multi-intelligent-agent genetic algorithm Download PDFInfo
- Publication number
- CN104504469A CN104504469A CN201410799112.8A CN201410799112A CN104504469A CN 104504469 A CN104504469 A CN 104504469A CN 201410799112 A CN201410799112 A CN 201410799112A CN 104504469 A CN104504469 A CN 104504469A
- Authority
- CN
- China
- Prior art keywords
- intelligent body
- chest
- maximum
- generation
- volume utilization
- 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
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"
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Processing Or Creating Images (AREA)
Abstract
The invention discloses a box loading method based on a three-dimensional moving mode sequence and a multi-intelligent-agent genetic algorithm. The method mainly solves the problem of low volume utilization rate during the box placement in the prior art. According to the method, the three-dimensional moving mode sequence is used as a box loading decoding process, and a multi-intelligent-agent system is combined with the genetic algorithm, and is used for solving the three-dimensional box loading problem. The method comprises the following realization steps that firstly, each intelligent agent in an intelligent agent lattice is randomly initialized according to requirements; then, the three-dimensional moving mode sequence is designed for decoding each intelligent agent; finally, a designed neighborhood competition operator, a designed neighborhood crossing operator, a designed mutation operator and a designed self learning operator are used for optimizing the intelligent agents, and the box loading optimum result is obtained. The box loading method has the advantages that the time complexity is reduced, the volume utilization rate of a container is improved, and the method can be used for solving the three-dimensional box loading problem with different constraint conditions and optimization targets.
Description
Technical field
The invention belongs to computer optimization field, relate to a kind of packing method based on three-dimensional Move Mode sequence 3DMBS and multi-Agent Genetic Algorithm MAGA further, can be used for the optimization realizing the industrial circle bin packings such as logistics.
Background technology
Three-Dimensional Packing Problem 3D-BPP appears at the industrial circles such as logistics, cutting processing, the placement of circuit board chip, transport widely.How to improve the efficiency of vanning, become the important topic paid special attention in scientific research and practice, Efficient Solution algorithm has very positive realistic meaning to the profitability reducing costs, improve enterprise.Three-Dimensional Packing Problem is again a NP-hard problem simultaneously, although the history of research is very long, is difficult to obtain effective derivation algorithm.
" A hybrid genetic algorithm for packing in 3D with deepest bottom leftwith fill method " (" Advances in Information Systems " that Karabulut etc. deliver, article is numbered: 441-450 (2004)), the core of this article is placed in three dimensions by chest, make coordinate points z minimum, then y is minimum, last x is minimum, this algorithm can produce good effect, but often put into a chest, will record and delete limit, namely be the point that chest can be placed in record, and placed the point of chest and deleted, finally again to give Ordination, time complexity is very high.
" solving the hybrid analog-digital simulation annealing algorithm of Three-Dimensional Packing Problem " (" Chinese journal of computers " that Zhang Defu etc. deliver, article is numbered: propose composite block as Placement Strategy 2147-2156 (2009) 32-11), adopts Simulated Anneal Algorithm Optimize 3D-BPP.This algorithm research be single Three-Dimensional Packing Problem with stability and direction constrain, to the filling rate that the weak isomery chest of part can obtain, and for strong isomery chest, according to the Placement Strategy of composite block, can space be produced in composite block and can not again fill for each composite block, thus filling rate is reduced.
The paper " the binding cave degree algorithm based on rectangular parallelepiped Packing problem " that He Kun etc. deliver (" Journal of Software ", number: disclose the method that a kind of binding cave degree algorithm is optimized Three-Dimensional Packing Problem 842-851 (2011) 2-5) by article.The implementation procedure of the method is first bundle all chests, forms new chest combination, secondly adopts cave degree algorithm to be optimized Three-Dimensional Packing Problem.The weak point of the method is, is only applicable to the Three-Dimensional Packing Problem that chest kind is few, and bin packing its volume utilization of obtaining many for chest kind is low.
Summary of the invention
The object of the invention is to the deficiency for above-mentioned prior art, propose a kind of packing method based on three-dimensional Move Mode sequence and multi-Agent evolutionary Algorithm, to reduce time complexity, improve container volume utilization factor.
