Information processing device, information processing system, information processing method, and information processing program

Abstract

In a natural disaster such as a large-scale earthquake, it is difficult to solve a route optimization problem of determining in which moving route, to which building, and in which order, each of a plurality of observers is sent for confirming damage states of a plurality of buildings, in a short time using a classical computer.　In view of the problem, the present invention provides an information processing device and the like that divide all observation destinations on a map into a plurality of grids as necessary, formulates constrained conditions such as estimation difficulty of various buildings with structure including timber, a steel frame, and a reinforced concrete, skills of observers, and desired observation time frame of each observation destination to provide the conditions to a quantum annealing machine in a QUBY form to enable clustering in consideration of various conditions for each grid and calculation of an optimal solution of the route optimization problem of the observation destination in each cluster.

Description

[Technical field]

[0001] The present invention relates to an information processing device, an information processing system, an information processing method, and an information processing program (hereinafter referred to as "information processing device, etc.") for solving the problem of optimizing the movement route (attendance route) of a non-life insurance adjuster (witness) in the event of an earthquake disaster, using a QUBO solver including a quantum computer, etc.

[0002] Specifically, the present invention relates to an information processing device, etc. that, in order to assess the damage status of a building caused by a disaster such as an earthquake, optimizes the travel routes and travel time required for each of multiple visitors to multiple destinations (desired attendance locations) by formulating various conditions taking into consideration the desired visit time frame (desired attendance time frame) and matching between the building to be visited and the appraiser (or attendee) in a Quadratic Unconstrained Binary Optimization (QUBO) format and providing the conditions to a quantum annealing machine. [Background technology]

[0003] In the event of a natural disaster such as an earthquake, non-life insurance companies calculate the amount of insurance payout based on the damage to the structure covered by the non-life insurance, so they basically check the damage to the structure through on-site inspections. In the event of a large-scale earthquake, for example, many inspectors, about 200 per office, are dispatched to the site, and on busy days, they carry out about 1,000 inspections (for assessing damage to structures, etc.).

[0004] When dispatching a large number of witnesses to the site, in order to efficiently conduct multiple witnesses in one day, it is necessary to optimize the travel route for each witness, such as which of the multiple witnesses to visit and in what order, and to minimize the required time, which includes at least travel time and witness time. This type of problem, such as optimizing the travel route and required time to multiple visits, is a type of mathematical optimization problem, and is known as the Traveling Salesman Problem (TSP) or Vehicle Routing Problem (VRP).

[0005] In recent years, systems have been proposed that use quantum computers to solve mathematical optimization problems such as TSP and VRP. For example, the system described in Patent Document 1 formulates a VRP related to logistics distribution in the QUBO format and solves it using a quantum solver in a quantum computing environment based on information (constraints) on "location" and "route preference." In addition, the system described in Patent Document 2 formulates a TSP or VRP in the QUBO format and uses an Ising solver such as a quantum computer to find the optimal solution.

[0006] A method of formulating constraints and the like to solve TSP or VRP, converting them into QUBO, and finding an optimal solution using a quantum computer such as D-Wave is also described in, for example, Non-Patent Document 1. For example, PyQUBO (Non-Patent Document 3) can be used as a domain-specific language that automatically constructs QUBO required to solve combinatorial optimization problems such as TSP or VRP using a quantum annealing quantum computer (quantum annealing machine) such as D-Wave (Non-Patent Document 2). [Prior art documents] [Patent documents]

[0007] [Patent Document 1] Patent Publication No. 2021-036417 [Patent Document 2] JP 2020-149403 A [Non-patent literature]

[0008] [Non-Patent Document 1] Irie, H., et al. "Quantum Annealing of Vehicle Routing Problem with Time, State and Capacity," arXiv (quant ph ):1903.06322, March 18, 2019(URL https: / / arxiv.org / pdf / 1903.06322.pdf)(Retrieved October 1, 2021) [Non-Patent Document 2] D-Wave System Introduction User Manual (URL https: / / dwavejapan.com / app / uploads / 2019 / 12 / 09-1076A-T_J_GettingStarted.pdf) (Retrieved October 1, 2021) [Non-Patent Document 3] PyQUBO (URL https: / / pyqubo.readthedocs.io / en / latest / ) (Retrieved October 1, 2021) Summary of the Invention [Problem to be solved by the invention]

[0009] Conventionally, in the event of a natural disaster such as a large-scale earthquake, in order to plan which route to take for each of multiple witnesses to check the damage status of multiple buildings, and in what order to dispatch them, a route calculation solver is executed on a general-purpose computer (classical computer) to obtain an optimal solution. However, since the calculation of an optimal solution using a classical computer takes an enormous amount of time, the calculation is basically stopped after a certain period of time (e.g., 3 to 4 hours) has passed, and the optimal solution among those obtained at that time is simply output as the optimal solution, so there is a possibility that a sufficiently optimal solution is not obtained.

