JP2002288624A - Route optimization program, route-optimizing method and route-optimizing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a route optimization technology, capable of deciding a plural number of optimizing routes, while keeping diversity of the routes. SOLUTION: This route-optimizing program is a program for carry out formation of a plural number of gene individuals to express the routes different from each other (S02), classification of the gene individuals in a plural number of individual groups according to similarity (S02), and formation of following generation gene individuals from the aforementioned gene individuals (S04, S05). The next generation gene individuals include naturally selected next generation gene individuals selected from the gene individuals according to the adaptability of the gene individuals. At this time, at least one of the gene individuals is selected from each of the individual groups as the naturally selected next generation gene individual.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] The present invention relates to a route optimization technique. The present invention particularly relates to genetic algorithms (GA:
The present invention relates to a route optimization technology for optimizing a route by using a genetic algorithm.

[0002]

2. Description of the Related Art It is desired that a moving body such as a vehicle or an aircraft move along an optimal route. Therefore, a route optimization method for searching for an optimal route has been studied.

As a route optimization method, a method using a genetic algorithm has been proposed. The route optimization method using a genetic algorithm has various advantages, such as that the amount of calculation does not increase exponentially, that a search space is wide, and that an evaluation function for evaluating a solution can be set flexibly.

Further, in the route optimization method using a genetic algorithm, a plurality of solution candidates can be determined. In the route optimization method using a genetic algorithm, a solution candidate is selected and determined from a large number of genes indicating a route. At this time, if a plurality of solution candidates are selected, a plurality of solution candidates can be determined. The characteristic that a plurality of solution candidates can be determined is difficult to realize by a route optimization method using another algorithm.

At this time, it is sometimes desired that the plurality of determined solution candidates are not similar to each other. For example, in a car navigation system, it is required that another solution candidate indicated when a user is not satisfied with one solution candidate is not similar to the one solution candidate.

A route optimization method using a genetic algorithm for determining a plurality of solution candidates that are not similar to each other is described in "Method for Determining Multiple Route Candidates in Route Search Using Genetic Algorithm", Jun Inagaki et al. Academic journal,
DI, Vo. J82-DI, No. 8, pp. 11
02-1111 (August 1999).

In the known route optimization method, a plurality of solution candidates are determined using a plurality of weighted evaluation functions. More specifically, a plurality of solution candidates are determined as follows. In the known route determination method, the search range is divided into n regions. Further, n evaluation functions weighted for each region and one evaluation function not weighted for any region are defined. That is, n + 1 evaluation functions are determined. All gene individuals are divided into n + 1 groups. n +
One group contains the same number of genes. Selection selection is performed using different evaluation functions for each group, and solution candidates are determined from each group. As a result, n + 1 solution candidates are determined. in this way,
In the known route determination method, n + 1 solution candidates that are not similar to each other are determined by performing selective selection using n + 1 evaluation functions having different weighting methods.

In the above-described known route optimization method, it is necessary to appropriately determine a method of weighting the evaluation function in order to maintain the diversity of the solution candidates. However, when determining a large number of solution candidates, it is necessary to determine a large number of evaluation functions while guaranteeing that the diversity of the solution candidates is maintained. Exists.

[0009]

SUMMARY OF THE INVENTION An object of the present invention is to provide another route optimizing technique capable of determining a plurality of optimized routes while maintaining the diversity of the routes.

Another object of the present invention is to provide a route optimization technique that can determine a plurality of optimization routes while maintaining the diversity of the routes, and that has less mathematical difficulty.

[0011]

Means for solving the problem are expressed as follows. The technical items appearing in the expression are appended with numbers, symbols, and the like in parentheses (). The numbers, symbols, and the like refer to technical matters constituting at least one of the embodiments of the present invention, particularly, technical matters expressed in the drawings corresponding to the embodiments. The reference numbers, reference symbols, and the like attached to are the same. Such reference numbers,
Reference symbols clarify the correspondence and bridging between the technical matters described in the claims and the technical matters in the embodiments. Such correspondence / bridge does not mean that the technical matters described in the claims are interpreted as being limited to the technical matters of the embodiments.

