JPH1027169A - Boring arrangement system using genetic algorithm - Google Patents

Abstract

PROBLEM TO BE SOLVED: To easily find the optimum arrangement of boring by generating a gene character string from specific boring candidate places around structures and determining the optimum boring position on the basis of the gene character string by using genetic algorithm. SOLUTION: An initial group is generated first (step 201). Then it is decided whether an end condition is met (step 202). When the end condition is not met (step 202), adaptivity is calculated for each gene (step 203). Once the adaptivity is calculated for each gene, genes are selected according to the adaptivity (step 204). Then a process called crossing which generates different genes by replacing parts of a character string of the selected genes (step 205). Further, a process named mutation which changes the character string of genes obtained in the step 205 with certain possibility is performed (step 206).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a boring arrangement system using a genetic algorithm.

[0002]

2. Description of the Related Art In general, survey drilling is performed to evaluate the safety of the ground such as an oil tank (hereinafter referred to as a tank). Survey drilling must meet the standards specified by a Cabinet Order. The conditions for the arrangement of survey bowling specified by a Cabinet Order are as follows.

(A) In principle, three or more borings are performed for one tank. (B) However, if the stratigraphy of the ground is uniform, multiple tanks are allowed to share survey drilling under the following conditions. -When the area covered by the triangle connecting the three boring points covers more than half the ground area of the tank, and the distance between the boring points is within 70 m.・ When there is overlap in the ground area of the tank (a range of 10m from the tank outer periphery)・ For one tank, one boring must be installed within the ground area of the tank.

FIG. 18 is a diagram showing an example of an optimal arrangement of boring. In FIG. 18, 101 is a tank, 103
Indicates a ground area of the tank, and 105 indicates a boring position.

[0005]

By the way, there is a demand for further minimizing the number of boring drills (boring layout optimization problem) while satisfying the criteria specified by the above-mentioned Cabinet Order.

However, if a human tries to optimize the boring arrangement manually, a complicated operation involving trial and error is required. If the number of tanks is large and the size and arrangement of the tanks are irregular, manual operations are required. It is very difficult to find the optimal solution.

The present invention has been made in view of such a problem, and an object of the present invention is to provide a boring arrangement system capable of easily finding an optimum arrangement of boring.

[0008]

SUMMARY OF THE INVENTION In order to achieve the above-mentioned object, the present invention is a system for determining an arrangement of a boring with respect to a plurality of structures in a certain area. A genetic character string is generated from the boring candidate site, and an optimal boring position is determined using a genetic algorithm based on the gene character string.

[0009]

Embodiments of the present invention will be described below in detail with reference to the drawings. FIG. 1 is a schematic configuration diagram of hardware of a boring arrangement system according to an embodiment of the present invention, and includes a computer 1, a display 3, a printer 5, a mouse 7, a workstation 9, and the like. The workstation 9 performs main calculations using a genetic algorithm, and the computer 1 performs input and output of data between the workstation 9 and a human.

FIG. 2 is a flowchart showing the outline of the processing of the genetic algorithm (GA) in the present embodiment, and mainly shows the processing of the workstation 9. FIG. 3 is a diagram showing the arrangement of the target tank 11.
1 is a tank, 13 is a ground area of the tank.

As shown in FIG. 2, first, an initial group is generated (step 201). The initial population indicates an initial set of genes, and the genes represent solution candidates of the boring arrangement by character strings.

FIG. 4 shows a character string in the tank 11. In the present embodiment, the circle indicating the tank 11 and the ground area 13 of the tank is divided into six equal parts, points [1] to [6] are defined around the tank 11, and a point A point [9] is determined from [7]. This point [1]
To [9] are boring arrangement candidate points.

Then, for example, points [1], [5], [7]
It is assumed that boring is to be performed, and the point at which boring is performed is set to "1", and the point at which boring is not performed is set to "0". Therefore, the character string for the tank 11 is “100
010100 ".

For example, when there are four tanks, a character string is defined for each tank. That is, a character string “010100100” is defined for the second tank, “001010001” is defined for the third tank, and “0000100000” is defined for the fourth tank. A combination of these four character strings is referred to as a gene.
Therefore, in this case, the gene is "1000101000
1010010000101000100001000
0 ", which is one solution candidate.

