Information processing method and information processing apparatus

Abstract

To efficiently perform stacking.SOLUTION: An information processing method to be executed by a computer of an arrangement plan creation system for creating an arrangement plan as to which floor a plurality of areas are arranged among a plurality of floors in a building includes steps of: creating a plurality of initial arrangement plans at random; generating an optimum arrangement plan by optimizing the plurality of initial arrangement plans on the basis of at least the areas of the floors, occupied areas of the areas, and an evaluation value calculated by using congeniality between different areas; generating a plurality of classes by clustering the plurality of optimum arrangement plans; extracting a feature amount of a class on the basis of a common item between the optimum arrangement plans included in the class for each of the plurality of classes; and mutually associating the feature amount of the class with the optimum arrangement plan included in the class to output them.SELECTED DRAWING: Figure 4

Description

[[Technical Field]]

[0001] The present invention relates to an information processing method and an information processing apparatus. [[Background Art]]

[0002] In the basic planning stage of a building, there is a process called stacking for deciding on which floor of the building a plurality of areas should be arranged. As a method for efficiently performing stacking, a method like that disclosed in Non-Patent Document 1 has been proposed. The method described in Non-Patent Document 1 aims to obtain an optimal layout plan by optimizing an initial layout plan using a genetic algorithm based on evaluation values based on factors such as floor area and affinity between areas. [[Prior Art Documents]] [[Patent Documents]]

[0003] [[Non-Patent Document 1]] Hiromasa Akagi, Shigeyoshi Tanaka, Katsuhiko Watanabe, "Application of Genetic Algorithm to Stacking Problems", Abstracts of Academic Lectures at the Annual Meeting of the Architectural Institute of Japan, Vol. 1997, 1997 [[Summary of the Invention]] [[Problem to be Solved by the Invention]]

[0004] In stacking, it is necessary to determine a layout plan in consideration of not only affinity between areas but also various other elements (viewpoints). However, it is difficult to reflect all elements to be considered for determining a layout plan in the evaluation value. For this reason, there is a need to search for a better layout plan by generating a plurality of layout plans with different characteristics and comparing them. Since only one layout plan can be obtained with conventional methods, designers have to execute the conventional method multiple times to create a plurality of layout plans, which requires a large amount of man-hours.

[0005] An object of the present invention is to efficiently perform stacking. [[Means for Solving the Problem]]

[0006] This invention relates to multiple areas within one or more buildings Located Among the multiple floors of An information processing method performed by a computer in a layout plan generation system for creating layout plans for which floors to place items on, wherein multiple initial layout plans are randomly generated. 1 Based on the steps and evaluation values ​​obtained using at least the floor area, the occupied area of ​​each area, and the proximity between different areas, multiple initial layout options are selected. one Generate the optimal layout plan 2nd Steps and The first and second steps are repeated while changing the initial seed value of the random numbers to generate multiple optimal placement options. The method includes the steps of: clustering multiple optimal placement options to generate multiple classes; extracting features for each of the multiple classes based on commonalities among the optimal placement options contained within the class; and outputting the correspondence between the class features and the optimal placement options contained within the class.

[0007] This invention relates to multiple areas within one or more buildings Located Among the multiple floors of An information processing device that generates layout plans for which floor to place items on, and randomly generates multiple initial layout plans. Execute the first process. The initial plan generation unit generates multiple initial layout plans based on evaluation values ​​obtained using at least the floor area, the occupied area of ​​each area, and the proximity between different areas. From one Generate the optimal layout plan Execute the second process. The system comprises an optimization processing unit, a class generation unit that generates multiple classes by clustering multiple optimal placement options, a feature extraction unit that extracts feature quantities for each of the multiple classes based on commonalities among the optimal placement options contained within the class, and an output unit that outputs the feature quantities of the class and the optimal placement options contained within the class in correspondence with each other. The system then repeatedly performs the first processing by the initial design generation unit and the second processing by the optimization unit, while changing the initial seed value of the random numbers, to generate multiple optimal placement designs. . [Effects of the Invention]

[0008] According to the present invention, stacking can be performed efficiently. [Brief explanation of the drawing]