[0010] In addition, the systems for finding optimal solutions to mathematical optimization problems as shown in Patent Documents 1 and 2 each formulate TSP or VRP under specific constraints in the QUBO format to find the optimal solution, and cannot be directly applied to solving the problem of optimizing the travel routes and travel times required for each of multiple traders to visit multiple trading sessions (hereinafter simply referred to as the "route optimization problem").

[0011] Therefore, in this invention, in order to solve such a route optimization problem, all trading venues on a map are divided into multiple grids as necessary, and constraints such as the appraisal difficulty (estimated difficulty) of various structures such as wooden, steel-framed, and reinforced concrete, the appraisal time required for various structures (estimated required time), the appraisal skill level (specifications) of the witnesses, and the desired time frame for each trading venue are formulated and given to a quantum annealing machine or the like in QUBO format, and clustering is performed for each grid taking into account various conditions, after which the route optimization problem for the trading venues in each cluster is solved, and an information processing device or the like is provided that can determine multiple trading venues for each witness and calculate the travel route to and visit order for those buildings. [Means for solving the problem]

[0012] As one embodiment of an information processing device according to the present invention, an information processing device for determining a plurality of buildings to be visited by each of a plurality of visitors and an order in which to visit the plurality of buildings includes: an input data setting unit that sets input data including at least an appraisal difficulty level for each of the plurality of buildings, an appraisal time required for each of the plurality of buildings, an appraisal proficiency level for each of the plurality of visitors, and a desired visiting time frame for each of the plurality of buildings; a clustering calculation unit that performs a clustering process to divide the plurality of buildings corresponding to the position information into a plurality of clusters using position information of the plurality of buildings based on a predetermined cluster constraint condition, or that determines the destinations of each of the plurality of visitors by having a quantum computer connected via a network perform the clustering process; a route optimization problem calculation unit that causes the quantum computer to execute an optimization process for determining an order of visiting a plurality of buildings to be visited in each of the plurality of clusters based on a predetermined optimization constraint condition; Including, The predetermined cluster constraint condition is expressed by a formulated first formula, and the first formula includes a constraint term for a structure with a maximum appraisal difficulty to belong to a cluster of visitors with a maximum appraisal proficiency, a constraint term for a structure with an intermediate appraisal difficulty to belong to a cluster of visitors with an appraisal proficiency level equal to or higher than the intermediate level, and a constraint term for dispersing a desired visiting time frame and an appraisal time required for route optimization; The predetermined optimization constraint condition is expressed by a formulated second mathematical expression, and the second mathematical expression includes a constraint term related to the desired visiting time slot.

[0013] In one embodiment of the information processing device according to the present invention, the clustering calculation unit performs a grid division process for dividing the image into a plurality of grids using only position information of the plurality of buildings before executing the clustering process, The clustering process is performed for each of the plurality of grids.

[0014] As one embodiment of the information processing device according to the present invention, the first formula further includes a cost term that reduces the travel distance taking into account the routes between multiple structures that are to be visited and included in the cluster, a cost term that reduces the subtotal of the estimated time in the desired visiting time slot for each of the multiple structures that are to be visited and included in the cluster, and a constraint term that ensures that each of all clusters includes a predetermined number of locations to be visited.

[0015] As one embodiment of the information processing method according to the present invention, there is provided an information processing method for determining a plurality of buildings to be visited by each of a plurality of visitors and an order in which to visit the plurality of buildings, the information processing method being executed by a terminal device connected to a quantum computer via a network, setting input data including at least an appraisal difficulty level for each of the plurality of buildings, an appraisal skill level for each of the plurality of visitors, and a desired visiting time frame for each of the plurality of buildings; A step of determining destinations of each of the plurality of visitors by performing a clustering process using position information of the plurality of buildings based on a predetermined cluster constraint condition, to divide the plurality of buildings corresponding to the position information into a plurality of clusters, or by having the quantum computer perform the clustering process; causing the quantum computer to execute an optimization process for determining an order of visiting a plurality of buildings to be visited in each of the plurality of clusters based on a predetermined optimization constraint condition; Including, The predetermined cluster constraint condition is expressed by a formulated first formula, and the first formula includes at least a constraint term for a structure having a maximum appraisal difficulty to belong to a cluster of visitors having a maximum appraisal proficiency level, and a constraint term for a structure having an intermediate appraisal difficulty to belong to a cluster of visitors having an appraisal proficiency level equal to or higher than the intermediate appraisal proficiency level, The predetermined optimization constraint condition is expressed by a formulated second mathematical expression, and the second mathematical expression includes at least a constraint term related to the desired visiting time slot.

