WO2023238225A1

WO2023238225A1 - Route finding device, route finding method, and program

Info

Publication number: WO2023238225A1
Application number: PCT/JP2022/022934
Authority: WO
Inventors: 俊介金井; 正崇佐藤; 和陽明石; 麻悠山添; まな美小川
Original assignee: 日本電信電話株式会社
Priority date: 2022-06-07
Filing date: 2022-06-07
Publication date: 2023-12-14

Abstract

A route finding device according to an embodiment of the present invention is provided with a processor and a storage unit that stores a plurality of assignment patterns corresponding to a plurality of parameter combinations assigned to a power supply vehicle and a plurality of locations. The processor is provided with a condition processing unit, a data analysis processing unit, and a data determination unit. The condition processing unit extracts an assignment pattern satisfying a given constraint condition from among the plurality of assignment patterns. The data analysis processing unit optimizes an objective function for the patrol pattern of the power supply vehicle to the plurality of locations, in accordance with the extracted assignment pattern. The data determination unit determines the patrol route for the power supply vehicle from the optimization result. The condition processing unit has a function for performing a first step for extracting assignment patterns satisfying a static first constraint condition, and a function for performing a second step for further extracting an assignment pattern satisfying a dynamic second constraint condition from among the assignment patterns that satisfy the first constraint condition.

Description

Route searching device, route searching method, and program

One aspect of the present invention relates to a route searching device, a route searching method, and a program.

Communication infrastructure is required to operate stably even during disasters. In particular, there is an urgent need to secure power supplies. Telecommunications carriers are therefore considering a system in which multiple power supply vehicles are kept on standby and patrolled by power supply vehicles to disaster-affected bases (such as buildings with power outages).
In order to quickly restore services in the event of a disaster, it is essential to calculate routes for multiple power supply vehicles to visit multiple locations. However, simply calculating the route is not enough; there are various factors (parameters) such as the ability of the driver of the power supply vehicle, the ability of the worker supplying the power to work (such as ability to drive, ability to perform electrical work, etc.), and work style. ) needs to be taken into account. This appears to be a so-called NP-hard problem, and because the number of power supply vehicle deployment (touring) route patterns increases explosively, it takes a very long time to obtain a solution.
In order to solve this kind of difficulty, for example, a technique described in Non-Patent Document 1 is known.

Existing technology searches for power supply vehicle deployment routes using brute force to arrive at exact solutions. Therefore, it is necessary to evaluate and compare each tour route for all tour patterns, which tends to increase the amount of calculation. Even if you want to obtain an optimal solution rather than an exact solution, there are many parameters that can be assigned to a power supply vehicle, and it is necessary to calculate evaluation values from the assignable patterns to determine the optimal solution. It took a huge amount of time.
This invention was made in view of the above-mentioned circumstances, and aims to provide a technique that can shorten the time it takes to search for a route.

A route search device according to one aspect of the present invention includes a storage unit that stores a plurality of allocation patterns corresponding to combinations of a plurality of parameters allocated to a power supply vehicle and a plurality of bases, and a processor. The processor includes a condition processing section, a data analysis processing section, and a data determination section. The condition processing unit extracts a layout pattern that satisfies given constraint conditions from a plurality of layout patterns. The data analysis processing unit optimizes an objective function regarding a patrol pattern of power supply vehicles to a plurality of bases, using the extracted allocation pattern as a target. The data determination unit determines a tour route for the power supply vehicle based on the optimization result. The condition processing unit has a function of executing a first step of extracting an allocation pattern that satisfies a static first constraint condition, and extracts a layout pattern that satisfies a dynamic second constraint condition from an allocation pattern that satisfies the first constraint condition. and a function of executing a second step of further extracting a layout pattern.

According to one aspect of the present invention, it is possible to provide a technique that can shorten the travel route search time.

FIG. 1 is a diagram showing an example of a system including a tour route search device according to a first embodiment of the present invention. FIG. 2 is a flowchart showing an example of a processing procedure of the processor 11 shown in FIG. FIG. 3 is a diagram for explaining an example of a method for calculating the predetermined time in block S6 of FIG. FIG. 4 is a flowchart showing an example of the processing procedure in block S3 of FIG. FIG. 5 is a flowchart showing an example of the processing procedure in block S4 of FIG. FIG. 6 is a diagram for explaining the effects of the embodiment. FIG. 7 is a diagram showing an example of the calculated optimal solution. FIG. 8 is a diagram for explaining a specific example of the calculated approximate solution. FIG. 9 is a diagram for explaining a specific example of the calculated approximate solution.