[0009] [Figure 1] It is a block diagram showing a configuration of an information processing apparatus according to an embodiment of the present invention. [Figure 2] It is a schematic diagram showing an example of stacking in an embodiment of the present invention. [Figure 3] It is a map showing an example of affinity in an embodiment of the present invention. [Figure 4] It is a flowchart showing an information processing method according to an embodiment of the present invention. [Figure 5] It is a schematic diagram showing an outline of an optimization method according to an embodiment of the present invention. [Figure 6] It is a flowchart showing an optimal arrangement plan generation method of an information processing method according to an embodiment of the present invention. [Figure 7] It is a diagram showing an example of crossover in the information processing method according to an embodiment of the present invention. [Figure 8] It is a diagram showing an example of mutation in the information processing method according to an embodiment of the present invention, wherein (a) shows a first mutation and (b) shows a second mutation. [Figure 9] It is a flowchart showing a clustering method of the information processing method according to an embodiment of the present invention. [Figure 10] It is a diagram showing a clustering method of the information processing method according to an embodiment of the present invention, wherein (a) is a diagram schematically representing classes corresponding to respective optimal arrangement plans on two-dimensional coordinates, and (b) is a diagram showing a state where class 2 and class 3 are integrated. [Figure 11] It is a diagram showing a clustering method of the information processing method according to an embodiment of the present invention, and is a dendrogram showing classification. [Figure 12] It is a conceptual diagram showing a method for extracting feature amounts in the information processing method according to an embodiment of the present invention. DESCRIPTION OF EMBODIMENTS

[0010] Hereinafter, an arrangement plan creation system including an information processing apparatus according to an embodiment of the present invention and an information processing method will be described with reference to the drawings.

[0011] The information processing apparatus 100 and the information processing method of the present embodiment shown in FIG. 1 perform so-called stacking (or vertical zoning), which determines on which floor (of one or more buildings) a plurality of areas (arrangement elements) classified by function or property are to be arranged. The present embodiment will be described with an example where six areas A1 to A6 are arranged in one four-story building as shown in FIG. 2.

[0012] In this specification, an area refers to an arrangement element distinguished by spatial function or property, such as a corporate business department, a conference room, or a cafeteria. The present invention is not limited to a single building, and can also be applied to stacking for a plurality of floors in two or more buildings.

[0013] As shown in FIG. 1, the arrangement plan creation system 101 includes the information processing apparatus 100, an input device 30 operated by an operator or the like to input information to the information processing apparatus 100, and a display device 31 that displays information output from the information processing apparatus 100.

[0014] The information processing apparatus 100 is configured by a computer including: a CPU (Central Processing Unit) that executes control programs and the like; a ROM (Read-Only Memory) that stores control programs executed by the CPU; a RAM (Random Access Memory) that stores calculation results of the CPU and the like; and a communication device, etc. The information processing apparatus 100 implements various functions of the information processing apparatus 100 described in this specification when a control program stored in the ROM is loaded into the RAM and executed by the CPU on the RAM. The information processing apparatus 100 may be configured by a single computer, or may be configured by a plurality of microcomputers, and each control may be processed in a distributed manner by the plurality of computers.

[0015] The information processing device 100 includes a storage unit 10 that stores various information necessary to execute the information processing method of this embodiment, and a processing unit 20 that creates a layout plan for arranging multiple areas on multiple floors based on the information stored in the storage unit 10.

[0016] The memory unit 10 stores in advance the area of ​​multiple floors, the occupied area of ​​the area to be placed, and the proximity between different areas. The floor area, the occupied area of ​​the area, and the proximity are stored in the memory unit 10, for example, in map format. The proximity is arbitrarily set based on the amount of interaction between users of each area, and as shown in Figure 3, the value is set so that the stronger the connection between areas, the higher the value. In other words, the proximity represents the degree to which areas are to be placed close together. In addition, information necessary to execute the information processing method described later is input to the memory unit 10 through the input device 30.

[0017] The information processing method of this embodiment includes the steps of: randomly creating a plurality of initial layout options; optimizing the plurality of initial layout options based on evaluation values ​​obtained using the floor area, the occupied area of ​​each area, and the degree of proximity between different areas to generate an optimal layout option; clustering the plurality of optimal layout options to generate a plurality of classes; extracting feature quantities for each of the plurality of classes based on commonalities among the optimal layout options included in the class; and outputting the feature quantities of the classes and the optimal layout options included in the class in correspondence with each other.