[0016] In one embodiment of the information processing method according to the present invention, the step of determining the destinations of each of the plurality of visitors includes, before the clustering process, performing a grid division process for dividing the map into a plurality of grids using only position information of the plurality of buildings; The clustering process is performed for each of the plurality of grids.

[0017] As one embodiment of the information processing method according to the present invention, the first formula further includes a cost term that reduces the travel distance taking into account the routes between the multiple structures to be visited that are included in the cluster, a cost term that reduces the subtotal of the estimated time in the desired visiting time slot for each of the multiple structures included in the cluster, and a constraint term that ensures that each of all clusters includes a predetermined number of locations to be visited.

[0018] As one embodiment of an information processing system according to the present invention, the information processing system includes: The information processing device; a quantum computer connected to the information processing device via a network; Including, The quantum computer receives QUBOs obtained by converting the first formula and the second formula from the information processing device, and executes the clustering process and the optimization process.

[0019] As one embodiment of an information processing program according to the present invention, the program is executed by a computer to cause the computer to function as each unit of the information processing device.

[0020] As another embodiment of an information processing program according to the present invention, the program is executed by a computer to cause the computer to perform each step of the information processing method. Effect of the Invention

[0021] In order to plan which route to dispatch each of a plurality of witnesses to which buildings and in what order to dispatch them in order to check the damage status of a plurality of buildings during a natural disaster such as a large-scale earthquake, the information processing device according to the present invention inputs a QUBO, which formulates constraints such as the difficulty of appraisal of various buildings, the appraisal proficiency of the witnesses, and the desired time frame for each witness, into a quantum annealing machine and calculates the optimal solution to the route optimization problem, and it is possible to determine the optimal buildings to be visited by each of the plurality of witnesses and obtain the route to and the order of visit to those buildings without terminating the calculation after a certain period of time. In this way, the present invention can solve a previously unconventional route optimization problem to the visit destinations of witnesses (appraisers). As a result, it is possible to efficiently carry out on-site inspections to check the damage status of buildings during a disaster, and it is possible to contribute to the rapid calculation of the amount of insurance payment according to the damage status of buildings to which non-life insurance is applied. [Brief description of the drawings]

[0022] [Figure 1] 1 is a diagram illustrating an example of a configuration of an information processing device and a system according to an embodiment of the present invention. [Diagram 2] 11 is a flowchart showing a process flow for finding an optimal solution to a route optimization problem in an information processing device. [Diagram 3] FIG. 3 is a diagram showing an outline of grid division, clustering, and calculation of the traveling salesman problem (TS-TSP) in the flowchart shown in FIG. 2. [Figure 4] FIG. 13 is a diagram showing an example of grid division implemented based on position information (latitude and longitude coordinates) of buildings on a map that are to be attended. [Diagram 5] FIG. 13 is a diagram showing an example in which three clusters out of a plurality of clusters generated by clustering performed on one grid are displayed. [Figure 6] FIG. 13 is a diagram showing an example of displaying optimal travel routes to multiple destinations in one cluster. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0023] Hereinafter, an embodiment of the present invention will be described with reference to the accompanying drawings. The following embodiment is an example for explaining the present invention, and is not intended to limit the present invention to only the embodiment. Furthermore, the present invention can be modified in various ways without departing from the gist of the invention. Furthermore, the same components in each drawing are denoted by the same reference numerals as much as possible, and duplicated explanations are omitted.

[0024] Fig. 1 shows an example of the configuration of an information processing device and a system according to an embodiment of the present invention. In the embodiment shown in Fig. 1, a client computer 100, which is an information processing device, is connected to an annealing machine 200, which is an example of a quantum computer connected via a network N. A user can use the client computer 100 to cause the annealing machine 200 to execute calculations for route optimization problems such as TSP and VRP. The client computer 100 is an information processing device that can be connected to the network N, such as a personal computer, a notebook computer, a smartphone, or a mobile phone.