A first route optimization program according to the present invention
Is a computer-readable program. This
The route optimization program executes different routes (1-1
To 1-M) (G)₁~ G_M) Raw
And gene individuals (G^k ₁~ G^k _M), Each other
A plurality of individuals (S
₁~ S_N) And gene individuals (G^k ₁~ G
^k _M) To the next generation gene individuals (G^{k + 1} ₁~ G^{k + 1}
_M) Generate and execute a path optimization program
It is. Next-generation gene (G^{k + 1} ₁~ G^{k + 1} _M)
Gene individuals (G ^k ₁~ G^k _MGenetics based on fitness)
Offspring (G^k ₁~ G^k _M) Natural selection next generation selected from
Gene individuals (G^{k + 1} ₁~ G^{k + 1} _{M '})including. This
At the time of, the natural selection next generation gene individual (G^{k + 1} ₁~ G
^{k + 1} _{M '}) As the population (S₁~ S_N) Each
At least one gene individual (G^k ₁~ G^k _M) Is selected
Selected. Natural selection next generation gene individuals (G^{k + 1} ₁~ G
^{k + 1} _{M '}) As the population (S₁~ S_N) Each
At least one gene individual (G^k ₁~ G^k _M) Is selected
What is selected is a naturally selected next-generation gene individual (G^{k + 1}
₁~ G^{k + 1} _{M '}) To increase diversity.

At this time, the next-generation gene individuals (G ^{k + 1} ₁₁
ＧG ^{k + 1} _M ) further includes two gene individuals (G _i , G _i ) belonging to two different populations (S _{1 to} S _N ) (S _i , S _{i ′} , where i ≠ i ′). ') Crossed next-generation gene individuals (G ^{k + 1} _{M' + 1 to} G
^{k + 1} _M ). The crossed next-generation gene individuals (G ^{k + 1} _{M ′ + 1 to} G
^{k + 1} _M ) has great diversity.

The second route optimization program according to the present invention is a computer-readable program. The route optimization program executes different routes (1-1
To 1-M), and generating a plurality of gene individuals (G _{1 to} G _M ) according to the degree of similarity indicating the degree of similarity between the gene individuals (G ^k _{1 to} G ^k _M ). Population (S
_{1 to} S _N ) and gene individuals (G ^k _{1 to} G ^k
k _M ), the next-generation gene individuals (G ^{k + 1} _{1 to} G ^{k + 1)}
_M ) is a path optimization program that executes At this time, the next-generation gene individuals (G ^{k + 1} _{1 to} G
^{k + 1} _M ) are two different (S _{1 to} S _N ) populations (S _{1 to} S _N ).
_i , S _{i ′} , where i ≠ i ′), crossed next-generation gene individuals (G ^{k + 1} _{M ′ + 1 to} G ^{k + 1} _M ) generated by crossing two gene individuals (G _i , G _i ′) belonging to each. )
including.

In the first and second route optimization programs according to the present invention, the similarity may be determined based on a correlation with a predetermined reference gene.

[0016] In addition, the degree of similarity, gene individual _^(G k 1 ~G
^k _M ) may be determined based on the correlation between the path indicated by each of the predetermined reference paths and a predetermined reference path.

[0017] The first route optimization method according to the present invention comprises:
Plural genes representing different pathways (1-1 to 1-M)
Individual (G₁~ G_M) And gene individuals (G
^k ₁~ G^k _M) To the degree of similarity
A plurality of populations (S₁~ S_N), And
Gene individuals (G^k ₁~ G^k _M) From the next generation gene individuals
(G^{k + 1} ₁~ G^{k + 1} _M) To generate
I have. Next-generation gene (G^{k + 1} ₁~ G^{k + 1} _M) The remains
Transgenic individuals (G^k ₁~ G^k _M) Genes based on fitness
Individual (G^k ₁~ G^k _M) Natural selection next generation selected from
Gene individuals (G ^{k + 1} ₁~ G^{k + 1} _{M '})including. this
Sometimes, naturally selected next generation gene individuals (G^{k +1} ₁~ G
^{k + 1} _{M '}) As the population (S₁~ S_N) Each
At least one gene individual (G^k ₁~ G^k _M) Is selected
Selected.

