Delivery Location Selection Device

Abstract

To provide a delivery spot selection apparatus for properly selecting a spot to which a replacement fuel cartridge is to be delivered at predetermined timing.SOLUTION: In a delivery system, a processor 34 of a server serving as a delivery spot selection apparatus includes: a storage unit which stores, for each of multiple scheduled delivery spots where fuel cartridges filled with a predetermined fuel are to be delivered, collection history information indicating time and date at which a fuel cartridge which has been used is collected in a past predetermined period; and a delivery spot selection unit 42 which calculates, for each of the scheduled delivery spots, a collection occurrence probability of the scheduled delivery spot at predetermined timing by inputting the collection history information of the scheduled delivery spot to a predictor which is trained in advance, to calculate a collection occurrence probability which is a probability of occurrence of collection of a fuel cartridge which has been used, and selects, as a spot to which the fuel cartridge is delivered at the predetermined timing, a scheduled delivery spot having collection occurrence probability which is equal to or higher than a predetermined threshold, out of the multiple scheduled delivery spots.SELECTED DRAWING: Figure 4

Description

[Technical field]

[0001] The present invention relates to a delivery point selection device for selecting a point to which a fuel cartridge is to be delivered. [Background technology]

[0002] Fuel cartridges filled with fuels such as hydrogen or liquefied petroleum gas are used for various purposes. Since the amount of fuel filled in such a fuel cartridge is limited, a fuel cartridge with little fuel remaining must be replaced with a new fuel cartridge. Therefore, it is required that replacement fuel cartridges be delivered at an appropriate time to each location where the fuel cartridge is used.

[0003] The cart fuel management system disclosed in Patent Document 1 includes a hydrogen tank management room with a measuring means for managing the remaining amount in the hydrogen tank, and a dispatch room that dispatches golf carts based on data from the hydrogen tank management room, etc. When the remaining amount in the hydrogen tank after a round falls below a specified amount, the dispatch room orders a hydrogen tank from a fuel supplier. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2019-125006 A Summary of the Invention [Problem to be solved by the invention]

[0005] When a fuel cartridge is used in multiple locations, the rate at which the fuel in the fuel cartridge decreases may differ from location to location. Therefore, if fuel is delivered to a location where the remaining amount of fuel in the fuel cartridge falls below a certain amount each time an order is placed, the fuel cartridges will be delivered at different times for each location, which increases the effort required to deliver the fuel cartridges.

[0006] SUMMARY OF THE PRESENT EMBODIMENTS Accordingly, an object of the present invention is to provide a delivery point selection device capable of appropriately selecting a point to which a replacement fuel cartridge is to be delivered at a predetermined timing. [Means for solving the problem]

[0007] According to one embodiment, a delivery point selection device is provided, which includes a storage unit that stores collection history information indicating the date and time when a used fuel cartridge was collected during a predetermined period in the past for each of a plurality of scheduled delivery points to which a fuel cartridge filled with a predetermined fuel is scheduled to be delivered, and a delivery point selection unit that calculates the collection occurrence probability of the scheduled delivery point at a predetermined timing by inputting the collection history information of the scheduled delivery point into a predictor that has been trained in advance to calculate the collection occurrence probability of the used fuel cartridge being collected for each of the plurality of scheduled delivery points, and selects a scheduled delivery point from the plurality of scheduled delivery points whose collection occurrence probability is equal to or greater than a predetermined threshold as a point to which the fuel cartridge is to be delivered at the predetermined timing. Effect of the Invention

[0008] The delivery point selection device according to the present disclosure has the effect of being able to appropriately select a point to which a replacement fuel cartridge is to be delivered at a predetermined timing. [Brief description of the drawings]