[0025] An information processing system according to one embodiment of the present invention includes a client computer 100 and an annealing machine 200. In the embodiment shown in Fig. 1, there is one client computer 100, but there may be multiple client computers 100, and there may be not only one but also multiple annealing machines 200.

[0026] The network N may be, for example, an open network such as the Internet, an intranet connected by a dedicated line, or a closed network. The network N is not limited to this, and a combination of a closed network and an open network may be used as appropriate depending on the required security level, etc.

[0027] Typically, a client computer 100 includes one or more processing units (CPUs) 102, memory 104, one or more network interfaces (or other communication interfaces) 106, and one or more communication buses 114 for interconnecting these components.

[0028] The client computer 100 may also include a user interface 108, which may include, for example, a display 110 and an input device (such as a keyboard and / or a mouse or some other pointing device) 112. The input device 112 may be a touch panel.

[0029] Memory 104 may be, for example, a high-speed random access memory such as DRAM, SRAM, DDR RAM, or other random access solid-state storage device, or may be a non-volatile memory such as one or more magnetic disk storage devices, optical disk storage devices, flash memory devices, or other non-volatile solid-state storage devices.

[0030] The CPU 102 executes various processes according to a program recorded in the memory 104 or a program loaded from a storage device (not shown) to the memory 104. The CPU 102 can execute a program for causing the client computer 100 to function as an information processing device for determining a plurality of buildings to be visited by each of a plurality of visitors and the order in which the buildings are to be visited, by calculating a route optimization problem such as TSP or VRP using the annealing machine 200, for example. It is also possible to implement at least a part of the functions of the information processing device in hardware, such as an application specific integrated circuit (ASIC).

[0031] Operating system 116, for example, handles various basic system services and contains procedures for performing tasks using the hardware.

[0032] The network communications module 118 is used, for example, to connect the client computer 100 to other computers via one or more network interfaces 106 and one or more networks N, such as the Internet, other wide area networks, local area networks, metropolitan area networks, etc.

[0033] The annealing machine setting module 120 includes an input data setting unit 122, a clustering calculation unit 124, a TS-TSP calculation unit (route optimization problem calculation unit) 126 that calculates a route optimization problem such as a Time Scheduled Traveling Problem (TS-TSP) that takes into account various times such as a desired visiting time frame (desired session time frame), appraisal time (estimated time), and travel time for each cluster, and a calculation result visualization unit 128. The annealing machine setting module 120 is a program recorded or loaded in the memory 104, and can cause hardware resources such as the CPU 102 and the memory 104 to function as the input data setting unit 122, the clustering calculation unit 124, the TS-TSP calculation unit 126, and the calculation result visualization unit 128 by being executed by the client computer 100.

[0034] Fig. 2 is a flowchart showing a process flow for finding an optimal solution to a route optimization problem in an information processing device. Fig. 3 shows an outline of grid division, clustering, and calculation of the traveling salesman problem (TS-TSP) in the flowchart shown in Fig. 2. The functional configuration of each part of a client computer 100, which is an information processing device, and the process flow will be described with reference to Figs. 1 to 3.

[0035] In step S10, in order to plan which route to take for each of multiple witnesses (visitors) to take, to which buildings, and in what order to dispatch them in order to check the damage status of multiple buildings in the event of a natural disaster such as a large-scale earthquake, the client computer 100 accepts input from the user via the user interface 108 of the appraisal difficulty (estimation difficulty) of various buildings, the witnesses' appraisal proficiency (specifications), the desired time frame for visiting each destination (the desired time frame for attendance at each destination), etc., and sets (stores) the input data in the input data setting unit 122.

[0036] The annealing machine setting module 120 can perform settings for operating the annealing machine 200 in response to input received from a user. For example, the input data setting unit 122 can store the specifications of each of a plurality of witnesses. The specifications of the witnesses can be three types of values, namely, a maximum value such as "high", "medium", and "low", an intermediate value, and a minimum value.

[0037] The input data setting unit 122 can also store attributes of the difficulty of estimation, the desired time frame for attendance, and the appraisal time (estimated time) for each of the multiple buildings to be witnessed. The difficulty of estimation for a building can have three types of values, i.e., maximum value, median value, and minimum value, such as "difficult", "medium", and "easy", the desired time frame for attendance can have three types of values, such as morning "AM", early afternoon "PM1", and late afternoon "PM2", and the estimated time (hours (h)) can have three types of values, such as "0.5", "1.0", and "1.5". The input data setting unit 122 can also store the distance between the multiple buildings to be witnessed. Note that, although the specifications of the witness, the difficulty of estimation for a building, the desired time frame for attendance, and the estimated time are each set to three types of values, this is not limiting.