The second route optimization method according to the present invention is a
Plural genes representing different pathways (1-1 to 1-M)
Individual (G₁~ G_M) And gene individuals (G
^k ₁~ G^k _M) To the degree of similarity
A plurality of populations (S₁~ S_N), And
Gene individuals (G^k ₁~ G^k _M) From the next generation gene individuals
(G^{k + 1} ₁~ G^{k + 1} _M) To generate
I have. At this time, the next generation gene individual (G^{k + 1} ₁~ G
^{k + 1} _M) Indicates the population (S₁~ S_N) Two different (S
_i, S_{i '}, Provided that the two belonging to i'i ')
Gene individuals (G _i, G_i′)
Difference next generation gene individual (G^{k + 1} _{M '+ 1}~ G^{k + 1} _M)
including.

[0019]

BRIEF DESCRIPTION OF THE DRAWINGS FIG.
A route optimization method according to an embodiment of the present invention will be described.

This embodiment is a route optimization method for optimizing routes of a plurality of aircraft. In the present embodiment, routes of a plurality of aircraft flying from position A to position B are determined. When determining the route of the airplane flying from the position A to the position B, if a route that is too close is determined, a near miss occurs, which is not preferable. Therefore, in the route optimization method of the present embodiment, routes of a plurality of aircraft are determined so as not to be close to each other.

FIG. 3 shows hardware 10 for executing the route optimizing method according to the present embodiment. Hardware 10
Has an input unit 11, a data storage unit 12, a CPU 13, a program recording unit 14, and an output unit 15.

The input unit 11 receives constraint condition information a indicating a constraint condition for optimizing a route. The input constraint condition information a is stored in the data storage unit 12.

The data storage unit 12 stores the above-described constraint information a and genetic information c. The genetic information c indicates a gene individual used when executing the route optimization. There are a plurality of gene individuals, each of which indicates a route of one aircraft.

The CPU 13 performs an arithmetic operation on the genetic information c stored in the data storage unit 12 using a genetic algorithm. The calculation performed by the CPU 13 is performed according to a route optimization program for executing the route optimization method according to the present embodiment. The route optimization program is recorded in the program recording unit 14.

The output unit 15 outputs the optimized path b calculated by the calculation performed by the CPU 13.

Next, a route optimization method according to the present embodiment will be described with reference to FIG. In the route optimization method of the present embodiment, M gene individuals each showing one different route are used. M is an integer of 2 or more.
In the following description, the gene individual of the k generation is G ^k ₁
ＧG ^k _M.

FIG. 1 shows a route optimization method according to this embodiment. First, constraint condition information a indicating a constraint condition is input (step S01). Examples of the constraint condition include a position A at which the required route starts and a position B at which the route ends.

[0028] Subsequently, the first generation of gene individual ^G _{1 1} ~G
¹ _M is generated (step S02). Gene individual ^G _{1 1} ^~G _{1 M} generated are stored as genetic information c. Gene individual ^G _{1 1} ^~G _{1 M,} each, show the path of one aircraft. Figure 2 is a gene the individual ^G ₁ 1 ~G ¹
Show the path represented by _M. Routes 1-1 to 1-M are:
Each is represented by a gene individual ^G _{1 1} ^~G _{1 M.}

[0029] Subsequently, in accordance with the genetic individual ^G _{1 1} ^~G _{1 M} mutual similarity, gene individual ^G _{1 1} ^~G _{1 M} are classified into N populations _S 1 to S _N (step S03 ). N
Is an integer of 2 or more, and the similarity refers to the degree to which gene individuals are similar to each other. The similarity can be defined based on the correlation in the gene space formed by the gene individuals, or can be defined based on the correlation in the phenotype space formed by the pathway represented by the gene individuals.

[0030] In this embodiment, a gene the individual ^G ₁ 1 ~G
¹ _M mutual similarity is determined based on the correlation between the reference gene G _STD indicating a predetermined reference path. In this case, gene individual ^G ₁ 1 ^~G _{1 M} mutual similarity, gene individual G
And ¹ ₁ ~G ¹ path indicated by the respective _M, can also be determined based on the correlation with a predetermined reference path.