[0018] In other words, as shown in Figure 1, the processing unit 20 includes an initial layout generation unit 21 that randomly creates a plurality of initial layout options corresponding to each step of the information processing method of this embodiment; an optimization processing unit 22 that optimizes the plurality of initial layout options to generate an optimal layout option based on evaluation values ​​obtained using the floor area, the occupied area of ​​the area, and the degree of proximity between different areas; a class generation unit 23 that clusters the plurality of optimal layout options to generate a plurality of classes; a feature extraction unit 24 that extracts feature quantities of each of the plurality of classes based on commonalities among the optimal layout options included in the class; and an output unit 25 that outputs the feature quantities of the class and the optimal layout options included in the class in correspondence with each other. Note that each configuration of the information processing device 100 shown in Figure 1 represents each function of the information processing device 100 as a virtual unit and does not necessarily mean that it exists physically.

[0019] When the processing unit 20 receives an input signal based on the operator's operation input, it executes the information processing method according to this embodiment by the process shown in Figure 4.

[0020] The following describes in detail the information processing method of this embodiment for creating a layout plan that places multiple areas on multiple floors.

[0021] As shown in Figure 4, first, in step S10, a predetermined number (two or more) of initial layout plans are generated. In this embodiment, as an example, 300 initial layout plans are generated. The initial layout plans are generated by randomly arranging multiple areas on multiple floors based on random numbers. The layout plans are represented as n-dimensional (n is the number of areas to be arranged) vectors. In other words, in this embodiment, the layout plans are represented as 6-dimensional vectors corresponding to 6 areas. The value of each element indicates the floor on which each area is placed, in the order of area A1 to A6. For example, as shown in Figure 2, layout plan = [1,3,3,4,4,2] means that area A1 is placed on the 1st floor, areas A2 and A3 on the 3rd floor, areas A4 and A5 on the 4th floor, and area A6 on the 2nd floor.

[0022] In step S11, the multiple initial placement options generated in step S10 are optimized based on evaluation values.

[0023] The evaluation value is calculated using the following formula (1), based on the area of ​​each floor, the occupied area of ​​each area, the distance between areas, and the degree of proximity between areas. n represents the number of areas, and m represents the number of floors.

number

[0024] W1 and W2 are weights that can be arbitrarily set to values ​​greater than 0. k If the value is less than 0 (negative), it is treated as 0; otherwise, the value itself is used.

[0025] In the proposed layout, it is desirable for two areas with high affinity to be placed close together, and it is desirable to have as little surplus area as possible on each floor. Therefore, in this embodiment, the evaluation value increases as the distance between two areas with high affinity increases. Also, the evaluation value increases as the difference between the floor area and the total area of ​​the areas placed on that floor increases, in other words, as the surplus area on the floor increases. Therefore, in this embodiment, a lower evaluation value indicates a better layout.

[0026] The optimization method for the initial placement plan is based on a genetic algorithm, as shown in Figure 5. Each element of the vector-represented placement plan is associated with the position of a gene, and operations such as crossover, mutation, and selection are performed to generate individuals (placement plans) with superior evaluation scores.

[0027] The following describes step S11, which generates the optimal layout plan, with reference to Figure 6.

[0028] First, in generating the optimal layout plan, in step S20, the initial layout plan generated in step S10 of Figure 4 is set as the parent generation set.

[0029] Next, in step S21, a crossover operation is performed on the initial arrangement of the parent generation with a predetermined probability (crossover rate X [%]). In this embodiment, a uniform crossover is performed with a probability (operation rate x1 [%]) that each gene (element) will be crossed if a crossover occurs. As explained by the example shown in Figure 7, two individuals are selected from the parent generation and a crossover is performed with a predetermined crossover rate X. When a crossover is performed, each element of the two corresponding individuals is crossed (exchanged) with a predetermined operation rate x1. Next, two more individuals are selected from the remaining parent generation population and the same operation is performed. In step S21, this operation is continued until it is no longer possible to select two individuals from the parent generation population. Note that the crossover rate X and the operation rate x1 in the crossover are set to values ​​greater than 0%.

[0030] Next, in step S22, mutation operations are performed on the parent generation configuration and the configuration generated by crossover in step S21. In this embodiment, two types of mutations occur independently for each configuration with a predetermined probability (mutation rate Y[%]).