Embodiments of the present invention will be described below with reference to the drawings.
<Configuration>
FIG. 1 is a diagram showing an example of a system including a tour route search device according to a first embodiment of the present invention. In FIG. 1, a tour route search device 10 includes a processor 11, a storage 12, an interface section 13, and a memory 14. That is, the tour route search device 10 is a computer, and is implemented as, for example, a personal computer or a server computer.

The interface unit 13 is connected to the network 100 and can, for example, access the database 2 and obtain information such as disaster situations. Further, the interface unit 13 outputs the itinerary pattern 3 generated by the itinerary route search device 10 in response to a request from, for example, an operator of a vehicle allocation center.

The storage 12 is, for example, a nonvolatile storage medium (block device) such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive). The storage 12 stores a layout pattern 12a in addition to basic programs such as an OS (Operating System) and device drivers, and programs for realizing the functions of the itinerary route search device 10.

The allocation pattern 12a stores a plurality of allocation patterns corresponding to combinations of a plurality of parameters allocated to a power supply vehicle that is a target of route search and a plurality of bases that are patrol destinations of the power supply vehicle.

The memory 14 in FIG. 1 is, for example, a RAM (Random Access Memory), and stores the cyclic pattern 14c calculated by the processor 11 in addition to the program 14a loaded from storage.

The processor 11 is an arithmetic unit such as a Central Processing Unit (CPU) or a Micro Processing Unit (MPU), and realizes its functions by a program loaded into the memory 14.

By the way, the processor 11 includes a condition processing section 111, a data analysis processing section 112, and a data determination section 115 as functional blocks (program modules) according to the embodiment. These functional blocks are processing functions realized by the processor 11 executing instructions included in the program 14a. That is, the tour route search device 10 of the present invention can also be realized by a computer and a program. It is possible to record programs on recording media such as optical media and distribute them. Alternatively, it is also possible to provide the program through a network.

The condition processing unit 111 extracts a layout pattern that satisfies the given constraint conditions from the layout patterns 12a stored in advance in the storage 12. Here, the constraint conditions are, for example, information given from the database 2 or the operator, such as the number of bases (buildings, etc.) visited by the power supply vehicle, the departure time of the power supply vehicle, the remaining power amount of the power supply vehicle, and the remaining power amount of the base. , the worker's permitted work hours, or the work that the worker can perform.

The data analysis processing unit 112 optimizes the objective function regarding the patrol pattern of the power supply vehicle to the plurality of bases, targeting the allocation pattern extracted by the condition processing unit 111. Examples of the objective function include [reducing the power outage time in a specific building to 0], [minimizing the power outage time as a whole], and the like. A plurality of objective functions may be prepared for calculating the itinerary route.

The data determination unit 115 determines the patrol route of the power supply vehicle based on the optimization result by the data analysis processing unit 112.

<Effect>
FIG. 2 is a flowchart showing an example of a processing procedure of the processor 11 shown in FIG. In the embodiment, the search for the route pattern of each power supply vehicle is treated as a scheduling problem in which the best solution (approximate solution/optimal solution) is extracted. Genetic algorithms are well-known as a solution to this type of problem.

In FIG. 2, the processor 11 sets an objective function. (Block S1). Here, the evaluation target value in the optimization calculation is set to, for example, (A). Next, the processor 11 sets constraints (block S2).

Next, the processor 11 divides the calculation target allocation pattern 12a into a plurality of patterns (for example, 1 to N) (block S3), and performs processing (1 to N) on each pattern group in parallel (block S4). Block S3 is a process of extracting a layout pattern that satisfies statically set constraints (first constraints) (STEP 1). Through this step, layout patterns inconsistent with the first constraint are excluded, and the search space for the tour route is reduced.

In block S4, an algorithm that can perform processing independently, such as a genetic algorithm or a branch and bound method, is adopted (STEP 2). The process of STEP 2 is a process of further extracting a layout pattern that satisfies the dynamic second constraint condition from the layout pattern extracted in STEP 1.

In block S4, when processing is started (subblock S41), evaluation values are extracted after a certain period of time (subblock S42), and common parameters (for example, power outage time, per person) are extracted while performing parallel processing. business hours, etc.) are periodically checked (block S5). If a value better than the evaluation target value A is obtained (sub-block S43), the process is continued and the process of sub-block S42 is repeated again.