[0038] Furthermore, the input data setting unit 122 can also store constraints and the like necessary for calculating TS-TSP. Examples of constraints include (1) an observer leaves from a certain office and returns to the original office, (2) one observer visits one destination exactly once, (3) one observer visits five buildings, (4) an observer who can handle the difficulty of the destination visits, and (5) the same observer is in charge of the same area. Regarding the visit of an observer who can handle the difficulty of the destination visits, for example, when the estimation difficulty of the building is the maximum value "difficult", an observer whose specifications are the maximum value "high" can handle it, and when the estimation difficulty of the building is the intermediate value "medium", an observer whose specifications are the maximum value "high" or the intermediate value "medium" can handle it. Note that the number of buildings visited by an observer is not limited to five, and may be any number. For example, a predetermined number designated by a user can be the number of buildings visited by an observer.

[0039] In one embodiment of the present invention, the client computer 100 can cause the annealing machine 200 to execute calculations for a route optimization problem (TS-TSP) based on the inputs of the difficulty of estimating various structures, the specifications of the observers, the desired time slots for each observation destination, the distance between structures, etc., stored in the input data setting unit 122 as described above, and the constraints, with the aim of minimizing the total travel time (including the time required for observation, such as the estimated time) while satisfying the constraints.

[0040] Next, in step S11, the clustering calculation unit 124 can perform rough clustering based on the position information (e.g., latitude and longitude) of the buildings that are the destinations as necessary, and divide all the destinations into multiple grids on the map data. The map data includes at least the position information of the buildings that are the destinations and route information such as roads in the area that includes all the destinations, and can use an existing map API or the like. Since the amount of calculation in the annealing machine 200 depends on the number of buildings that are the destinations, if it takes a long time for the annealing machine 200 to obtain an optimal solution, rough clustering can be performed in advance to reduce the size of the problem.

[0041] In the annealing machine 200, since it is not practical to calculate the distance matrix between buildings visited and the QUBO for about 1,000 visited places at once, it is necessary to divide the system into rough grids in advance to reduce the problem size. In the information processing device according to one embodiment of the present invention, the implementation was performed under the assumption that (for simplicity) the distribution ratio of the witness specifications, the difficulty of estimating the buildings visited, the desired time frame for attendance, and the estimated time in the instance is maintained in each grid after division. The number of divisions into grids was set to 10 so that the clustering process (step S12) in the subsequent process can be performed in a realistic time. In other words, if the number of visits is 1,000, it can be divided into 100 × 10 grids and clustering can be performed for each grid. The grid division (step S11) can be performed based on latitude and longitude information (straight-line distance without considering the route on the map) using a conventional classical algorithm (for example, a conventional clustering algorithm such as the K-means method with constraints). The grid division (step S11) may also be performed by a classical computer such as the client computer 100. If the problem size is small, the grid division (step S11) can be omitted.

[0042] The left image in Fig. 3 shows the map divided into four grids based on the location information (e.g., latitude and longitude) of all visited buildings on the map. The second image from the left shows an enlarged view of the grid surrounded by a solid-line square among the four grids.

[0043] Next, in step S12 (FIG. 2), the clustering calculation unit 124 executes a process of clustering the destinations for each of the multiple grids created in the previous step (step S11) (or for all destinations if grid division has not been performed). In order to make it easier to obtain a feasible solution in the TS-TSP in the subsequent step (step S12), each constraint condition (cluster constraint condition) is taken into consideration at the time of clustering, and the problem is formulated in QUBO format based on these cluster constraint conditions as follows. · First term (coefficient A): A cost term that reduces the travel distance within a cluster (taking into account the route). Close destinations are placed in the same cluster. · Second term (coefficient B): A constraint term for all destinations to belong to one of the clusters. · The third term (coefficient C): A constraint term that ensures that every cluster contains five destinations. ·Fourth term (coefficient D): A cost term that reduces the subtotal of estimated times in each desired session time frame period {AM, PM1, PM2} included in the cluster. This prevents destinations with long desired times and estimated times from being biased within the cluster. Term 5 (coefficient E): A constraint term for when a visit to a place with a difficulty level of "difficult" belongs to a cluster of witnesses with a "high" skill, and when a visit to a place with a difficulty level of "medium" belongs to a cluster of witnesses with a "high" or "medium" skill.

