CN102855328A - Parallel variable neighborhood search method - Google Patents

Parallel variable neighborhood search method Download PDF

Info

Publication number
CN102855328A
CN102855328A CN 201210348245 CN201210348245A CN102855328A CN 102855328 A CN102855328 A CN 102855328A CN 201210348245 CN201210348245 CN 201210348245 CN 201210348245 A CN201210348245 A CN 201210348245A CN 102855328 A CN102855328 A CN 102855328A
Authority
CN
China
Prior art keywords
solution
current
optimum solution
iterative search
iterations
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
CN 201210348245
Other languages
Chinese (zh)
Inventor
张俊
颜秉珩
崔赢
张现忠
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.)
Inspur Beijing Electronic Information Industry Co Ltd
Original Assignee
Inspur Beijing Electronic Information Industry Co Ltd
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 Inspur Beijing Electronic Information Industry Co Ltd filed Critical Inspur Beijing Electronic Information Industry Co Ltd
Priority to CN 201210348245 priority Critical patent/CN102855328A/en
Publication of CN102855328A publication Critical patent/CN102855328A/en
Pending legal-status Critical Current

Links

Images

Landscapes

  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

Disclosed are variable neighborhood search method and device. The variable neighborhood search method includes configuring a plurality of examples and a data set shared by a plurality of the examples; selecting a neighborhood structure for each example from a neighborhood structure set, and performing iterative search by a current solution base from the current neighborhood set to obtain a local optimal solution; judging whether the local optimal solution is superior to a current optimal solution of the example or not, if yes, updating a current solution and the current optimal solution of the example to the local optimal solution, and replacing a worst historical optimal solution with the current optimal solution of the example if the current optimal solution of the example is superior to the worst historical optimal solution stored in the data set; otherwise, randomly selecting a historical optimal solution from the data set as the current solution of the example after judging whether iterated times of the example reach a threshold value of iterative times before updating the current optimal solution of the example or not; and selecting an optimal solution in the data set as a global optimal solution after times of iterative search of all the examples reach total iterative times.