For achieving the above object, technical scheme of the present invention comprises the steps:
(1) setting parameter:
If Po is neighborhood competition probability, Pc is crossover probability, and Pm is mutation probability; M is the value of Move Mode, and value is 0,1,2; L
size× L
sizefor the sizing grid of multiple agent, t be more than or equal to 0 integer, represent t generation, Max is maximum evolutionary generation, L
trepresent that t is for Agent Grid, L
t+1/3and L
t+2/3l
tand L
t+1between centre for Agent Grid, Best
tl
0, L
1..., L
tthe intelligent body that middle global volume utilization factor is maximum, CBest
tl
tthe intelligent body that middle volume utilization is maximum; SL
size× sL
sizefor the sizing grid of multiple agent during self study, SMax is the maximum evolutionary generation of self study, Agent Grid when sL is self study; L
i,jexpression is in the i-th row of Agent Grid, the intelligent body of jth row, energy (L
i,j) represent this intelligent body L
i,jenergy; M is the chest number in each intelligent body, M be more than or equal to 1 integer, represent the wide, high, long of container respectively with W, H, L; RX, RY, RZ represent the wide, high, long of minimum cube envelope respectively; Rand be random produce 0 to 1 between real number;
(2) initialization Agent Grid L
0, upgrade the intelligent body Best that initial global volume utilization factor is maximum
0, make t=0;
(3) recursive call is decoded to each intelligent body based on the algorithm of three-dimensional Move Mode sequence, calculates the volume utilization of each intelligent body, upgrades the intelligent body Best that initial global volume utilization factor is maximum
0;
(4) to Agent Grid L
tin each intelligent body, adopt neighborhood competition operator carry out first time optimize, obtain first time optimization after Agent Grid L
t+1/3;
(5) the Agent Grid L after first time being optimized
t+1/3in each intelligent body, if rand<Pc, then adopt neighborhood crossover operator to carry out second time optimization, obtain the Agent Grid L after second time optimization
t+2/3;
(6) the Agent Grid L after second time being optimized
t+2/3in each intelligent body, if rand<Pm, then adopt mutation operator carry out third time optimize, obtain third time optimization after Agent Grid L
t+1;
(7) the Agent Grid L after optimizing from third time
t+1in find out the maximum intelligent body CBest of the volume utilization in t generation
t;
(8) to the intelligent body CBest that the volume utilization in t generation is maximum
temploying self-learning operator is optimized, and obtains the intelligent body CBest that the volume utilization in t+1 generation is maximum
t+1;
(9) by intelligent body CBest maximum for the volume utilization in t+1 generation
t+1energy energy and the maximum intelligent body Best of the global volume utilization factor in front t generation
tenergy energy compare, if energy (CBest
t+1) >energy (Best
t), then perform step (10), otherwise perform step (11);
(10) intelligent body CBest maximum for the volume utilization in t+1 generation
t+1the intelligent body Best that before giving, the global volume utilization factor in t+1 generation is maximum
t+1, perform step (12);
(11) intelligent body Best maximum for the global volume utilization factor in front t generation
tthe intelligent body Best that global volume utilization factor before giving in t+1 generation is maximum
t+1with the intelligent body CBest that the volume utilization in t+1 generation is maximum
t+1, perform step (12);
(12) judge whether t reaches maximum evolutionary generation Max, if t<Max, making t from adding 1, returning step (4), otherwise, export the intelligent body Best that final global volume utilization factor is maximum
t+1, as vanning result, wherein comprise the sequence number of chest, rotary mode, Move Mode, size, coordinate points and maximum volume utilization factor.
The present invention has the following advantages compared with prior art:
1. the present invention is owing to adopting the decoding process of three-dimensional Move Mode sequence, and the design not being only genetic algorithm provides platform, and reduces the memory space of space internal memory.
2. the present invention is owing to adopting the optimal way of multi-Agent Genetic Algorithm, reduces the scale of population, accelerates speed of convergence, improve the stability of vanning.
3. the present invention is owing to adopting four kinds of Optimizing operator: neighborhood competition operator, neighborhood crossover operator, mutation operator and self-learning operator, be optimized Agent Grid by these four kinds of Optimizing operator, obtain larger volume utilization.
Accompanying drawing explanation
Fig. 1 of the present inventionly realizes general flow chart;
Fig. 2 is the schematic diagram of the upper covering of each chest in the present invention, right covering and front covering;
Fig. 3 is three kinds of Move Mode schematic diagram of the initial position of each chest in the present invention;
Fig. 4 is the decoding sub-process figure that the present invention is based on three-dimensional Move Mode sequence;
Fig. 5 is the optimal placement result schematic diagram emulated when different rotary 20 chests with the present invention;
Fig. 6 is the optimal placement result schematic diagram emulated when different rotary 60 chests with the present invention.
Embodiment
The Three-Dimensional Packing Problem of the present invention's research, refers to and installs in fixed container by chest as much as possible under the condition meeting feasible placement schemes, make volume of a container utilization factor maximum, and test data.
The present invention's important feature is, be that two-dimensional movement mode sequences is expanded to three-dimensional, and the method using chest to project to plane carries out record to the information of chest, it designs a model and is:
If Edgxz represents that chest projects to xoz plane, Edgxy represents that chest projects to xoy plane, and Edgyz represents that chest projects to yoz plane, wherein:
The information of Edgxz comprises:
Xl: when chest projects to xoz plane, the x coordinate figure that border is minimum;
Xr: when chest projects to xoz plane, the x coordinate figure that border is maximum;
Yt: the y coordinate figure of coordinate points when not projecting;
Zl: when chest projects to xoz plane, the z coordinate value that border is minimum;
Zr: when chest projects to xoz plane, the z coordinate value that border is maximum.