[0031] As shown in Figure 8(a), in the first mutation, each element of the proposed configuration is replaced with another randomly selected element with a predetermined probability (operation rate y1[%]). In the case of the first mutation, even if it is determined that a mutation occurs with a predetermined mutation rate Y, if it is determined that no replacement will occur in any element, the original configuration remains (the same configuration as the original is generated).

[0032] As shown in Figure 8(b), in the second mutation, two values ​​are randomly selected from the floor numbers (element values), and these two values ​​included in the layout plan are swapped. Even in the second mutation, the two values ​​swapped by the mutation may not be included in the layout plan, so even if a mutation is determined to have occurred based on the mutation rate Y, the original layout plan may remain (the same layout plan as the original is generated).

[0033] Thus, in this embodiment, two types of mutations can be performed on a single configuration.

[0034] Next, in step S23, evaluation values ​​are calculated for each of the initial parent generation configurations, configurations generated by crossover, and configurations generated by mutation, based on equation (1).

[0035] Next, in step S24, selection is performed on the initial parent generation configurations, configurations generated by crossover, and configurations generated by mutation. Specifically, as shown in Figure 5, first, 10 configurations with superior evaluation values ​​(low evaluation values) are selected from all target configurations (elite strategy). Next, 190 configurations are selected from the remaining configurations using a tournament selection method. In this way, a final total of 200 configurations are selected. In this embodiment, for example, the tournament size in tournament selection is set to 3.

[0036] Next, in step S25, 100 new placement options (new individuals) are generated in the same manner as in step S10 in Figure 4, which generates the initial placement options. Then, the 200 placement options selected by elimination in step S24 and the 100 new placement options generated in step S25, for a total of 300 placement options, are set as the next generation.

[0037] Next, in step S26, it is determined whether the termination condition is met, which is whether the next generation has been generated from the parent generation a predetermined number of times (i.e., the generation has advanced a predetermined number of times). If the termination condition is not met, in step S27, the current next generation is set as the new parent generation, and the process returns to step S21. In this way, steps S21 to S25 are repeatedly executed until it is determined in step S26 that the termination condition is met.

[0038] If the termination condition is met in step S26, the next generation of layout plans is evaluated. In the evaluation, an evaluation value is calculated for each next generation of layout plans, and the layout plan with the best (lowest) evaluation value is extracted as the optimal layout plan. In this way, one optimal layout plan is generated from multiple initial layout plans, and the process shown in Figure 6 is completed.

[0039] Next, in step S12 shown in Figure 4, it is determined whether a predetermined number (10 in this embodiment) of optimal placement options have been generated. If the predetermined number of optimal placement options have not been generated, the initial seed value of the random numbers is changed in step S13, and the process returns to step S10. In this way, steps S10 and S11 are repeatedly executed, changing the initial seed value of the random numbers, until it is determined in step S12 that a predetermined number of optimal placement options have been generated. Because the initial seed value of the random numbers is changed in step S13, the initial placement options generated in step S10 and the probability-based processing results used in the process shown in Figure 5 will be different in each loop that repeats steps S10 and S11. Therefore, it is possible to generate multiple optimal placement options that are different from each other.

[0040] If it is determined in step S12 that a predetermined number of optimal placement options have been generated, in step S14, the predetermined number of optimal placement options are clustered and classified. In this embodiment, hierarchical clustering is employed as the method for clustering multiple optimal placement options. More specifically, in this embodiment, hierarchical clustering using Ward's method is performed as an example of a clustering method. The processing in step S14 will be described in detail below with reference to Figures 9 to 11.

[0041] In step S14, the process shown in Figure 9 is executed. First, in step S30, a class is assigned individually to each of the 10 optimal placement options (see Figure 10(a)).

[0042] Next, in step S31, the distance between each class and the other classes is calculated. Since the calculation of the distance between classes in Ward's method is well known, a detailed explanation is omitted here.

[0043] Next, in step S32, the two classes with the smallest distance from each other are merged into a single class. For example, as shown in Figure 10(b), class 2 and class 3, which have the smallest distance from each other, are merged to create a new class.