Description

A kind of parallel change neighborhood search method
Technical field
The present invention relates to the heuritic approach field, be specifically related to a kind of parallel change neighborhood search method.
Background technology
Becoming neighborhood search algorithm (Variable Neighborhood Search, VNS), is a kind of path type meta-heuristic algorithm, is proposed first in 1997 by Hansen and Mladenovic.Become the neighbour structure that the neighborhood search algorithm has comprised dynamic change, algorithm is more general, and degree of freedom is large, can design multiple modification for specific question, is mainly used in finding the solution combinatorial optimization problem and Global Optimal Problem.
The method for solving of combinatorial optimization problem can be divided into two classes: a class is exact algorithm, and this class algorithm carries out complete search to solution space, can guarantee to find the on a small scale optimum solution of problem; Another kind of is heuritic approach, and this class algorithm has been abandoned the integrality to Searching Resolution Space, therefore can not guarantee the final Global Optimality of separating.Owing to exist in a large number the NP-Hard problem in the combinatorial optimization problem, therefore the precise search algorithm often is difficult to realize when problem scale is larger, although and heuritic approach can not prove the optimality of solution, but can find out preferably approximate optimal solution with rational calculation cost under many circumstances, therefore becoming the neighborhood search algorithm is the main algorithm that solves combinatorial optimization problem.
The basic thought that becomes the neighborhood search algorithm is in search procedure, based on the locally optimal solution that has found, systematically changes its neighbour structure, expands the hunting zone with this, and then finds another locally optimal solution.Become the neighborhood search algorithm and comprised Local Search (Local Search), disturbance (Shaking) and three processes of neighborhood conversion, utilize Local Search to seek locally optimal solution, improve search precision, adopt perturbation process to jump out the scope of locally optimal solution, seek new locally optimal solution, so that locally optimal solution is drawn close to globally optimal solution, the neighborhood conversion provides a kind of iterative manner and stopping criterion.
Become the basic step of neighborhood search algorithm:
Step 1, selected initial solution are X 0, set initial parameter---the total iterations of neighbour structure number and algorithm, wherein the neighbour structure number is K Max, the total iterations of algorithm is T Max, set the neighbour structure set and be N k(k=1 ..., K Max); Current optimum solution X is set BestFor: X Best=X 0, current solution X CurBe X Cur=X 0, selected neighbour structure numbering k=1, iterative search number of times t=0.
Step 2, carry out interative computation, as t>T MaxThe time, output result of calculation stops computing; Otherwise, at X CurK neighborhood N kIn choose at random and separate X Shake, and to separating X ShakeCarry out Local Search, obtain locally optimal solution X LocalIf f is (X Local)<f (X Best), X then Best=X Local, X Cur=X Local, k=1 namely upgrades locally optimal solution; Otherwise, k=k mod K is set Max+ 1, t=t+1 repeats described step of carrying out interative computation.Wherein, function f is the valuation functions in the practical application.
Summary of the invention
The technical problem to be solved in the present invention provides a kind of parallel change neighborhood search method, and the method is compared with present change neighborhood search algorithm, can effectively expand solution room, increases to flee from the possibility of local optimum, and then obtains better globally optimal solution.
In order to address the above problem, the embodiment of the invention provides a kind of change neighborhood search method, it is characterized in that, this change neighborhood search method is the parallel neighborhood search method that becomes, and comprises the steps:
Step 101, configuration Multi-instance and a data set that supplies a plurality of described examples to share store a plurality of historical locally optimal solutions in this data acquisition;
Step 102, identical neighbour structure number, total iterations, neighbour structure set and iterations threshold value are set for a plurality of described examples;
Step 103, for each example, from the set of described neighbour structure, choose a neighbour structure as current neighbour structure for this example, the current solution of this example is carried out iterative search based on this current neighbour structure, obtains locally optimal solution; Judge whether this locally optimal solution is better than the current optimum solution of this example, if, then current solution and the current optimum solution with this example all is updated to this locally optimal solution, if the current optimum solution of this example is better than the poorest historical optimum solution of storing in the described data acquisition, then the current optimum solution with this example replaces this poorest historical optimum solution; If this locally optimal solution is not better than the current optimum solution of this example, after judging this example iterations having reached described iterations threshold value before its current optimum solution is not upgraded, from described data acquisition, choose at random a historical optimum solution as the current solution of this example, the iterative search number of times of this example is added 1, re-start iterative search based on this current neighbour structure, until the iterative search number of times reaches described total iterations;
After the iterative search number of times of step 104, all examples all reaches described total iterations, choose solution optimum in the described data acquisition as globally optimal solution.
Alternatively, described step 101 also comprises:
Historical optimum solution in the described data acquisition is sorted according to size.