The information of Edgxy comprises:
Xl: when chest projects to xoy plane, the x coordinate figure that border is minimum;
Xr: when chest projects to xoy plane, the x coordinate figure that border is maximum;
Yl: when chest projects to xoy plane, the y coordinate figure that border is minimum;
Yr: when chest projects to xoy plane, the y coordinate figure that border is maximum;
Zt: the z coordinate value of coordinate points when not projecting.
The information of Edgyz comprises:
Xt: the x coordinate figure of coordinate points when not projecting;
Yl: when chest projects to yoz plane, the y coordinate figure that border is minimum;
Yr: when chest projects to yoz plane, the y coordinate figure that border is maximum;
Zl: when chest projects to yoz plane, the z coordinate value that border is minimum;
Zr: when chest projects to yoz plane, the z coordinate value that border is maximum.
Represent that the chest placed projects to xoz plane with BTT, the information that the information that it comprises and Edgxz comprise is identical; Represent that the chest placed projects to xoy plane with BTF, the information that the information that it comprises and Edgxy comprise is identical; Represent that the chest placed projects to yoz plane with BTF, the information that the information that it comprises and Edgyz comprise is identical.
With reference to Fig. 1, specific embodiment of the invention step is as follows:
Step 1. setting parameter.
If Po is neighborhood competition probability, Pc is crossover probability, and Pm is mutation probability; M is the value of Move Mode, and value is 0,1,2; L
size× L
sizefor the sizing grid of multiple agent, t be more than or equal to 0 integer, represent t generation, Max is maximum evolutionary generation, L
trepresent that t is for Agent Grid, L
t+1/3and L
t+2/3l
tand L
t+1between centre for Agent Grid, Best
tl
0, L
1..., L
tthe intelligent body that middle global volume utilization factor is maximum, CBest
tl
tthe intelligent body that middle volume utilization is maximum; SL
size× sL
sizefor the sizing grid of multiple agent during self study, SMax is the maximum evolutionary generation of self study, Agent Grid when sL is self study; L
i,jexpression is in the i-th row of Agent Grid, the intelligent body of jth row, energy (L
i,j) represent this intelligent body L
i,jenergy; M is the chest number in each intelligent body, M be more than or equal to 1 integer, represent the wide, high, long of container respectively with W, H, L; RX, RY, RZ represent the wide, high, long of minimum cube envelope respectively; Rand be random produce 0 to 1 between real number.
Step 2. initialization Agent Grid L
0, upgrade the intelligent body Best that initial global volume utilization factor is maximum
0, make t=0.
Step 3. is decoded to each intelligent body.
3.1) the upper covering of chest, right covering and front covering is defined:
With reference to Fig. 2, the present invention is defined as follows the upper covering of chest, right covering and front covering:
3.11) with reference to Fig. 2 a, the definition of upper covering: compared with the information Edgxz projected to by chest a in xoz plane projects to the information BTT in xoz plane with the chest b placed, if discontented foot formula:
(Edgxz.yt＜BTT.yt)or(Edgxz.xr＜＝BTT.xl)or(Edgxz.xl＞＝BTT.xr)or
(Edgxz.zr＜＝BTT.zl)or(Edgxz.zl＞＝BTT.zr)
Then illustrate and chest a is coated with the chest b placed, be the upper covering of chest;
3.12) with reference to Fig. 2 b, the definition of right covering: compared with the information Edgxy projected to by chest a in xoy plane projects to the information LTR in xoy plane with the chest b placed, if discontented foot formula:
(Edgxy.zt＜LTR.zt)or(Edgxy.xr＜＝LTR.xl)or(Edgxy.xl＞＝LTR.xr)or
(Edgxy.yr＜＝LTR.yl)or(Edgxy.yl＞＝LTR.yr)
Then illustrate that the chest a right side is coated with the chest b placed, be the right covering of chest;
3.13) with reference to Fig. 2 c, the definition of front covering: compared with the information Edgyz projected to by chest a in yoz plane projects to the information BTF in yoz plane with the chest b placed, if discontented foot formula:
(Edgyz.xt＜BTF.xt)or(Edgyz.yr＜＝BTF.yl)or(Edgyz.yl＞＝BTF.yr)or
(Edgyz.zr＜＝BTF.zl)or(Edgyz.zl＞＝BTF.zr)
Be coated with the chest b placed before then chest a being described, be the front covering of chest;
3.2) three-dimensional Move Mode is defined:
With reference to Fig. 3, the present invention is defined as follows three-dimensional Move Mode:
The first Move Mode is 0, and the second Move Mode is 1, and the third Move Mode is 2, and often kind of Move Mode has two kinds of mobile statuss, wherein:
Two kinds of mobile statuss of the first Move Mode 0: one is first moved down by chest to find out upper covering, more front covering is found out in movement backward, is finally moved to the left and finds out right covering; Two is first moved down by chest to find out upper covering, then is moved to the left and finds out right covering, and finally front covering is found out in movement backward;
Two kinds of mobile statuss of the second Move Mode 1: one is first be moved to the left by chest to find out right covering, more front covering is found out in movement backward, finally moves down and finds out covering; Two is first be moved to the left by chest to find out right covering, then moves down and find out covering, and finally front covering is found out in movement backward;
Two kinds of mobile statuss of the third Move Mode 2: one be by chest first backward movement find out front covering, then be moved to the left and find out right covering, finally move down and find out covering; Two be by chest first backward movement find out front covering, then move down and find out covering, be finally moved to the left and find out right covering;