If the evaluation value of the objective function is not better than the evaluation target value A as the specified value, that is, if the evaluation value is less than or equal to the predetermined value, the process is terminated midway (subblock S44). Furthermore, when all the processes for N=1 are completed (sub-block S45), the evaluation value (A) is updated and the process proceeds to the next block S6.
The processing in block S4 is performed in parallel on the N allocation pattern groups divided in block S3, thereby promoting reduction in calculation time.

By determining whether allocation is possible in multiple steps, the search space can be reduced (STEP 1). Further, common parameters are periodically checked while performing parallel processing, and it is determined whether each allocation pattern is optimal (STEP 2). With these two steps, the search space is reduced and the overall processing time can be completed in a short time.

In block S6, a specified time is confirmed, and once the specified time has elapsed, a part of the processing results are outputted and displayed on the monitor of the operator's terminal (FIG. 1), for example (block S7). When it is determined in block S6 that all processing is completed, a state in which all processing is completed is reached (block S8) and the processing is completed.

FIG. 3 is a diagram for explaining an example of a method for calculating the predetermined time in block S6 of FIG. 2. For example, in a graph plotting evaluation values over time, the time at which the evaluation values are saturated (saturation time) can be determined from past results and used as the specified time.

FIG. 4 is a flowchart illustrating an example of the processing procedure in block S3 of FIG. 2. After extracting the constraint parameters (block S11), the processor 11 checks the number of patterns (block S12). For example, if an optimal solution (exact solution) is required, an infinite value can be set as the number of patterns. If an approximate solution is sufficient, a finite default value can be set. The default value can also be calculated from, for example, the past constraint parameter amount and the number of patterns in which successful patterns were output.

Next, the processor 11 randomly assigns constraint parameters (block S13) and checks for inconsistencies from the assignment results (block S14). The inconsistency is, for example, a contradiction caused by assigning the type of patrol building (eg, high voltage) to the type of power supply vehicle (eg, only low voltage). In other words, it is not possible to patrol high-voltage buildings with low-voltage power supply vehicles. By eliminating such contradictory combinations, it becomes possible to reduce the search space for the itinerary route. Note that by implementing the process of block S14 as a separate module, it is also possible to flexibly respond to increases and decreases in combinations.

If there is a contradiction (yes in block S15), the processing procedure returns to block S13 again. If there is no contradiction ((absent) in block S15), the pattern obtained in block S13 is retained as an allocation pattern and stored in the storage 12 (block S16). The steps from block S13 to block S16 are repeated until the number of retained layout patterns reaches N (block S17). When the number of retained layout patterns reaches N, all layout patterns are retained (block S18).

FIG. 5 is a flowchart illustrating an example of the processing procedure in block S4 of FIG. 2. In block S4, optimization of the objective function based on the genetic algorithm is performed. In FIG. 5, the processor 11 determines the repetition generation (M) (block S21), and then executes the process of STEP 2 (FIG. 2) (block S22). As a method for determining the number of generations, for example, it is possible to calculate it from the past constraint parameter amount as a default value and the number of patterns that output successful patterns.

Next, the processor 11 extracts parameter A (block S23). There may be multiple parameters extracted here. If all of the extracted parameters A are the highest (block S24), the processor 11 checks the number of repetition generations (block S27), and if the number of generations reaches M (block S28), the process is completed. In block S24, regarding the parameter (power outage time), for example, power outage time = 0 time is the highest.

On the other hand, regarding the parameter (work hours per person), work hours per person = 6 hours is not the highest. Therefore, assuming that there is a better pattern, the processor 11 randomly changes the "parameters that are not at the highest level." Note that if there are multiple parameters that are not at the top level, one of them is fixed and changed at random.

If No in block S24, the processor 11 changes parameters other than the highest level as crossover or mutation. In the genetic algorithm, one-point crossover, two-point crossover, and mutation rate can be treated as separate processing so that they can be changed separately.

Next, the processor 11 determines the presence or absence of a past pattern (block S26), and if (absent), the processing procedure returns to block S22. If there are past patterns (all are present), the processing procedure reaches block S27.

<Effect>
6 and 7 are diagrams for explaining the effects obtained by this embodiment. As shown in Figure 6(a), with existing technology, it is necessary to allocate many parameters to the power supply vehicle and calculate evaluation values from the patterns that can be allocated to determine the optimal solution. There were cases where it took a huge amount of time.

On the other hand, as shown in FIG. 6(b), in the embodiment, the allocation possibility is processed in STEP 1, and patterns that do not match the common parameter from the evaluation value are periodically pruned from among the patterns that can be allocated. Accordingly, it is possible to calculate an optimal (approximate) solution while reducing processing time. FIG. 7 shows an example of the calculated approximate solution.