Alternatively, described step 103 also comprises:
For this example arranges an initial solution, current optimum solution and current solution that this example is set is equal to this initial solution, and the current optimum solution that the iterative search number of times of this example and this example be set do not upgrade before iterations.
Alternatively, described step 103 also comprises:
Replace re-starting iterative search based on this current neighbour structure after this poorest historical optimum solution with the current optimum solution of this example, the iterative search number of times of this example is added 1, until the iterative search number of times reaches described total iterations.
Alternatively, also comprise in the described step 103:
If this locally optimal solution is not better than the current optimum solution of this example, when judging this example iterations not reached described iterations threshold value before its current optimum solution is not upgraded, before the current optimum solution of this example do not upgraded iterations add 1, after the iterative search number of times adds 1, from described neighbour structure set, choose new neighbour structure and proceed iterative search, until the iterative search number of times reaches described total iterations.
Alternatively, described step 103 also comprises:
When the poorest historical optimum solution of storing in Multi-instance needs simultaneously to described data acquisition is upgraded, then from the current optimum solution of this Multi-instance, select one of optimum and replace this poorest historical optimum solution.
Alternatively, in a plurality of described examples, the historical optimum solution that has at least an example to be based on the optimum of storing in the described data acquisition is carried out iterative search.
The embodiment of the invention also provides a kind of change neighborhood search device, comprising: configuration module, iterative search module, judge module, update module and globally optimal solution determination module, wherein:
Described configuration module is arranged to: configuration Multi-instance and a data set that supplies a plurality of described examples to share store a plurality of historical optimum solutions in this data acquisition; Also for a plurality of described examples identical neighbour structure number, total iterations, neighbour structure set and iterations threshold value are set;
Described iterative search module is arranged to: for each example, from described neighbour structure set, choose a neighbour structure as current neighbour structure for this example, the current solution of this example is carried out iterative search based on this current neighbour structure, obtain locally optimal solution; Also from described data acquisition, choose at random a historical optimum solution as the current solution of this example, the iterative search number of times of this example is added 1, re-start iterative search based on this current neighbour structure;
Described judge module is arranged to: for each example, after described iterative search module obtains the locally optimal solution of this example, judge whether locally optimal solution that this example iterative search obtains is better than the current optimum solution of this example, if then notify described update module to replace this current optimum solution with the locally optimal solution of this example; Whether the current optimum solution of also judging this example is better than the poorest historical optimum solution of storing in the described data acquisition, if then notify described update module to replace this poorest historical optimum solution with the current optimum solution of this example; After judging that also this locally optimal solution is not better than the current optimum solution of this example, after judging this example iterations having reached described iterations threshold value before its current optimum solution is not upgraded, notify described iterative search module from described data acquisition, to choose at random a historical optimum solution as the current solution of this example, the iterative search number of times of this example is added 1, re-start iterative search based on this current neighbour structure; After the iterative search number of times of judging a certain example reaches described total iterations, notify described iterative search module to stop search for this example; After the iterative search number of times of judging all examples all reaches described total iterations, notify described globally optimal solution determination module to determine globally optimal solution;
Described update module is arranged to: the locally optimal solution that obtains with the example iterative search replaces the current optimum solution of this example; Replace the poorest historical optimum solution of storing in the described data acquisition with the current optimum solution of this example;
Described globally optimal solution determination module is arranged to: choose solution optimum in the described data acquisition as globally optimal solution.
Alternatively, described configuration module also is arranged to:
For each example all arranges an initial solution, current optimum solution and current solution that this example is set is equal to this initial solution, and the current optimum solution that the iterative search number of times of this example and this example be set do not upgrade before iterations.
Alternatively, described configuration module is all different for the initial solution that each example arranges.
Alternatively, described iterative search module also is arranged to: after described update module replaces the poorest historical optimum solution with the current optimum solution of example, re-start iterative search based on this current neighbour structure, the iterative search number of times of this example is added 1.