3.3) decoding based on three-dimensional Move Mode sequence is carried out to each intelligent body:
With reference to Fig. 4, being implemented as follows of this step:
(3.3a) length of initialization M chest, random generation Move Mode m, makes the initial coordinate point of all chests for (-1 ,-1 ,-1); If temp is a number produced at random, value is 0 or 1; If k is counter;
(3.3b) coordinate points of the 0th chest be set to (0,0,0), wide RX, high RY, the long RZ of initialization minimum cube envelope, make k=1;
(3.3c) judge whether to meet k<M, if meet, then perform step (3.3d), otherwise perform step (3.3r);
(3.3d) according to the value of Move Mode m, determine subsequent operation: if m=0, as temp=0, perform step (3.3e), as temp=1, perform step (3.3g); If m=1, as temp=0, perform step (3.3i), as temp=1, perform step (3.3k); If m=2, as temp=0, perform step (3.3m), as temp=1, perform step (3.3o);
(3.3e) moved down respectively by chest and find out upper covering, front covering is found out in movement backward, is moved to the left and finds out right covering; (3.3f) judge that can chest also move again, if can, then return step (3.3e), otherwise perform step (3.3q);
(3.3g) moved down respectively by chest and find out upper covering, be moved to the left and find out right covering, front covering is found out in movement backward;
(3.3h) judge that can chest also move again, if can, then return step (3.3g), otherwise perform step (3.3q);
(3.3i) chest is found out right covering respectively to moving left, front covering is found out in movement backward, moves down and finds out covering;
(3.3j) judge that can chest also move again, if can, then return step (3.3i), otherwise perform step (3.3q);
(3.3k) chest being found out right covering respectively to moving left, moving down and finding out covering, front covering is found out in movement backward;
(3.3l) judge that can chest also move again, if can, then return step (3.3k), otherwise perform step (3.3q);
(3.3m) chest is found out front covering respectively to rear movement, be moved to the left and find out right covering, move down and find out covering;
(3.3n) judge that can chest also move again, if can, then return step (3.3m), otherwise perform step (3.3q);
(3.3o) chest is found out front covering respectively to rear movement, move down and find out covering, be moved to the left and find out right covering;
(3.3p) judge that can chest also move again, if can, then return step (3.3o), otherwise perform step (3.3q);
(3.3q) judge whether chest loads in container completely, and if so, k is from adding 1, returns step (3.3c), otherwise, upgrading the wide RX of minimum cube envelope, high RY, long RZ, k from adding 1, returning step (3.3c);
(3.3r) decode procedure of vanning is terminated.
Each intelligent body in step 4. pair Agent Grid carries out three suboptimization.
(4a) neighborhood is adopted to compete operator to Agent Grid L
tin each intelligent body carry out first time optimize, obtain first time optimize after Agent Grid L
t+1/3, its optimal way is as follows:
By intelligent body L
i,jenergy energy and its neighborhood in the maximum intelligent body L of energy
i,j maxenergy energy compare, if energy (L
i,j) >energy (L
i,j max), then intelligent body L
i,jcontinue survival on grid, otherwise, by intelligent body L maximum for energy in neighborhood
i,j maxa part of information exchange, obtain first time optimize after Agent Grid L
t+1/3;
(4b) the Agent Grid L after first time being optimized
t+1/3in the random number rand that produces of each intelligent body and neighborhood intersect probability P c and compare, if rand<Pc, then neighborhood crossover operator is adopted to carry out second time optimization to each intelligent body, otherwise, terminate second time optimization, obtain the Agent Grid L after second time optimization
t+2/3, its optimal way is as follows:
4b1) for intelligent body L
i,jthe order of middle chest: Stochastic choice point, by intelligent body L
i,jchest sequence number at that point and the maximum intelligent body L of its neighboring region energy
i,j maxchest sequence number at that point exchanges;
4b2) for intelligent body L
i,jthe rotary mode of middle chest and Move Mode adopt the method for two-point crossover: Stochastic choice two points, by intelligent body L
i,jthe intelligent body L that rotary mode between these 2 or Move Mode are maximum with its neighboring region energy respectively
i,j maxrotary mode between these 2 or Move Mode exchange;
Through above-mentioned 4b1) and 4b2) after obtain second time optimize after Agent Grid L
t+2/3;
(4c) the Agent Grid L after second time being optimized
t+2/3in the random number rand that produces of each intelligent body compared with mutation probability Pm, if rand<Pm, then adopt mutation operator to carry out third time to each intelligent body and optimize, otherwise, terminate third time to optimize, obtain the Agent Grid L after second time optimization
t+1, its optimal way is as follows:
4c1) for intelligent body L
i,jthe order of middle chest: Stochastic choice two point, exchanges intelligent body L
i,jin the sequence number of this chest of 2;
4c2) for intelligent body L
i,jthe rotary mode of middle chest: Stochastic choice a bit, selects an integer to replace intelligent body L at random from the integer between 0 to 5
i,jat the rotary mode of this point;
4c3) for intelligent body L
i,jthe Move Mode of middle chest: a bit, Stochastic choice integer replaces intelligent body L to Stochastic choice among 0,1,2
i,jat the Move Mode of this point;
Through above-mentioned 4c1), 4b2) and 4b3) after obtain third time optimize after Agent Grid L
t+1.