[0031] Each of the populations _S 1 to S _N includes at least one of the genes the individual ^G _{1 1} ^~G _{1 M.} Gene individuals similar to each other belong to one of the individual groups S _{1 to} S _N.

[0032] Subsequently, the second generation of gene individual ^G ₂ 1 ~G
² _M is generated (step S04, step S0
5).

[0033] Gene individual ^G ₂ 1 ~G ² _M of the second generation of genetic individual ^G ₂ 1 ^~G _{2 M 'is} populations _S 1 to S _N
It is generated by performing a genetic operation for each (step S04). Here, M ′ is an integer of N or more and less than M.

The genetic manipulation includes selection, crossover, and mutation. Selection Selection For each population S ₁ to S _N is performed, from each of the populations S ₁ to S _N, at least one gene individuals are selected as a second-generation gene individuals. Selection Selection is made based on the genetic individual ^G _{1 1} ^~G _{1 M} each of fitness. The calculation of the fitness is performed using an evaluation function. Gene individual ^G _{1 1} ^~G _{1 M}
The value of the evaluation function is obtained for each of the chromosomes indicated by, and the value of the evaluation function is used as the fitness. By selecting selection for each population _S 1 to S _N is performed, diversity second generation genes individual ^G ₂ 1 ~G ² _{M 'is} maintained.

Further, the intersection is performed for each of the individual groups S _{1 to SN} . Two first generation gene individuals included in one of the populations S _{1 to SN} are crossed to generate two second generation gene individuals.

Furthermore, a second-generation gene individual is generated by mutating the first-generation gene individual.

Second-generation gene individuals G² ₁~ G² _MHorse
Chi no G individual² _{M '+ 1}~ G² _MIs the population S₁~
S_NTwo belonging to two different populations of
Generated by crossing gene individuals
S05).

More specifically, as described below,
Individual G² _{M '+ 1}~ G² _MIs generated. First, the population
S₁~ S_NTwo different populations S from among_i,
S_{i '}Is selected. Here, i ≠ i ’. In addition,
Body group S_iFrom the first generation gene individuals belonging to
Child individual G_iIs selected. Furthermore, the population S_{i '}Belongs to
One gene individual G from the first generation gene individuals_{i '}Is selected
Selected. Genetic individual G_iAnd gene individual G_{i '}But any
Crossed at the position, the gene individual G² _{M '+ 1}~ G ² _MHorse
The other two are generated. Genetic individual G² _{M '+ 1}~ G²
_MAre generated in the same manner.
Gene individuals G thus generated ² _{M '+ 1}~ G
² _MIs highly diverse.

[0039] By the above process, all the second-generation gene individuals ^G ₂ 1 ^~G _{2 M} is produced.

Subsequently, an end determination is made (step S).
06). The end determination is made based on whether an end condition is satisfied. Steps S03 to S05 are repeated until the end condition is satisfied. That is, until the end condition is satisfied, the genes individual k-th generation ^G _k 1 ^~G _{k M,} the same process of the (k + 1) th generation gene individual ^{G k
+1} ₁ ~G ^{k +} _{1 M} is, sequentially, continue to be generated.

In the present embodiment, a condition that gene individuals have been generated up to the k-th _MAX generation is used as the termination condition. Other conditions can be used as the end condition.

Subsequently, when the termination condition is satisfied, the optimization path b as a solution is determined, and the optimization path b is output to the output unit 1.
5 (step S07). From the gene individuals of the k _MAX generation generated last, according to the fitness,
A plurality of gene individuals are selected. The routes represented by the selected gene individuals are output as the optimized route b.

In the route optimization method according to the present embodiment, at least one k-th gene individual from each of the populations S _{1 to} S _N is selected as a (k + 1) -th gene individual,
The diversity of the (k + 1) th generation gene individuals is largely maintained.

Further, the k + 1 generation gene individual is generated by crossing two k generation gene individuals belonging respectively to two different population groups S _{1 to} S _N , and the diversity of the k + 1 generation gene individuals is generated. However, it is kept larger.

As described above, by maintaining the diversity of gene individuals, the routes of a plurality of aircraft are determined so as not to be close to each other.