Alternatively, described judge module also is arranged to: when the locally optimal solution that judgement example iterative search obtains is not better than the current optimum solution of this example, when judging this example iterations not reached described iterations threshold value before its current optimum solution is not upgraded, before notifying described iterative search module that the current optimum solution of this example is not upgraded iterations add 1, after the iterative search number of times adds 1, from described neighbour structure set, choose new neighbour structure and proceed iterative search;
Described iterative search module also is arranged to: before the current optimum solution of example is not upgraded iterations add 1, after the iterative search number of times adds 1, from the set of described neighbour structure, choose new neighbour structure and proceed iterative search.
Alternatively, described judge module also is arranged to: when the poorest historical optimum solution of storing in the judgement Multi-instance needs simultaneously to described data acquisition is upgraded, notify described update module to select one of optimum to replace historical optimum solution the poorest in the described data acquisition from the current optimum solution of this Multi-instance;
Described update module also is arranged to: select optimum one to replace historical optimum solution the poorest in the described data acquisition in the current optimum solution of the Multi-instance that simultaneously the poorest historical optimum solution of storing the described data acquisition is upgraded from needs.
The parallel change neighborhood search method of the embodiment of the invention can effectively be expanded solution room, increases to flee from the possibility of local optimum, and then obtains better globally optimal solution.
Description of drawings
Fig. 1 is the method flow synoptic diagram of the embodiment of the invention;
Fig. 2 is the change neighborhood search device synoptic diagram of the embodiment of the invention.
Embodiment
Below in conjunction with drawings and Examples technical scheme of the present invention is described in detail.
Need to prove that if do not conflict, each feature among the embodiment of the invention and the embodiment can mutually combine, all within protection scope of the present invention.In addition, although there is shown logical order in flow process, in some cases, can carry out step shown or that describe with the order that is different from herein.
A kind of parallel change neighborhood search method of the embodiment of the invention as shown in Figure 1, mainly comprises the steps:
Step 101, configuration Multi-instance and a data set that supplies a plurality of described examples to share store a plurality of historical optimum solutions in this data acquisition.
Wherein, historical optimum solution is that a plurality of described examples find in iterative search procedures.
Step 102, identical neighbour structure number, total iterations, neighbour structure set and iterations threshold value are set for a plurality of described examples;
Wherein, described iterations threshold value is the maximum iteration time that method allows optimum solution not upgrade.
Step 103, for each example, from the set of described neighbour structure, choose a neighbour structure as current neighbour structure for this example, the current solution of this example is carried out iterative search based on this current neighbour structure, obtains locally optimal solution; Judge whether this locally optimal solution is better than the current optimum solution of this example, if, then current solution and the current optimum solution with this example is updated to this locally optimal solution, if the current optimum solution of this example is better than the poorest historical optimum solution of storing in the described data acquisition, then the current optimum solution with this example replaces this poorest historical optimum solution; If this locally optimal solution is not better than the current optimum solution of this example, after judging this example iterations having reached described iterations threshold value before its current optimum solution is not upgraded, from described data acquisition, choose at random a historical optimum solution as the current solution of this example, re-start iterative search based on this current neighbour structure, the iterative search number of times of this example is added 1, until the iterative search number of times of this example reaches described total iterations;
Each example all possesses the independent ability of finding the solution, and is absorbed in local time when single instance, by with the message exchange of the data acquisition of sharing, realized mutual between example, thereby can effectively expand solution room, increase the possibility that example is fled from local optimum.
After the iterative search number of times of step 104, all examples all reaches described total iterations, choose solution optimum in the described data acquisition as globally optimal solution.
Alternatively, described step 101 also comprises: the historical optimum solution in the described data acquisition is sorted according to size;
Alternatively, described step 103 also comprises: for each example all arranges an initial solution, current optimum solution and current solution that this example is set is equal to this initial solution, and the current optimum solution that the iterative search number of times of this example and this example be set do not upgrade before iterations.
Alternatively, the initial solution of each example is all different.
Alternatively, the iterative search number of times that this example is set is 1, before the current optimum solution that this example is set is not upgraded iterations be 0.
Alternatively, described step 103 also comprises: replace after this poorest historical optimum solution with the current optimum solution of this example, re-start iterative search based on this current neighbour structure, the iterative search number of times of this example is added 1, until the iterative search number of times reaches described total iterations.