Step 5. is from the Agent Grid L after third time optimization
t+1in find out the maximum intelligent body CBest of the volume utilization in t generation
t.
The intelligent body CBest that the volume utilization of step 6. to t generation is maximum
temploying self-learning operator is optimized, and obtains the intelligent body CBest that the volume utilization in t+1 generation is maximum
t+1.
(6a) by the intelligent body CBest that the volume utilization in t generation is maximum
trecursive call mutation operator produces Agent Grid sL during self study, and this mutation operator is identical with the mutation operator in step 4;
(6b) set st as evolutionary generation during self study, make st=0;
(6c) random number rand and neighborhood are competed probability P o to compare, if rand<Po, then adopt neighborhood competition operator and mutation operator to be optimized Agent Grid sL during self study, obtain the intelligent body CBest that the volume utilization in t+1 generation is maximum
t+1, it is identical with mutation operator that this neighborhood competition operator and mutation operator and the neighborhood in step 4 compete operator;
(6d) by evolutionary generation st during self study compared with maximum evolutionary generation SMax, if st<SMax, st from adding 1, return step (6c), otherwise, export the intelligent body CBest that the volume utilization in t+1 generation is maximum
t+1.
Step 7. is by intelligent body CBest maximum for the volume utilization in t+1 generation
t+1energy energy and the maximum intelligent body Best of the global volume utilization factor in front t generation
tenergy energy compare, if energy (CBest
t+1) >energy (Best
t), then perform step 8, otherwise, perform step 9.
Step 8. is intelligent body CBest maximum for the volume utilization in t+1 generation
t+1the intelligent body Best that before giving, the global volume utilization factor in t+1 generation is maximum
t+1, perform step 10.
Step 9. is intelligent body Best maximum for the global volume utilization factor in front t generation
tthe intelligent body Best that global volume utilization factor before giving in t+1 generation is maximum
t+1with the intelligent body CBest that the volume utilization in t+1 generation is maximum
t+1, perform step 10.
Step 10. judges whether t reaches maximum evolutionary generation Max, if t<Max, making t from adding 1, returning step 4, otherwise, export the intelligent body Best that final global volume utilization factor is maximum
t+1, as vanning result, wherein comprise the sequence number of chest, rotary mode, Move Mode, size, coordinate points and maximum volume utilization factor.
Effect of the present invention can be verified by following emulation experiment:
1. test running environment and condition setting
The environment that experiment runs: processor is Intel (R) Core (TM) i3CPU 550@3.2GHz 3.19GHz, inside save as 3.05GB, hard disk is 1T, operating system is Microsoft windows XP Professional 2002, and programmed environment is visual c++ 6.0.
Experiment condition is arranged: according to the method producing data random in existing document, produce the data of 20 and 60 chests.The size L of Agent Grid in experiment
size× L
sizebe designed to 8 × 8, maximum evolutionary generation Max is 1000, Agent Grid size sL during self study
size× sL
sizebe designed to 4 × 4, the maximum evolutionary generation SMax of intelligent body during self study is 20, and neighborhood competition operator probability P o is 0.2, and neighborhood intersection probability P c is 0.8, and mutation probability Pm is 0.05.The present invention considers the randomness of algorithm, solve often organize data time, rerun 10 times.
2. experiment content and interpretation of result
Emulation experiment 1, carry out placement emulation with the present invention to the chest of 20 under different rotary state, result is as Fig. 5, and wherein, Fig. 5 a represents the optimal placement result that 20 chests emulate when not having rotation status; Fig. 5 b represents that the optimal placement result that 20 chests emulate when 2 kinds of rotation status, Fig. 5 c represent the optimal placement result that 20 chests emulate when 6 kinds of rotation status.
As can be seen from Fig. 5 a, Fig. 5 b and Fig. 5 c, 20 chests are when 2 kinds of rotation status and when not having a rotation status, and container can produce residual volume space, and 20 chests are when 6 kinds of rotation status, volume of a container space is occupied completely by chest, obtains the volume utilization of 100%.
Emulation experiment 2, with the present invention, placement emulation is carried out to the chest of 60 under different rotary state, result is as Fig. 6, wherein Fig. 6 a represents the optimal placement result that 60 chests emulate when not having rotation status, Fig. 6 b represents that the optimal placement result that 60 chests emulate when 2 kinds of rotation status, Fig. 6 c represent the optimal placement result that 60 chests emulate when 6 kinds of rotation status.