FIGS. 8 and 9 are diagrams for explaining specific examples of calculated approximate solutions. As shown in FIG. 8, in the patrol route calculated according to the embodiment, for example, worker changes are set at a building (north building) along the way. This prevents work from being concentrated on one worker, and enables efficient processing by distributing work among a plurality of workers, as shown in FIG. Therefore, the entire work can be completed in a short time. That is, it is possible to predict each processing time, level out the processing of the entire process, and shorten the processing completion time.

As described above, in the embodiment, by processing the patrol route of the power supply vehicle as a schedule problem and excluding patterns where the conditions do not match ("inconsistent points"), the allocation pattern to be searched (tour pattern) [STEP 1] Next, the number of cyclic patterns to be searched is reduced by evaluating the cyclic patterns that are being searched at regular intervals and interrupting the search for patterns whose evaluation results (*parameter A) are below a certain value [STEP 2 ]. Through the processing in STEP 1 and STEP 2, the number of cyclic patterns to be searched can be significantly reduced, and the search time can be shortened by pruning.

That is, in the embodiment, the search for a tour route is formulated as a schedule problem that extracts the best solution (approximate solution/optimal solution) for each power supply vehicle's tour pattern, and the allocation is determined in multiple steps (STEP 1, STEP 2). . This allows the search space to be reduced. Furthermore, by periodically checking the common parameters while performing parallel processing, the search space can be further reduced, and the overall processing time required to search for the itinerary route can be further reduced.

For these reasons, according to the embodiment, it is possible to provide a technique that can shorten the travel route search time. In addition, it will be possible to quickly propose a deployment route for power supply vehicles in the event of a disaster, contributing to reductions in operation. It is also possible to contribute to strengthening the BCP (Business Continuity Plan).

Note that the present invention is not limited to the above-described embodiments as they are, and in the implementation stage, the components can be modified and embodied without departing from the gist of the invention. Moreover, various inventions can be formed by appropriately combining the plurality of components disclosed in the above embodiments. For example, some components may be deleted from all the components shown in the embodiments. Furthermore, components from different embodiments may be combined as appropriate.

2...Database 3...Cycling pattern 10...Cycling route search device 11...Processor 12...Storage 12a...Allocation pattern 13...Interface section 14...Memory 14a...Program 14c...Cycling pattern 100...Network 111...Condition processing section 112...Data analysis processing Section 115...Data determination section.

Claims

a storage unit that stores a plurality of assignment patterns corresponding to combinations of a plurality of parameters assigned to the power supply vehicle and the plurality of locations;
and a processor;
The processor includes:
a condition processing unit that extracts a layout pattern that satisfies given constraints from the plurality of layout patterns;
a data analysis processing unit that optimizes an objective function regarding a patrol pattern of the power supply vehicle to the plurality of bases, using the extracted allocation pattern as a target;
a data determining unit that determines a patrol route for the power supply vehicle based on the optimization result;
The condition processing unit is
a function of executing a first step of extracting an allocation pattern that satisfies a static first constraint;
A route searching device comprising: a function of executing a second step of further extracting a layout pattern satisfying a dynamic second constraint condition from a layout pattern satisfying the first constraint condition.
The route search device according to claim 1, wherein the first step is a step of reducing the search space of the itinerary route by excluding a layout pattern inconsistent with the first constraint.
The route search device according to claim 1, wherein the second step is a step of interrupting the search for the itinerary route for an allocation pattern in which the evaluation value of the objective function is equal to or less than a predetermined value.
The route search device according to claim 1, wherein the data analysis processing unit optimizes the objective function based on a genetic algorithm.
The route search device according to claim 1, wherein the data analysis processing unit optimizes the objective function using a branch and bound method.
A patrol route search for generating a patrol route for the power supply vehicle using a computer including a processor and a storage unit that stores a plurality of assignment patterns corresponding to combinations of a plurality of parameters assigned to the power supply vehicle and a plurality of bases. A method,
an extraction step in which the processor extracts a layout pattern that satisfies given constraint conditions from the plurality of layout patterns;
a step in which the processor optimizes an objective function regarding a patrol pattern of the power supply vehicle to the plurality of bases, using the extracted allocation pattern as a target;
the processor determines a patrol route for the power supply vehicle from the optimization result,
The extraction process includes:
a step in which the processor extracts an allocation pattern that satisfies a static first constraint;
A route searching method comprising: the processor further extracting a layout pattern satisfying a dynamic second constraint condition from a layout pattern satisfying the first constraint condition.
A program that causes a computer to function as each section of the route search device according to claim 5.