Alternatively, described step 103 also comprises: if this locally optimal solution is not better than the current optimum solution of this example, when judging this example iterations not reached described iterations threshold value before its current optimum solution is not upgraded, before the current optimum solution of this example do not upgraded iterations add 1, after the iterative search number of times adds 1, from described neighbour structure set, choose new neighbour structure and proceed iterative search, until the iterative search number of times reaches described total iterations.
Alternatively, described step 103 also comprises: when the poorest historical optimum solution of storing in simultaneously to described data acquisition when the Multi-instance needs is upgraded, then select optimum one to replace this poorest historical optimum solution from the current optimum solution of this Multi-instance.
Alternatively, in a plurality of described examples, the historical optimum solution that has at least an example to be based on the optimum of storing in the described data acquisition is carried out iterative search.
For Multi-instance identical neighbour structure number, total iterations and neighbour structure set are set in the above-mentioned example, carry out the parallel iteration search based on set neighbour structure number, total iterations and neighbour structure set, and in iterative search procedures separately, data are upgraded, and finish the mutual of information between each example by the data acquisition of sharing, thereby can effectively expand solution room, increase and flee from the possibility of local optimum, and then obtain better globally optimal solution.
A kind of change neighborhood search device of the embodiment of the invention as shown in Figure 2, comprising: configuration module 201, iterative search module 202, judge module 203, update module 204 and globally optimal solution determination module 205, wherein:
Described configuration module 201 is arranged to: configuration Multi-instance and a data set that supplies a plurality of described examples to share store a plurality of historical optimum solutions in this data acquisition; Also for a plurality of described examples identical neighbour structure number, total iterations, neighbour structure set and iterations threshold value are set;
Wherein, historical optimum solution is that a plurality of described examples find in iterative search procedures.
Described iterative search module 202 is arranged to: for each example, from described neighbour structure set, choose a neighbour structure as current neighbour structure for this example, the current solution of this example is carried out iterative search based on this current neighbour structure, obtain locally optimal solution; From described data acquisition, choose at random a historical optimum solution as the current solution of this example, the iterative search number of times of this example is added 1, re-start iterative search based on this current neighbour structure.
Described judge module 203 is arranged to: for each example, after described iterative search module 202 obtains the locally optimal solution of this example, judge whether locally optimal solution that this example iterative search obtains is better than the current optimum solution of this example, if then notify the locally optimal solution of described update module 204 these examples of usefulness to replace this current optimum solution; Whether the current optimum solution of also judging this example is better than the poorest historical optimum solution of storing in the described data acquisition, if then notify the current optimum solution of described update module 204 these examples of usefulness to replace this poorest historical optimum solution; After judging that also this locally optimal solution is not better than the current optimum solution of this example, after judging this example iterations having reached described iterations threshold value before its current optimum solution is not upgraded, notify described iterative search module 202 from described data acquisition, to choose at random a historical optimum solution as the current solution of this example, re-start iterative search based on this current neighbour structure, the iterative search number of times of this example is added 1; After the iterative search number of times of judging a certain example reaches described total iterations, notify described iterative search module 202 to stop iterative search for this example; After the iterative search number of times of judging all examples all reaches described total iterations, notify described globally optimal solution determination module 205 to determine globally optimal solutions.
Wherein, the maximum iteration time of iterations threshold value for allowing optimum solution not upgrade.
Described update module 204 is arranged to: the locally optimal solution that obtains with the example iterative search replaces the current optimum solution of this example; Replace the poorest historical optimum solution of storing in the described data acquisition with the current optimum solution of this example;
Described globally optimal solution determination module 205 is arranged to: choose solution optimum in the described data acquisition as globally optimal solution.
Alternatively, described configuration module 201 also is arranged to: for each example all arranges an initial solution, current optimum solution and current solution that this example is set is equal to this initial solution, and the current optimum solution that the iterative search number of times of this example and this example be set do not upgrade before iterations.
Alternatively, the iterative search number of times of described configuration module 201 these examples is set to 1, before the current optimum solution of this example is not upgraded iterations be set to 0.
Alternatively, described configuration module 201 is all different for the initial solution that each example arranges.