As can be seen from Fig. 6 a, Fig. 6 b and Fig. 6 c, the residual volume space of container is little, and volume of a container utilization factor is large.
Emulation experiment 3, the maximum volume utilization factor of HGA algorithm, HGAI algorithm, DBLF-GA algorithm gained in the maximum volume utilization factor of gained of the present invention and prior art contrasted, comparing result is as shown in table 1.
The Comparative result table of table 1 four kinds of methods
Result is not provided in "------" expression document
As can be seen from Table 1, volume utilization of the present invention is better than existing algorithm when 20 chests, and 60 chests are not when rotating, and volume utilization is higher than existing algorithm, and visible the present invention can obtain high volume utilization.
To sum up, when adopting the present invention to place chest, high volume utilization can be obtained, demonstrate validity of the present invention further.
Claims (5)
1., based on a packing method for three-dimensional Move Mode sequence and multi-Agent Genetic Algorithm, comprise the steps:
(1) setting parameter:
If Po is neighborhood competition probability, Pc is crossover probability, and Pm is mutation probability; M is the value of Move Mode, and value is 0,1,2; L
size× L
sizefor the sizing grid of multiple agent, t be more than or equal to 0 integer, represent t generation, Max is maximum evolutionary generation, L
trepresent that t is for Agent Grid, L
t+1/3and L
t+2/3l
tand L
t+1between centre for Agent Grid, Best
tl
0, L
1..., L
tthe intelligent body that middle global volume utilization factor is maximum, CBest
tl
tthe intelligent body that middle volume utilization is maximum; SL
size× sL
sizefor the sizing grid of multiple agent during self study, SMax is the maximum evolutionary generation of self study, Agent Grid when sL is self study; L
i,jexpression is in the i-th row of Agent Grid, the intelligent body of jth row, energy (L
i,j) represent this intelligent body L
i,jenergy; M is the chest number in each intelligent body, M be more than or equal to 1 integer, represent the wide, high, long of container respectively with W, H, L; RX, RY, RZ represent the wide, high, long of minimum cube envelope respectively; Rand be random produce 0 to 1 between real number;
(2) initialization Agent Grid L
0, upgrade the intelligent body Best that initial global volume utilization factor is maximum
0, make t=0;
(3) recursive call is decoded to each intelligent body based on the algorithm of three-dimensional Move Mode sequence, calculates the volume utilization of each intelligent body, upgrades the intelligent body Best that initial global volume utilization factor is maximum
0;
(4) to Agent Grid L
tin each intelligent body, adopt neighborhood competition operator carry out first time optimize, obtain first time optimization after Agent Grid L
t+1/3;
(5) the Agent Grid L after first time being optimized
t+1/3in each intelligent body, if rand<Pc, then adopt neighborhood crossover operator to carry out second time optimization, obtain the Agent Grid L after second time optimization
t+2/3;
(6) the Agent Grid L after second time being optimized
t+2/3in each intelligent body, if rand<Pm, then adopt mutation operator carry out third time optimize, obtain third time optimization after Agent Grid L
t+1;
(7) the Agent Grid L after optimizing from third time
t+1in find out the maximum intelligent body CBest of the volume utilization in t generation
t;
(8) to the intelligent body CBest that the volume utilization in t generation is maximum
temploying self-learning operator is optimized, and obtains the intelligent body CBest that the volume utilization in t+1 generation is maximum
t+1;
(9) by intelligent body CBest maximum for the volume utilization in t+1 generation
t+1energy energy and the maximum intelligent body Best of the global volume utilization factor in front t generation
tenergy energy compare, if energy (CBest
t+1) >energy (Best
t), then perform step (10), otherwise perform step (11);
(10) intelligent body CBest maximum for the volume utilization in t+1 generation
t+1the intelligent body Best that before giving, the global volume utilization factor in t+1 generation is maximum
t+1, perform step (12);
(11) intelligent body Best maximum for the global volume utilization factor in front t generation
tthe intelligent body Best that global volume utilization factor before giving in t+1 generation is maximum
t+1with the intelligent body CBest that the volume utilization in t+1 generation is maximum
t+1, perform step (12);
(12) judge whether t reaches maximum evolutionary generation Max, if t<Max, making t from adding 1, returning step (4), otherwise, export the intelligent body Best that final global volume utilization factor is maximum
t+1, as vanning result, wherein comprise the sequence number of chest, rotary mode, Move Mode, size, coordinate points and maximum volume utilization factor.