[0044] QUBO, which formulates clustering taking into account the above items 1 to 5, is as follows.

number

[0045] The first term (coefficient A) groups nearby destinations into the same cluster. u and v are destinations, and D uv is the travel distance between destination u and destination v. In the second term (coefficient B), all destinations are assigned to one cluster, and in the third term (coefficient C), the points included in one cluster (i.e., the number of buildings that one person visits), and M in the third term represents the number of destinations included in the cluster, and in one embodiment, M=5. By setting M=5, five destinations are included in all clusters. In other words, when five destinations are included in all clusters, the third term is expressed as follows:

number

[0046] The fourth term (coefficient D) is a term that spreads the desired time and estimated time for route optimization, ensuring diversity. In this embodiment, the sum of squares of estimated times for the same time slot (time frame) in the same cluster is taken. Specifically, the fourth term can be expressed as follows:

number

[0047] The fifth term (coefficient E) is a constraint term for matching clusters with the points contained therein. Let V = K × N, and for k, n ∈ V, it is defined that k, n ∈ Ematch ⇔ k and n are consistent. In this embodiment, being consistent means the consistency between the specifications of the witness and the difficulty of estimating the building. The fifth term adds a consistent n to k. If k is divided and written down for each difficulty of estimation, in this embodiment, the fifth term can be expressed as follows, divided into a coefficient E1 term and a coefficient E2 term.

number

[0048] The optimal solution in clustering (step S12) may be found not only by using a quantum algorithm (quantum annealing) by a quantum computer such as the annealing machine 200, but also by using a classical algorithm (simulated annealing) by a classical computer such as the client computer 100.

[0049] An overview of the clustering in step S12 is shown in the two central figures in Figure 3 (the second figure from the left and the second figure from the right). In the example shown in Figure 3, destinations 1 to 9 around starting point D are divided into clusters shown with the same pattern, such as {1, 2, 3, 4}, {5, 6, 9}, and {7, 8}, through the clustering process.

[0050] Next, in step S13 (see FIG. 2), the TS-TSP calculation unit (route optimization problem calculation unit) 126 calculates the traveling salesman problem (TS-TSP) taking into account various times (travel time, estimated time, time required for session, etc.) for each cluster. In this embodiment, optimization of visiting routes (TS-TSP calculation) was performed for each cluster (each including five visiting destinations) obtained in the clustering process (step S12). The TS-TSP calculation (optimization process) can take into account the desired session time frame and estimated time. Details of the optimization constraint conditions are as follows:

[0051] Regarding the division by hour, 9:00~17:00 (8h) is Δ t The time was divided by (h) and the processing unit τ (0~T) was set as follows: When Δt=0.5(h), AM: 0≦τ≦5 (0.0~2.5(h)), PM1: 6≦τ≦11 (3.0~5.5(h)), PM2: 12≦τ≦15 (6.0~7.5(h)).

[0052] Travel time was calculated by dividing the route distance between destinations (a, b) by the average travel speed.

number

[0053] For the estimated time, each destination has Δ t The unit session time was treated as a cost term.

number

[0054] The estimated time was taken into account together with travel time as follows:

number

[0055] We assumed a route that starts from a starting point (depot) and finally returns to the same depot. The QUBO that formulates the TS-TSP calculation by considering the above constraints (optimization constraints) is as follows. The following formula is based on existing methods (for example, Non-Patent Document 1).

number

[0056] The first term is a cost term, the second term is a constraint to prohibit any early arrival from the depot, and the third term is a constraint to not exceed a time limit when arriving (returning) to the depot.

[0057] The fourth term represents the constraint on the desired session time, where TG is a set of time slots, TG={AM, PM1, PM2} in this embodiment. a∈T represents that the desired time of location (destination) a is included in time slot T. The fourth term can be written as follows, divided into each time slot.

number

[0058] An overview of the optimization process (TS-TSP calculation) in step S13 is shown in the right diagram of Figure 3. In the example shown in Figure 3, the travel routes starting from the starting point D and returning to D are shown to be determined for each of the clusters {1, 2, 3, 4}, {5, 6, 9}, and {7, 8}.

[0059] Finally, in step S14 (see FIG. 2), the calculation result visualization unit 128 can display the travel route determined for each cluster on the display 110 of the client computer 100, for example, by overlaying it on map data.