Alternatively, described iterative search module 202 also is arranged to: after the current optimum solution of described update module 204 usefulness examples replaces the poorest historical optimum solution, re-start iterative search based on this current neighbour structure, the iterative search number of times of this example is added 1.
Alternatively, described judge module 203 also is arranged to: when the locally optimal solution that judgement example iterative search obtains is not better than the current optimum solution of this example, when judging this example iterations not reached described iterations threshold value before its current optimum solution is not upgraded, before notifying described iterative search module 202 that the current optimum solution of this example is not upgraded iterations add 1, after the iterative search number of times adds 1, from described neighbour structure set, choose new neighbour structure and proceed iterative search;
Described iterative search module 202 also is arranged to: before the current optimum solution of example is not upgraded iterations add 1, after the iterative search number of times adds 1, from the set of described neighbour structure, choose new neighbour structure and proceed iterative search.
Alternatively, described judge module 203 also is arranged to: when the poorest historical optimum solution of storing in the judgement Multi-instance needs simultaneously to described data acquisition is upgraded, notify described update module 204 to select one of optimum to replace historical optimum solution the poorest in the described data acquisition from the current optimum solution of this Multi-instance;
Described update module 204 also is arranged to: select optimum one to replace historical optimum solution the poorest in the described data acquisition in the current optimum solution of the Multi-instance that simultaneously the poorest historical optimum solution of storing the described data acquisition is upgraded from needs.
Alternatively, in a plurality of described examples, the historical optimum solution that has at least an example to be based on the optimum of storing in the described data acquisition is carried out iterative search.
The below uses example with one of the present invention and further is illustrated.
Step 1, configuration Multi-instance and a data set D who shares for a plurality of described examples; Set the example number, be that all examples are set the initial parameters such as identical neighbour structure number, total iterations, neighbour structure set.
Wherein, store a plurality of historical optimum solutions in the data acquisition;
The example number is M, and the neighbour structure number is K Max, total iterations is T Max, the neighbour structure set is N k(k=1 ..., K Max);
Step 2, for each example an initial solution is set all, current optimum solution and current solution that this example is set be equal to this initial solution, and be all selected neighbour structures from the set of described neighbour structure of each example, iterations before the current optimum solution that also for each example iterative search number of times and this example is set is not all upgraded.
Suppose that m is any one in the Multi-instance, for example m (m=1 ..., the initial solution of M) setting is
Figure BDA00002154229600101
If the current optimum solution of this example For: Current solution
Figure BDA00002154229600104
For:
Figure BDA00002154229600105
Figure BDA00002154229600106
The neighbour structure selected for example m is numbered k m=1, the iterative search number of times of example m is t m=1, example m iterations i before its current optimum solution is not upgraded m=0.
The iterative search number of times i.e. the total degree crossed of iteration.
Step 3, initialization data set D and iterations threshold value Iter.
Wherein, iterations threshold value Iter is the maximum iteration time that algorithm allows optimum solution not upgrade.
Step 4, for example m (m=1 ..., M), carry out following operation:
Work as t m>T MaxThe time, example m stops computing;
Work as t mBe not more than T MaxThe time,
Figure BDA00002154229600111
K mNeighborhood
Figure BDA00002154229600112
In choose at random solution
Figure BDA00002154229600113
And to separating
Figure BDA00002154229600114
Carry out local iteration's search, obtain locally optimal solution
Figure BDA00002154229600115
If Then
Figure BDA00002154229600117
Figure BDA00002154229600118
If
Figure BDA00002154229600119
Be better than gathering the poorest solution among the D, the poorest solution that then will gather among the D is updated to
Figure BDA000021542296001110
k m=1, i m=0; If
Figure BDA000021542296001111
Be not less than
Figure BDA000021542296001112
And i m>Iter then selects a historical optimum solution at random from set D, give k m=1, i m=0; Otherwise, k=k mod K Max+ 1, i m=i m+ 1, t=t+1, repeating step four;
Other examples are all carried out the operation identical with example m, do not repeat them here.
After the iterative search number of times of step 5, all examples all reached described total iterations, the optimum solution among the data acquisition D was the globally optimal solution that algorithm finds.
One of ordinary skill in the art will appreciate that all or part of step in the said method can come the instruction related hardware to finish by program, described program can be stored in the computer-readable recording medium, such as ROM (read-only memory), disk or CD etc.Alternatively, all or part of step of above-described embodiment also can realize with one or more integrated circuit.Correspondingly, each the module/unit in above-described embodiment can adopt the form of hardware to realize, also can adopt the form of software function module to realize.The present invention is not restricted to the combination of the hardware and software of any particular form.
Certainly; the present invention also can have other various embodiments; in the situation that does not deviate from spirit of the present invention and essence thereof; those of ordinary skill in the art work as can make according to the present invention various corresponding changes and distortion, but these corresponding changes and distortion all should belong to the protection domain of claim of the present invention.