2. the packing method based on three-dimensional Move Mode sequence and multi-Agent Genetic Algorithm according to claim 1, it is characterized in that: each intelligent body is decoded with the algorithm based on three-dimensional Move Mode sequence described in step (3), carry out as follows:
(3a) length of initialization M chest, random generation Move Mode m, makes the initial coordinate point of all chests for (-1 ,-1 ,-1); If temp is a number produced at random, value is 0 or 1; If k is counter;
(3b) coordinate points of the 0th chest be set to (0,0,0), wide RX, high RY, the long RZ of initialization minimum cube envelope, make k=1;
(3c) judge whether to meet k<M, if meet, then perform step (3d), otherwise perform step (3r);
(3d) according to the value of Move Mode m, determine subsequent operation: if m=0, as temp=0, perform step (3e), as temp=1, perform step (3g); If m=1, as temp=0, perform step (3i), as temp=1, perform step (3k); If m=2, as temp=0, perform step (3m), as temp=1, perform step (3o);
(3e) moved down respectively by chest and find out upper covering, front covering is found out in movement backward, is moved to the left and finds out right covering; (3f) judge that can chest also move again, if can, then return step (3e), otherwise perform step (3q);
(3g) moved down respectively by chest and find out upper covering, be moved to the left and find out right covering, front covering is found out in movement backward;
(3h) judge that can chest also move again, if can, then return step (3g), otherwise perform step (3q);
(3i) chest is found out right covering respectively to moving left, front covering is found out in movement backward, moves down and finds out covering;
(3j) judge that can chest also move again, if can, then return step (3i), otherwise perform step (3q);
(3k) chest being found out right covering respectively to moving left, moving down and finding out covering, front covering is found out in movement backward;
(3l) judge that can chest also move again, if can, then return step (3k), otherwise perform step (3q);
(3m) chest is found out front covering respectively to rear movement, be moved to the left and find out right covering, move down and find out covering;
(3n) judge that can chest also move again, if can, then return step (3m), otherwise perform step (3q);
(3o) chest is found out front covering respectively to rear movement, move down and find out covering, be moved to the left and find out right covering;
(3p) judge that can chest also move again, if can, then return step (3o), otherwise perform step (3q);
(3q) judge whether chest loads in container completely, and if so, k is from adding 1, returns step (3c), otherwise, upgrading the wide RX of minimum cube envelope, high RY, long RZ, k from adding 1, returning step (3c);
(3r) decode procedure of vanning is terminated.
3. the packing method based on three-dimensional Move Mode sequence and multi-Agent Genetic Algorithm according to claim 1, is characterized in that: step (5) described neighborhood crossover operator carries out second time optimization to each intelligent body, carries out according to the following rules:
For intelligent body L
i,jthe order of middle chest: Stochastic choice point, by intelligent body L
i,jchest sequence number at that point and the maximum intelligent body L of its neighboring region energy
i,j maxchest sequence number at that point exchanges;
For intelligent body L
i,jthe rotary mode of middle chest and Move Mode adopt the method for two-point crossover: Stochastic choice two points, by intelligent body L
i,jthe intelligent body L that rotary mode between these 2 or Move Mode are maximum with its neighboring region energy respectively
i,j maxrotary mode between these 2 or Move Mode exchange.
4. the packing method based on three-dimensional Move Mode sequence and multi-Agent Genetic Algorithm according to claim 1, is characterized in that: carrying out third time with mutation operator to each intelligent body and optimize described in step (6), carries out according to the following rules:
For intelligent body L
i,jthe order of middle chest: Stochastic choice two point, exchanges intelligent body L
i,jin the sequence number of this chest of 2;
For intelligent body L
i,jthe rotary mode of middle chest: Stochastic choice a bit, selects an integer to replace intelligent body L at random from the integer between 0 to 5
i,jat the rotary mode of this point;
For intelligent body L
i,jthe Move Mode of middle chest: a bit, Stochastic choice integer replaces intelligent body L to Stochastic choice among 0,1,2
i,jat the Move Mode of this point.