[0060] 4 to 6 show the results of solving the TS-TSP for the base case using the present invention. FIG. 4 shows an example of grid division implemented based on the position information (latitude and longitude coordinates) of buildings on a map to be attended. In FIG. 4, 1,000 destinations within an 8 km radius of Kawajiri Station in Kumamoto Prefecture, which was set as the starting point Depot in the base case, were divided into 10 based on latitude and longitude (without considering routes such as roads on the map) so that there are 100 destinations per grid. In FIG. 4, destination points in the same grid are shown with the same shape and pattern. Dividing into grids is equivalent to dividing the calculation domain, and the amount of calculations performed at one time by the annealing machine 200 (or the client computer 100) can be reduced.

[0061] FIG. 5 shows an example of displaying three clusters among a plurality of clusters generated by clustering performed in one grid, and destinations that are matched with the specifications of the observers and the difficulty of estimating the buildings and appropriately distributed desired attendance times are displayed in three clusters in a circled area on the map. FIG. 6 shows an example of displaying an optimal travel route to a plurality of destinations in one cluster, and displays the visit order and visit route starting from the starting point D, from visit points 1 to 4, from 4 to 2, from 2 to 3, from 3 to 5, and from 5 back to D. The displays shown in FIGS. 4 to 6 are examples of visualization of the calculation results in step S14 of the flowchart shown in FIG. 2, and can be performed by the calculation result visualization unit 128 of the client computer 100. In the embodiment shown in FIGS. 4 to 6, OpenStreetMap is used as map data, but the present invention is not limited to this, and any map database that stores information for solving the route optimization problem, such as distances between buildings and route information, such as Google Maps and Zenrin Maps, may be used.

[0062] In this way, the client computer 100 according to one embodiment of the present invention provides the annealing machine 200 with a QUBO that formulates constraint conditions such as the difficulty of estimating various structures at the destination, the specifications of the traders, and the desired time frame for each trader, and performs grid division, clustering, and TS-TSP calculation (steps S11 to S13 in FIG. 2), thereby determining the optimal structures at the destination for each of the traders and obtaining the travel route to and the order of visiting the structures. In other words, the present invention is the first to solve the problem of optimizing the travel route (trading route) of traders, which has been difficult to solve until now, by using a QUBO solver including a quantum computer.

[0063] For example, in the event of a natural disaster such as a large-scale earthquake, the travel routes and visiting order shown in FIG. 6 can be presented to the user of client computer 100, thereby assisting in the rapid planning of which travel route each of multiple witnesses should take, to which buildings, and in what order to dispatch them in order to check the damage status of multiple buildings.

[0064] Table 1 below shows the breakdown of the TS-TSP results for one cluster in the base case visualized as shown in Figure 6. Table 1 shows the visit order, travel time [h], arrival time [h], arrival category, estimated time [h], and travel distance [km]. The arrival time [h] is the time obtained by adding the arrival time [h] value to the session start time. For example, if the session start time is 9:00 a.m., the arrival time at visit 1 is 10:30 a.m. (=9+1.5), and the arrival time at visit 2 is 11:15 a.m. (=9+2.25). Table 3 shows the travel route starting from the departure point D, visiting visits 1 and 4 in the morning (AM), visiting visits 2 and 3 in the first half of the afternoon (PM1), visiting visit 5 in the second half of the afternoon (PM2), and returning to the departure point D, together with the travel distance, etc. [Table 1]

[0065] As described above, the information processing system according to the present invention uses a quantum computer (annealing machine) to solve a route optimization problem, and can find an optimal solution to the route optimization problem in a shorter time than the calculation time required by a classical algorithm, without terminating the calculation at a certain time, as is the case when an optimal solution is calculated by a classical computer (classical algorithm). This can contribute to planning, such as which route to send each of multiple witnesses to which buildings and in what order to check the damage status of multiple buildings during a natural disaster such as a large-scale earthquake. [Industrial Applicability]

[0066] The information processing device and the like according to the present invention can be used for purposes such as determining the optimal travel route (attendance route) for a non-life insurance adjuster (witness) in the event of an earthquake disaster. [Explanation of symbols]

[0067] 100: Client computer 102 :CPU 104: Memory 106: Network Interface 108: User Interface 110: Display 112: Input device 114: Communication bus 116: Operating System 118: Network communication module 120: Annealing machine setting module 122: Input data setting section 124: Clustering calculation section 126:TSP calculation section 128: Calculation result visualization part 200: Annealing machine