Next, the boring execution point is changed for the tank 11 shown in FIG.
0100 "and appropriate character strings for the other three tanks, for example," 0011001 ".
0001010001010010010000100
A gene such as "1001" is defined. Similarly, a predetermined number of genes are determined by appropriately determining the boring execution point, and an initial population of genes is generated.

In the above, the case of four tanks is taken as an example.
Similarly, genes are determined for the 18 tanks shown in FIG. 3, and an initial population is generated. In the case of 18 tanks, the gene is 9 × 18 bits.

Next, it is determined whether the termination condition is satisfied in FIG. 2 (step 202). The termination condition is
For example, it is repeated a predetermined number of times, or the following fitness value converges.

If the termination condition is not satisfied (step 202), the fitness FV (Fitness Valu
e) is calculated. The fitness is calculated by the following equation. FV = NB × WNB + VT / TK × WVT + BO / CP × WBO… (1) NB (No Boring Tanks): Number of tanks with no boring in the tank ground area VT (Violte Tanks): Half or more of the tank ground area TK (Tanks): Number of tanks BO (Borings): Number of boring CP (Candidate Points): Number of candidate boring locations W NB (NB Weight): Weight for NB W VT (VT Weight): For VT Weight W BO (BO Weight): Weight for BO Each weight represents the contribution to the fitness of each condition.

When the fitness is calculated for each gene in this way, a gene is selected according to the fitness (step 204).

The selection processing includes, for example, a roulette strategy. FIG. 5 is an explanatory diagram of the roulette strategy. There are four genes A, B, C, and D. A roulette is created according to the fitness, and an arrow is fired twice to select two genes.

Next, a process called crossover for replacing the character string portion of the selected gene to create a different gene is performed (step 205). FIG. 6 is an explanatory diagram of the crossover process. In the crossover process, at step 204, a child gene k is created from the selected two genes (paternal gene i, mother gene j). For example, if the father gene i is "00000111"
1 "(in order to avoid redundancy, the gene is described as 9 bits) and the mother gene j is" 11111 ".
0000 ".

Here, a character string "0101" called a mask
01010 ”is generated. The mask is a random character string of “0” and “1”. Child gene k has mask “1”
When the mask is "0", it is inherited from the parent gene i. In this way, a child gene is generated from the two genes selected in step 204 through a process called crossover. Step 204,
In the process of 205, the portion of the gene having higher fitness is more prosperous to the offspring.

Next, the gene obtained in step 205 is subjected to a mutation process that changes the character string of the gene at a certain probability (step 206).

FIG. 7 shows the mutation. As shown in FIG. 7, the position of the gene obtained in step 205 is appropriately determined by random numbers, and the bits are inverted. The mutation is generated with a low probability of about 1% to 5% with respect to the gene obtained in step 205.

Then, the process returns to step 202. And
The above processing is repeated, and when the termination condition is satisfied (step 202), excellent genes remain and a solution is obtained.
FIG. 8 is a diagram illustrating the boring position 19 finally determined by the above-described processing, and FIG. 9 is a diagram illustrating the boring position 20 when the boring is performed manually. FIG.
FIG. 5 is a diagram showing processing time and solution accuracy using the present embodiment. As shown in FIG. 10, according to the present embodiment,
An accurate solution can be obtained in a short time.

FIG. 11 is an explanatory diagram in the case of performing parallel processing using a plurality of computers. C1 is a process performed by the main computer, C2 is a process performed by the subordinate computer, and there may be a plurality of subordinate computers.

The main computer confirms the connection of the subordinate computer (step 1101) and starts calculation of a genetic algorithm (GA). Then, a part of the gene is transmitted to a slave computer (step 110).
2). The slave computer receives the gene (step 1106) and calculates the fitness (step 110).
7) The fitness is transmitted to the main computer (step 1108).

On the other hand, the main computer calculates the fitness of the remaining genes (step 1103), and receives the fitness sent from the subordinate computer (step 1104). Then, the main computer performs processes such as selection, crossover, and selection (step 1105).
After repeating the calculation of the genetic algorithm for the generation loop, the result is output (step 1109).