Claims (13)

1. one kind becomes the neighborhood search method, it is characterized in that, this change neighborhood search method is the parallel neighborhood search method that becomes, and comprises the steps:
Step 101, configuration Multi-instance and a data set that supplies a plurality of described examples to share store a plurality of historical locally optimal solutions in this data acquisition;
Step 102, identical neighbour structure number, total iterations, neighbour structure set and iterations threshold value are set for a plurality of described examples;
Step 103, for each example, from the set of described neighbour structure, choose a neighbour structure as current neighbour structure for this example, the current solution of this example is carried out iterative search based on this current neighbour structure, obtains locally optimal solution; Judge whether this locally optimal solution is better than the current optimum solution of this example, if, then current solution and the current optimum solution with this example all is updated to this locally optimal solution, if the current optimum solution of this example is better than the poorest historical optimum solution of storing in the described data acquisition, then the current optimum solution with this example replaces this poorest historical optimum solution; If this locally optimal solution is not better than the current optimum solution of this example, after judging this example iterations having reached described iterations threshold value before its current optimum solution is not upgraded, from described data acquisition, choose at random a historical optimum solution as the current solution of this example, the iterative search number of times of this example is added 1, re-start iterative search based on this current neighbour structure, until the iterative search number of times reaches described total iterations;
After the iterative search number of times of step 104, all examples all reaches described total iterations, choose solution optimum in the described data acquisition as globally optimal solution.
2. change neighborhood search method as claimed in claim 1 is characterized in that described step 101 also comprises:
Historical optimum solution in the described data acquisition is sorted according to size.
3. change neighborhood search method as claimed in claim 1 is characterized in that described step 103 also comprises:
For this example arranges an initial solution, current optimum solution and current solution that this example is set is equal to this initial solution, and the current optimum solution that the iterative search number of times of this example and this example be set do not upgrade before iterations.
4. change neighborhood search method as claimed in claim 1 is characterized in that described step 103 also comprises:
Replace re-starting iterative search based on this current neighbour structure after this poorest historical optimum solution with the current optimum solution of this example, the iterative search number of times of this example is added 1, until the iterative search number of times reaches described total iterations.
5. such as each described change neighborhood search method among the claim 1-4, it is characterized in that, also comprise in the described step 103:
If this locally optimal solution is not better than the current optimum solution of this example, when judging this example iterations not reached described iterations threshold value before its current optimum solution is not upgraded, before the current optimum solution of this example do not upgraded iterations add 1, after the iterative search number of times adds 1, from described neighbour structure set, choose new neighbour structure and proceed iterative search, until the iterative search number of times reaches described total iterations.
6. change neighborhood search method as claimed in claim 5 is characterized in that described step 103 also comprises:
When the poorest historical optimum solution of storing in Multi-instance needs simultaneously to described data acquisition is upgraded, then from the current optimum solution of this Multi-instance, select one of optimum and replace this poorest historical optimum solution.
7. change neighborhood search method as claimed in claim 1 is characterized in that, in a plurality of described examples, the historical optimum solution that has at least an example to be based on the optimum of storing in the described data acquisition is carried out iterative search.
8. one kind becomes the neighborhood search device, it is characterized in that, comprising: configuration module, iterative search module, judge module, update module and globally optimal solution determination module, wherein:
Described configuration module is arranged to: configuration Multi-instance and a data set that supplies a plurality of described examples to share store a plurality of historical optimum solutions in this data acquisition; Also for a plurality of described examples identical neighbour structure number, total iterations, neighbour structure set and iterations threshold value are set;
Described iterative search module is arranged to: for each example, from described neighbour structure set, choose a neighbour structure as current neighbour structure for this example, the current solution of this example is carried out iterative search based on this current neighbour structure, obtain locally optimal solution; Also from described data acquisition, choose at random a historical optimum solution as the current solution of this example, the iterative search number of times of this example is added 1, re-start iterative search based on this current neighbour structure;
Described judge module is arranged to: for each example, after described iterative search module obtains the locally optimal solution of this example, judge whether locally optimal solution that this example iterative search obtains is better than the current optimum solution of this example, if then notify described update module to replace this current optimum solution with the locally optimal solution of this example; Whether the current optimum solution of also judging this example is better than the poorest historical optimum solution of storing in the described data acquisition, if then notify described update module to replace this poorest historical optimum solution with the current optimum solution of this example; After judging that also this locally optimal solution is not better than the current optimum solution of this example, after judging this example iterations having reached described iterations threshold value before its current optimum solution is not upgraded, notify described iterative search module from described data acquisition, to choose at random a historical optimum solution as the current solution of this example, the iterative search number of times of this example is added 1, re-start iterative search based on this current neighbour structure; After the iterative search number of times of judging a certain example reaches described total iterations, notify described iterative search module to stop search for this example; After the iterative search number of times of judging all examples all reaches described total iterations, notify described globally optimal solution determination module to determine globally optimal solution;
Described update module is arranged to: the locally optimal solution that obtains with the example iterative search replaces the current optimum solution of this example; Replace the poorest historical optimum solution of storing in the described data acquisition with the current optimum solution of this example;
Described globally optimal solution determination module is arranged to: choose solution optimum in the described data acquisition as globally optimal solution.
9. change neighborhood search device as claimed in claim 8 is characterized in that described configuration module also is arranged to:
For each example all arranges an initial solution, current optimum solution and current solution that this example is set is equal to this initial solution, and the current optimum solution that the iterative search number of times of this example and this example be set do not upgrade before iterations.
10. change neighborhood search device as claimed in claim 9 is characterized in that, described configuration module is all different for the initial solution that each example arranges.
11. such as each described change neighborhood search device in the claim 8, it is characterized in that:
Described iterative search module also is arranged to: after described update module replaces the poorest historical optimum solution with the current optimum solution of example, re-start iterative search based on this current neighbour structure, the iterative search number of times of this example is added 1.
12. such as each described change neighborhood search device among the claim 8-11, it is characterized in that:
Described judge module also is arranged to: when the locally optimal solution that judgement example iterative search obtains is not better than the current optimum solution of this example, when judging this example iterations not reached described iterations threshold value before its current optimum solution is not upgraded, before notifying described iterative search module that the current optimum solution of this example is not upgraded iterations add 1, after the iterative search number of times adds 1, from described neighbour structure set, choose new neighbour structure and proceed iterative search;
Described iterative search module also is arranged to: before the current optimum solution of example is not upgraded iterations add 1, after the iterative search number of times adds 1, from the set of described neighbour structure, choose new neighbour structure and proceed iterative search.
13. such as each described change neighborhood search device among the claim 8-11, it is characterized in that:
Described judge module also is arranged to: when the poorest historical optimum solution of storing in the judgement Multi-instance needs simultaneously to described data acquisition is upgraded, notify described update module to select one of optimum to replace historical optimum solution the poorest in the described data acquisition from the current optimum solution of this Multi-instance;
Described update module also is arranged to: select optimum one to replace historical optimum solution the poorest in the described data acquisition in the current optimum solution of the Multi-instance that simultaneously the poorest historical optimum solution of storing the described data acquisition is upgraded from needs.
CN 201210348245 2012-09-18 2012-09-18 Parallel variable neighborhood search method Pending CN102855328A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201210348245 CN102855328A (en) 2012-09-18 2012-09-18 Parallel variable neighborhood search method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201210348245 CN102855328A (en) 2012-09-18 2012-09-18 Parallel variable neighborhood search method