Although the route of the aircraft is optimized in the embodiment, the route optimization method of the embodiment is adapted to the optimization of the route of another moving object. It is also possible.

The route optimizing method of the present embodiment can be used for optimizing the route of a truck in a physical distribution system.

Further, the route optimization method according to the present embodiment
It can be used to optimize the routes of multiple vehicles in a car navigation system.

[0049]

According to the present invention, there is provided a route optimization technique capable of determining a plurality of optimized routes while maintaining the diversity of the routes.

Further, according to the present invention, there is provided a route optimization technique capable of determining a plurality of optimization routes while maintaining the diversity of the routes, and having less mathematical difficulty.

[Brief description of the drawings]

FIG. 1 is a flowchart illustrating a route optimization method according to an embodiment of the present invention.

FIG. 2 shows a route determined by the route optimization method.

FIG. 3 shows hardware 10 for executing the route optimization method.

[Description of Signs] 1-1 to 1-M: Path 11: Input Unit 12: Data Storage Unit 13: CPU 14: Program Storage Unit 15: Output Unit

Continued on the front page (72) Inventor Shoji Sugioka 1200, Higashi Tanaka, Komaki City, Aichi Prefecture Mitsubishi Heavy Industries, Ltd. Nagoya Guidance Propulsion System Works

Claims (9)

[Claims] 1. Generating a plurality of gene individuals representing different pathways from each other; classifying the gene individuals into a plurality of individuals according to a similarity indicating a degree of similarity to each other; Generating a next-generation gene individual, wherein the next-generation gene is a naturally selected next-generation gene individual selected from the gene individual based on the fitness of the gene individual. A route optimization program, wherein at least one gene individual is selected from each of the populations as the naturally selected next-generation gene individual. 2. The route optimization program according to claim 1, wherein the next-generation gene individual further has a different population from the population.
A route optimization program including an intersecting next-generation gene individual generated by intersecting two of the gene individuals belonging to each of the two. 3. Generating a plurality of gene individuals representing different pathways from each other; classifying the gene individuals into a plurality of individuals according to a similarity indicating a degree of similarity to each other; Generating a next-generation gene individual, wherein the next-generation gene individual is generated by crossing two gene individuals belonging to two different individuals of the population. A route optimization program that includes crossed next-generation gene individuals. 4. The route optimization program according to claim 1, wherein the similarity is determined based on a correlation with a predetermined reference gene. 5. The route optimization program according to claim 1, wherein the similarity is based on a correlation between a route indicated by each of the gene individuals and a predetermined reference route. Route optimization program defined. 6. Generating a plurality of gene individuals representing different paths from each other, classifying the gene individuals into a plurality of individuals according to a degree of similarity indicating a degree of similarity to each other, and Generating a next-generation gene individual, wherein the next-generation gene includes a naturally selected next-generation gene individual selected from the gene individual based on the fitness of the gene individual, A route optimization method in which at least one gene individual is selected from each of the individual groups as an individual. 7. Generating a plurality of gene individuals representing different pathways from each other, classifying the gene individuals into a plurality of individuals according to a degree of similarity indicating a degree of similarity to each other, and Generating a next-generation gene individual, wherein the next-generation gene individual includes a crossed next-generation gene individual generated by crossing two of the gene individuals belonging to two different individuals of the population. Optimization method. 8. A first generating means for generating a plurality of gene individuals representing different routes, a classifying means for classifying the gene individuals into a plurality of individual groups according to a degree of similarity indicating a degree of similarity to each other, A second generation of a next-generation gene individual from the gene individual
Generation means, wherein the next generation gene includes a naturally selected next generation gene individual selected from the gene individual based on the fitness of the gene individual, and the second generation means includes the naturally selected next generation gene A route optimization device for selecting at least one gene individual from each of the individual groups as an individual. 9. A first generating means for generating a plurality of gene individuals representing different pathways, a classifying means for classifying the gene individuals into a plurality of individual groups according to a degree of similarity indicating a degree of similarity to each other, A second generation of a next-generation gene individual from the gene individual
A route optimization device comprising: a generation unit, wherein the next-generation gene individual includes a crossed next-generation gene individual generated by crossing two gene individuals belonging to two different individuals of the population.