In the genetic algorithm, the fitness calculation occupies most of the calculation time. Therefore, by dividing and parallelizing this processing, the calculation time can be reduced by 20 to 30%.

Further, in this embodiment, fine adjustment can be performed. As described above, after the calculation of the genetic algorithm is performed and the result is plotted, fine adjustment is performed on the result. FIG. 12 and FIG. 13 are explanatory diagrams in the case of adding boring. In FIG. 12, 21-1, 21-
2, 21-3 are tanks, 23-1, 23-2, 2
3-3 is a boring position. When newly boring to the point 23-4, if the point 23-4 on the screen is clicked, the boring is performed at the position of the point 23-4 as shown in FIG.

FIGS. 14 and 15 show the deletion of a boring point. In FIG. 14, reference numeral 31 denotes a tank, reference numeral 33 denotes a boring point, and when deleting the boring point 33-5, the point 33-5 is set. By clicking and inputting a command such as deletion, the point 33-5 is deleted as shown in FIG.

FIGS. 16 and 17 show the movement of the boring point. In FIG. 16, reference numeral 41 denotes a tank, and reference numeral 43 denotes a boring point. When the boring point 43-3 is moved, the user clicks on the point 43-3 on the screen and inputs a command such as movement, so that the point 43-3 is displayed as shown in FIG.
Go to In the present invention, a genetic algorithm other than the genetic algorithm shown in the present embodiment may be used. For example, the selection process may use a selection method other than the roulette strategy. Further, the termination conditions and the like may be changed as appropriate. In addition, the present invention is not limited to existing oil tanks and the like,
It can also be applied to oil tanks and the like to be constructed.

[0033]

As described above, according to the present invention, it is possible to easily determine the optimum arrangement of boring.

[Brief description of the drawings]

FIG. 1 is a schematic configuration diagram of hardware of a boring arrangement system according to an embodiment;

FIG. 2 is a flowchart illustrating a schematic process according to the embodiment;

FIG. 3 is a diagram showing an arrangement of a target tank 11;

FIG. 4 is a diagram showing genes for tank 11.

FIG. 5 is a diagram showing a roulette strategy.

FIG. 6 is a diagram showing crossover processing;

FIG. 7 is a diagram showing a mutation process.

FIG. 8 is a layout diagram of the boring obtained in the present embodiment.

FIG. 9 is a layout diagram of boring performed manually.

FIG. 10 is a diagram showing processing time and accuracy of a solution according to the present embodiment.

FIG. 11 is a flowchart in the case of performing parallel calculation.

FIG. 12 is an explanatory diagram of adding a boring point;

FIG. 13 is an explanatory diagram of adding a boring point.

FIG. 14 is an explanatory view of deleting a boring point.

FIG. 15 is an explanatory diagram of deleting a boring point;

FIG. 16 is an explanatory diagram of a movement of a boring point.

FIG. 17 is an explanatory diagram of a movement of a boring point.

FIG. 18 is a diagram showing an optimum arrangement of boring;

[Explanation of symbols]

 1 Computer 3 Display 5 Printer 7 Mouse 9 Workstation 11 Tank 13 Ground area of tank

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Kenji Kato Kashima Construction Co., Ltd. 1-2-7 Moto-Akasaka, Minato-ku, Tokyo

Claims (4)

[Claims] 1. A system for determining an arrangement of boring for a plurality of structures in a certain area, comprising: generating a gene character string from a predetermined boring candidate site around the plurality of structures; A boring arrangement system using a genetic algorithm, characterized in that an optimal boring position is determined using a genetic algorithm based on the boring. 2. Genetic character strings are generated by setting a point where a boring is performed and a point where a boring is not performed as “0” among candidate boring sites around a structure, and generating a character string. 2. The boring arrangement system using a genetic algorithm according to claim 1, wherein a character string generated for an object is combined, and a plurality of said gene character strings are generated in an initial stage. 3. The method according to claim 1, wherein the genetic algorithm selects, crosses, or mutates a predetermined gene from among a plurality of genes to select the gene.
A boring arrangement system using the described genetic algorithm. 4. The system of claim 1, wherein the structure is an oil tank.