Publications (1)

Publication Number Publication Date
CN102855328A true CN102855328A (en) 2013-01-02

Family

ID=47401916

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201210348245 Pending CN102855328A (en) 2012-09-18 2012-09-18 Parallel variable neighborhood search method

Country Status (1)

Country Link
CN (1) CN102855328A (en)

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103366021A (en) * 2013-08-07 2013-10-23 浪潮(北京)电子信息产业有限公司 Variable neighborhood search method and system on cloud computing platform
CN104751016A (en) * 2015-04-16 2015-07-01 大连大学 Variable neighborhood search based DNA label structuring method
CN105157712A (en) * 2015-08-18 2015-12-16 浙江工商大学 Vehicle routing planning method and planning system
CN108009013A (en) * 2017-12-25 2018-05-08 湖南大学 For a kind of parallel neighborhood search method of collaboration of separation constraint knapsack problem
CN112632723A (en) * 2020-12-22 2021-04-09 中国农业大学 Water pump blade parameter obtaining method and device and electronic equipment
CN114091717A (en) * 2021-01-11 2022-02-25 北京京东振世信息技术有限公司 Site selection method and device

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103366021A (en) * 2013-08-07 2013-10-23 浪潮(北京)电子信息产业有限公司 Variable neighborhood search method and system on cloud computing platform
CN104751016A (en) * 2015-04-16 2015-07-01 大连大学 Variable neighborhood search based DNA label structuring method
CN104751016B (en) * 2015-04-16 2018-03-27 大连大学 Building method based on the DNA labels for becoming neighborhood search
CN105157712A (en) * 2015-08-18 2015-12-16 浙江工商大学 Vehicle routing planning method and planning system
CN108009013A (en) * 2017-12-25 2018-05-08 湖南大学 For a kind of parallel neighborhood search method of collaboration of separation constraint knapsack problem
CN112632723A (en) * 2020-12-22 2021-04-09 中国农业大学 Water pump blade parameter obtaining method and device and electronic equipment
CN112632723B (en) * 2020-12-22 2024-04-16 中国农业大学 Water pump blade parameter acquisition method and device and electronic equipment
CN114091717A (en) * 2021-01-11 2022-02-25 北京京东振世信息技术有限公司 Site selection method and device

Similar Documents

Publication Publication Date Title
CN102855328A (en) Parallel variable neighborhood search method
CN103366021A (en) Variable neighborhood search method and system on cloud computing platform
CN107710193B (en) Data placement control for distributed computing environments
Pearce et al. Faster parallel traversal of scale free graphs at extreme scale with vertex delegates
Meyerhenke et al. Partitioning complex networks via size-constrained clustering
CN109522428B (en) External memory access method of graph computing system based on index positioning
CN106649391B (en) Method and device for processing graph data
CN102148871B (en) Storage resource scheduling method and device
KR102357863B1 (en) A memory access method and apparatus
Mohamadi et al. Learning automata-based algorithms for solving the target coverage problem in directional sensor networks
CN102880724A (en) Method and system for processing Hash collision
WO2016206578A1 (en) Region guided and change tolerant fast shortest path algorithm and graph preprocessing framework
Mohamadi et al. Solving target coverage problem using cover sets in wireless sensor networks based on learning automata
US9075533B2 (en) Binary tree storage method and system
JP6751376B2 (en) Optimal solution search method, optimal solution search program, and optimal solution search device
EP3059670A1 (en) Data storage method and storage server
Devismes et al. Optimal torus exploration by oblivious robots
CN101963510B (en) Rapid path planning method and enhancement method for random route map
Mansour et al. An iterative solution for the coverage and connectivity problem in wireless sensor network
US9116751B2 (en) Reconfigurable device, processing assignment method, processing arrangement method, information processing apparatus, and control method therefor
Dereniowski et al. Fast collaborative graph exploration
US20140089613A1 (en) Management of data elements of subgroups
US10198529B2 (en) Apparatus and method of processing graphic data using index based triangle listing
CN105515818B (en) The method and system of cyclic structure are split in a kind of network topology layout
CN111737347B (en) Method and device for sequentially segmenting data on Spark platform

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C02 Deemed withdrawal of patent application after publication (patent law 2001)
WD01 Invention patent application deemed withdrawn after publication

Application publication date: 20130102