Claims

1. An information processing device for determining a plurality of buildings to be visited by each of a plurality of visitors and an order in which the plurality of buildings are to be visited, the information processing device comprising: an input data setting unit that sets input data including at least an appraisal difficulty level for each of the plurality of buildings, an appraisal time required for each of the plurality of buildings, an appraisal proficiency level for each of the plurality of visitors, and a desired visiting time frame for each of the plurality of buildings; a clustering calculation unit that performs a clustering process to divide the plurality of buildings corresponding to the position information into a plurality of clusters using position information of the plurality of buildings based on a predetermined cluster constraint condition, or that determines the destinations of each of the plurality of visitors by having a quantum computer connected via a network perform the clustering process; a route optimization problem calculation unit that causes the quantum computer to execute an optimization process for determining an order of visiting a plurality of buildings to be visited in each of the plurality of clusters based on a predetermined optimization constraint condition; Including, The predetermined cluster constraint condition is expressed by a formulated first formula, and the first formula includes a constraint term for a structure having a maximum appraisal difficulty to belong to a cluster of visitors having a maximum appraisal proficiency, a constraint term for a structure having an intermediate appraisal difficulty to belong to a cluster of visitors having an appraisal proficiency equal to or higher than the intermediate value, and a constraint term for dispersing a desired visiting time frame and an appraisal time required for route optimization; An information processing device, wherein the predetermined optimization constraint condition is expressed by a formulated second mathematical expression, and the second mathematical expression includes a constraint term related to the desired visiting time slot.

2. the clustering calculation unit executes a grid division process for dividing the image into a plurality of grids using only position information of the plurality of buildings before executing the clustering process; The information processing apparatus according to claim 1 , wherein the clustering process is executed for each of the plurality of grids.

3. 3. The information processing device according to claim 1 or 2, wherein the first formula further includes a cost term that reduces a travel distance taking into account paths between multiple structures included in a cluster that are destinations, a cost term that reduces a subtotal of an estimated time in a desired visiting time slot for each of multiple structures included in a cluster, and a constraint term that ensures that each of all clusters includes a predetermined number of destinations.

4. An information processing method for determining a plurality of buildings to be visited by each of a plurality of visitors and an order in which to visit the plurality of buildings, the information processing method being executed by a terminal device connected to a quantum computer via a network, setting input data including at least an appraisal difficulty level for each of the plurality of buildings, an appraisal skill level for each of the plurality of visitors, and a desired visiting time frame for each of the plurality of buildings; A step of determining destinations of each of the plurality of visitors by performing a clustering process using position information of the plurality of buildings based on a predetermined cluster constraint condition, to divide the plurality of buildings corresponding to the position information into a plurality of clusters, or by having the quantum computer perform the clustering process; causing the quantum computer to execute an optimization process for determining an order of visiting a plurality of buildings to be visited in each of the plurality of clusters based on a predetermined optimization constraint condition; Including, The predetermined cluster constraint condition is expressed by a formulated first formula, and the first formula includes at least a constraint term for a structure having a maximum appraisal difficulty to belong to a cluster of visitors having a maximum appraisal proficiency level, and a constraint term for a structure having an intermediate appraisal difficulty to belong to a cluster of visitors having an appraisal proficiency level equal to or higher than the intermediate appraisal proficiency level, An information processing method, wherein the predetermined optimization constraint condition is expressed by a formulated second mathematical expression, the second mathematical expression including at least a constraint term regarding the desired visiting time slot.

5. The step of determining the destinations of each of the plurality of visitors includes, before the clustering process, performing a grid division process for dividing the map into a plurality of grids using only position information of the plurality of buildings; The information processing method according to claim 4 , wherein the clustering process is performed for each of the plurality of grids.

6. 6. The information processing method according to claim 4 or 5, wherein the first formula further includes a cost term that reduces a travel distance taking into account paths between multiple structures included in a cluster that are visited, a cost term that reduces a subtotal of an estimated time in a desired visiting time slot for each of multiple structures included in a cluster, and a constraint term that ensures that each of all clusters includes a predetermined number of visited locations.

7. An information processing device according to any one of claims 1 to 3; a quantum computer connected to the information processing device via a network; Including, The quantum computer receives QUBOs obtained by converting the first formula and the second formula from the information processing device, and executes the clustering process and the optimization process.

8. An information processing program, which, when executed by a computer, causes the computer to function as each of the units of the information processing device according to any one of claims 1 to 3.

9. 7. An information processing program, which, when executed by a computer, causes the computer to carry out each step of the information processing method according to any one of claims 4 to 6.

Images