5. the packing method based on three-dimensional Move Mode sequence and multi-Agent Genetic Algorithm according to claim 1, is characterized in that: the intelligent body CBest maximum with the volume utilization of self-learning operator to t generation described in step (8)
tbe optimized, carry out according to the following rules:
(8a) by the intelligent body CBest that the volume utilization in t generation is maximum
trecursive call mutation operator produces Agent Grid sL during self study, and this mutation operator is identical with the mutation operator in step (6);
(8b) set st as evolutionary generation during self study, make st=0;
(8c) random number rand and neighborhood are competed probability P o to compare, if rand<Po, then adopt neighborhood competition operator and mutation operator to be optimized Agent Grid sL during self study, obtain the intelligent body CBest that the volume utilization in t+1 generation is maximum
t+1, it is identical that this neighborhood competition operator and the neighborhood in step (4) compete operator, and this mutation operator is identical with the mutation operator in step (6);
(8d) by evolutionary generation st during self study compared with maximum evolutionary generation SMax, if st<SMax, st from adding 1, return step (8c), otherwise, export the intelligent body CBest that the volume utilization in t+1 generation is maximum
t+1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410799112.8A CN104504469A (en) | 2014-12-19 | 2014-12-19 | Box loading method based on three-dimensional moving mode sequence and multi-intelligent-agent genetic algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410799112.8A CN104504469A (en) | 2014-12-19 | 2014-12-19 | Box loading method based on three-dimensional moving mode sequence and multi-intelligent-agent genetic algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104504469A true CN104504469A (en) | 2015-04-08 |
Family
ID=52945864
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410799112.8A Pending CN104504469A (en) | 2014-12-19 | 2014-12-19 | Box loading method based on three-dimensional moving mode sequence and multi-intelligent-agent genetic algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104504469A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110490371A (en) * | 2019-07-26 | 2019-11-22 | 西安理工大学 | Packing method based on genetic algorithm |
| CN111079987A (en) * | 2019-11-28 | 2020-04-28 | 电子科技大学 | Semiconductor workshop production scheduling method based on genetic algorithm |
| CN112150049A (en) * | 2019-06-27 | 2020-12-29 | 厦门链石网络科技股份有限公司 | Stone plate boxing method based on application software |
| CN114781014A (en) * | 2022-06-17 | 2022-07-22 | 腾讯科技（深圳）有限公司 | Object processing method, device, equipment and computer readable storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2004030413A (en) * | 2002-06-27 | 2004-01-29 | Fujitsu Ltd | Optimization processing apparatus |
| CN103020730A (en) * | 2012-11-01 | 2013-04-03 | 西安电子科技大学 | Resource-constrained project scheduling method based on multi-agent evolutionary algorithm |
-
2014
- 2014-12-19 CN CN201410799112.8A patent/CN104504469A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2004030413A (en) * | 2002-06-27 | 2004-01-29 | Fujitsu Ltd | Optimization processing apparatus |
| CN103020730A (en) * | 2012-11-01 | 2013-04-03 | 西安电子科技大学 | Resource-constrained project scheduling method based on multi-agent evolutionary algorithm |
Non-Patent Citations (1)
| Title |
|---|
| 朱园: "基于多智能体进化算法的布图方法及三维装箱方法", 《中国知网》 * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112150049A (en) * | 2019-06-27 | 2020-12-29 | 厦门链石网络科技股份有限公司 | Stone plate boxing method based on application software |
| CN112150049B (en) * | 2019-06-27 | 2023-01-17 | 厦门链石网络科技股份有限公司 | Stone plate boxing method based on application software |
| CN110490371A (en) * | 2019-07-26 | 2019-11-22 | 西安理工大学 | Packing method based on genetic algorithm |
| CN111079987A (en) * | 2019-11-28 | 2020-04-28 | 电子科技大学 | Semiconductor workshop production scheduling method based on genetic algorithm |
| CN114781014A (en) * | 2022-06-17 | 2022-07-22 | 腾讯科技（深圳）有限公司 | Object processing method, device, equipment and computer readable storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104504468A (en) | Three-dimensional box loading method based on three-dimensional moving mode sequence and memetic algorithm | |
| CN104504469A (en) | Box loading method based on three-dimensional moving mode sequence and multi-intelligent-agent genetic algorithm | |
| CN102426424B (en) | Visual emulation method of near-space flying vehicle based on distributed framework | |
| CN101630380A (en) | Job-shop scheduling method based on multi-population evolution mechanism | |
| CN107977756B (en) | Ternary tree planning calculation method for solving three-dimensional packing problem | |
| CN104200130B (en) | It is a kind of that the Advances in protein structure prediction assembled with fragment is exchanged based on tree construction copy | |
| CN109246176A (en) | Based on the multi-controller synchronous method and device of block chain in software defined network | |
| Liu et al. | A new evolutionary multi-objective algorithm to virtual machine placement in virtualized data center | |
| CN102855271A (en) | Storage and traceable management method for multi-version power grid model | |
| CN109447264B (en) | Virtual machine placement genetic optimization method based on VHAM-R model in cloud computing environment | |
| Shang et al. | Spatial scheduling optimization algorithm for block assembly in shipbuilding | |
| CN105874511B (en) | Analog machine | |
| CN109732593A (en) | A kind of far-end control method of robot, device and terminal device | |
| CN105589814A (en) | Fork type access method for Path ORAM | |
| CN105976306B (en) | It is a kind of based on energy consumption-error budget real-time drawing method | |
| CN101739713A (en) | Method for detecting parallel collision based on subdivision | |
| CN104346441B (en) | A kind of power distribution network information data dynamic integrity exchange method | |
| CN103714384A (en) | Integrated circuit floor planning method based on moving modal sequence and Memetic algorithm | |
| CN102411792A (en) | Multilevel dynamic loading-unloading method for virtual simulation scene | |
| CN103714211A (en) | Integrated circuit floorplanning method based on moving block sequence and multi-agent particle swarm | |
| CN110209354A (en) | For handling the method, apparatus, equipment and medium of data | |
| CN109636888A (en) | 2D special effect making method and device, electronic equipment, storage medium | |
| CN110274608A (en) | Intelligent paths planning method, device and computer readable storage medium | |
| CN116451592B (en) | Extensible intelligent shelter layout optimization method | |
| CN109451016A (en) | Data downloading management method, system and relevant device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20150408 |
|
| WD01 | Invention patent application deemed withdrawn after publication |