[0044] Steps S31 and S32 are repeatedly executed until it is determined in step S33 that the classes have been merged into one. This process divides the multiple optimal placement options into classes, as shown in the tree diagram in Figure 11, from the initial classes (number of classes = 10) to a single class (number of classes = 1). In other words, the multiple optimal placement options can be divided into any number of classes (natural numbers) from 1 to 10, which is the same number as the optimal placement options. In this way, the multiple optimal placement options are clustered, and the process shown in Figure 9 is completed.

[0045] Next, in step S15 shown in Figure 4, features (schemas) are extracted from the classes of optimal placement proposals, which have been divided into a predetermined number of classes, based on the commonalities between the optimal placement proposals. The number of classes into which the optimal placement proposals are divided represents the number of output classes and is set in advance. For example, if the number of classes to be divided is set to 3, as shown in Figure 11, the optimal placement proposals are divided into three classes: Class A, which includes optimal placement proposals 2, 3, 0, and 1; Class B, which includes optimal placement proposals 4, 7, and 8; and Class C, which includes optimal placement proposals 5, 6, and 9. Then, the elements of the optimal placement proposals included in Classes A to C are compared with each other to find commonalities. Specifically, each optimal placement proposal is compared element by element, and elements whose values ​​are common in optimal placement proposals of a predetermined percentage (70% in this embodiment) or more are extracted. Note that this predetermined percentage is not limited to the numerical value in this embodiment and can be set arbitrarily.

[0046] To explain using the calculation of features for class A as an example, as shown in Figure 12, in optimal placement options 0 to 3, the values ​​of the 1st to 3rd elements are common in 70% or more (3 or more options). Therefore, in class A, features are extracted where the value of the 1st element is 1, the value of the 2nd element is 3, and the value of the 3rd element is 3. In this embodiment, the extracted features are represented by an array as [1, 3, 3, *, *, *]. This array represents the element number that becomes a feature and its value corresponding to the vector representation of the placement options, and * is a wildcard symbol that means it is not a feature. These processes are performed for all the divided classes to extract the features for each class.

[0047] Next, in step S16, the feature quantities of each class extracted in step S15 are associated with the optimal placement options included in that class and output, and the process ends. The output information is input to the display device 31 (see Figure 1) and displayed. In this way, multiple optimal placement options are provided to the operator along with their features.

[0048] According to the above embodiments, the following effects and advantages are achieved.

[0049] In this embodiment, multiple optimal placement options are clustered, and then features are extracted from the classes representing each cluster. Therefore, since the generation of multiple classified optimal placement options and the extraction of their features are performed automatically, the optimal placement options can be easily obtained along with their features. According to this embodiment, features (schemas), which are subsets of genotypes that greatly influence the evaluation value (fitness), are visualized, making it easier to compare and examine placement options and improving stacking efficiency.

[0050] More specifically, in combinatorial optimization problems like stacking, as the number of variables increases, the number of combinations becomes enormous, the computational complexity increases exponentially, and it is often impossible to find a solution in a realistic amount of time. Therefore, even if it is not the exact optimal solution, a near-optimal solution (an approximate solution of the optimal solution) is sometimes adopted as an acceptable solution that can be obtained in a relatively short time. However, since the near-optimal solution can converge to various values ​​depending on the initial conditions (initial arrangement), it is difficult to determine which combination pattern was effective.

[0051] In contrast, in this embodiment, the functional placement optimization problem is solved multiple times, multiple near-optimal solutions are clustered, and features (commonality) are extracted from the classes representing the solution clusters to visualize the classes (set of solutions). Therefore, by systematically representing the commonality and similarity of each class, the characteristics of combination patterns of highly-rated placement proposals can be easily grasped, and the comparative examination of placement proposals becomes more efficient.

[0052] Furthermore, in this embodiment, two types of mutation operations are performed in the optimization using a genetic algorithm. In addition, in this embodiment, the next generation is generated by adding new individuals, which are generated independently of the parent generation, to individuals generated by crossover and mutation from the parent generation. With these configurations, it is possible to suppress falling into local optima with low evaluation values ​​and obtain a variety of solutions with high evaluation values.

[0053] Although embodiments of the present invention have been described above, these embodiments only represent a part of the application examples of the present invention, and are not intended to limit the technical scope of the present invention to the specific configurations of the above embodiments.

[0054] The following describes modifications of this embodiment. The following modifications are also within the scope of the present invention. Furthermore, it is possible to combine the configurations shown in the modifications with the configurations described in the above embodiment, or to combine the configurations described in the following different modifications.

[0055] In the above embodiment, the optimization method for generating the optimal layout from the initial layout is a genetic algorithm. However, the optimization method is not limited to a genetic algorithm; any method that solves a combinatorial optimization problem may be used. For example, simulated annealing, a metaheuristic that is independent of the specific problem, similar to a genetic algorithm, may be used.

[0056] Furthermore, in the above embodiment, hierarchical clustering is used as a method for clustering multiple optimal placement options. However, the clustering method is not limited to hierarchical clustering and may be other methods. For example, the k-means method, which is a non-hierarchical clustering method, may be used as the clustering method.

[0057] Furthermore, the evaluation formula used to calculate the evaluation value of the layout plan is not limited to formula (1) above, and may utilize at least the floor area, the occupied area of ​​each area, and the degree of proximity between areas.

[0058] Furthermore, the method for obtaining class features is not limited to the above embodiment, but can be any method.

[0059] Furthermore, the information processing device 100 may be configured, in whole or in part, as a cloud server located in a cloud environment.

[0060] Furthermore, the series of processes in the information processing device 100 described above may be provided to a computer as a program for execution.

[0061] In other words, the program according to the above embodiment causes the information processing device 100, which is a computer, to perform the following steps: randomly create a plurality of initial layout options; generate an optimal layout option from the plurality of initial layout options based on evaluation values ​​obtained using at least the floor area, the occupied area of ​​each area, and the degree of proximity between different areas; cluster the plurality of optimal layout options to generate a plurality of classes; for each of the plurality of classes, extract feature quantities of the class based on commonalities among the optimal layout options included in the class; and output the feature quantities of the class and the optimal layout options included in the class in correspondence with each other.

[0062] Furthermore, the program for executing the series of processes described above is provided by a storage medium readable by the information processing device 100. Alternatively, the program may be provided to the information processing device 100 via a network line. [Explanation of Symbols]

[0063] 100 Information Processing Devices 10 Storage section 20 Processing Units 21 Initial plan generation section 22 Optimization Processing Unit 23 Class Generation Unit 24 Feature extraction unit 25 Output section

Claims

1. An information processing method performed by a computer in a layout plan creation system for creating a layout plan for determining which of the multiple floors in one or more buildings to place multiple areas on, The first step is to randomly generate multiple initial placement options, A second step of generating one optimal layout plan from a plurality of initial layout plans based on evaluation values ​​obtained using at least the floor area, the occupied area of ​​the area, and the degree of proximity between different areas, The first and second steps are repeated while changing the initial seed value of the random numbers to generate a plurality of optimal arrangement proposals. The steps include: clustering multiple optimal placement proposals to generate multiple classes; For each of the multiple classes, the steps include extracting the features of the class based on the commonalities among the optimal arrangement proposals included within the class, The process includes the step of outputting the corresponding feature quantities of the class and the optimal arrangement proposals included in the class. Information processing methods.

2. The information processing method according to claim 1, The aforementioned feature quantities are represented as the areas and floors in which the arrangement is common in a predetermined proportion or more of the optimal arrangement proposals within the class. Information processing methods.

3. The information processing method according to claim 1, The optimization method for generating the aforementioned optimal arrangement is based on a genetic algorithm. Information processing methods.

4. The information processing method according to claim 1, The method for clustering multiple optimal placement options is based on hierarchical clustering. Information processing methods.

5. An information processing device that creates a layout plan for which of the multiple floors in one or more buildings to place multiple areas, An initial layout generation unit that performs a first process to randomly create multiple initial layout options, An optimization processing unit that performs a second process to generate one optimal layout from a plurality of initial layout options based on evaluation values ​​obtained using at least the floor area, the occupied area of ​​the area, and the degree of proximity between different areas, A class generation unit that clusters multiple optimal arrangement proposals to generate multiple classes, A feature extraction unit that extracts feature quantities for each of the multiple classes based on commonalities among the optimal arrangement proposals included within the class, The system includes an output unit that outputs the feature quantities of the class and the optimal arrangement proposals included in the class in correspondence with each other, The first process by the initial plan generation unit and the second process by the optimization processing unit are repeated while changing the initial seed value of the random numbers to generate a plurality of optimal arrangement plans. Information processing device.

Images