[0009] [Figure 1] FIG. 1 is a schematic diagram of a fuel cartridge delivery system including a delivery point selection device. [Diagram 2] FIG. 2 is a schematic diagram of a delivery vehicle. [Diagram 3] FIG. 2 is a schematic configuration diagram of a server which is an example of a delivery point selection device. [Figure 4] FIG. 13 is a functional block diagram of a processor of the server relating to a delivery point selection process. [Diagram 5] FIG. 13 is a diagram for explaining an outline of a delivery point selection process. [Figure 6] 13 is an operational flowchart of a delivery point selection process. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0010] Hereinafter, a delivery point selection device, a delivery point selection method executed on the delivery point selection device, and a computer program for selecting a delivery point will be described with reference to the drawings. The delivery point selection device selects a point to which a replacement fuel cartridge is to be delivered at a scheduled timing for delivery of a replacement fuel cartridge, among a plurality of scheduled delivery points where a fuel cartridge is used. Hereinafter, the scheduled timing for delivery of a replacement fuel cartridge may be simply referred to as a scheduled delivery timing. For this purpose, the delivery point selection device collects collection information on the collection date and time and the collection location for each used fuel cartridge collected when a replacement fuel cartridge is delivered by a delivery vehicle. In addition, the delivery point selection device updates collection history information indicating the date and time when used fuel cartridges were received and the number of used fuel cartridges received during a predetermined period in the past (for example, several weeks to one year) for a plurality of scheduled delivery points based on the collection information. Furthermore, the delivery point selection device calculates, for each scheduled delivery point, a collection occurrence probability that collection of a used fuel cartridge occurs at the scheduled delivery timing of interest, and an expected number of fuel cartridges to be collected, based on the collection history information. Then, at the scheduled delivery timing, the delivery point selection device sends a delivery vehicle loaded with more than the predicted number of fuel cartridges to be collected to a scheduled delivery point among the scheduled delivery points where the collection occurrence probability is equal to or greater than a predetermined delivery threshold.

[0011] In this embodiment, the fuel cartridge is filled with hydrogen as fuel. For this purpose, the fuel cartridge is formed, for example, in a cylindrical shape, and has a storage tank equipped with a hydrogen storage material such as a hydrogen storage alloy or a high-pressure storage tank therein, and is provided with a mechanism for removing the filled hydrogen or for filling the fuel cartridge with hydrogen. In addition, the fuel cartridge is provided with an RFID tag including a non-volatile semiconductor memory that stores the identification number. Note that the fuel filled in the fuel cartridge is not limited to hydrogen, and may be any of various liquids or gases that can be used as fuel.

[0012] The planned delivery point of the fuel cartridge may be, for example, a location of an individual house, an apartment building, or a specific facility that uses the gas or liquid filled in the fuel cartridge as fuel, or a location in the vicinity thereof where a delivery vehicle can be temporarily parked.Furthermore, the planned delivery point may be a temporary store or an event venue where vehicles that use the gas or liquid filled in the fuel cartridge as fuel temporarily gather when some event is held, or a location in the vicinity thereof where a delivery vehicle can be temporarily parked.

[0013] FIG. 1 is a schematic diagram of a fuel cartridge delivery system including a delivery point selection device. In this embodiment, the delivery system 1 has at least one delivery vehicle 2 and a server 3 which is an example of a delivery point selection device. The delivery vehicle 2 is connected to the server 3 via the wireless base station 5 and the communication network 4 by accessing, for example, a wireless base station 5 connected to the communication network 4 to which the server 3 is connected via a gateway (not shown). Note that, although only one delivery vehicle 2 is illustrated in FIG. 1, the delivery system 1 may have a plurality of delivery vehicles 2. Similarly, a plurality of wireless base stations 5 may be connected to the communication network 4. In addition, the server 3 may be communicably connected to other devices (not shown) that manage the type, date and time, and location of an event, or other devices (not shown) that manage environmental information such as the weather, via the communication network 4.

[0014] 2 is a schematic configuration diagram of the delivery vehicle 2. The delivery vehicle 2 has a GPS receiver 11, a wireless communication terminal 12, a cartridge storage box 13, a wireless reader 14, a management device 15, and a display device 16.

[0015] The GPS receiver 11 receives GPS signals from GPS satellites at predetermined intervals and determines the self-position of the delivery vehicle 2 based on the received GPS signals. Then, the GPS receiver 11 outputs positioning information representing the positioning result of the self-position of the delivery vehicle 2 based on the GPS signals to the management device 15 via the in-vehicle network at predetermined intervals. Note that the delivery vehicle 2 may have a receiver that complies with a satellite positioning system other than the GPS receiver 11. In this case, it is sufficient that the receiver determines the self-position of the delivery vehicle 2.

[0016] The wireless communication terminal 12 is a device that executes wireless communication processing conforming to a predetermined wireless communication standard, and for example, by accessing the wireless base station 5, it is connected to the server 3 via the wireless base station 5 and the communication network 4. The wireless communication terminal 12 receives a downlink wireless signal from the wireless base station 5, and passes information contained in the wireless signal indicating a delivery route, etc., to the management device 15. The wireless communication terminal 12 also generates an uplink wireless signal that includes collection information received from the management device 15, in which identification information of each collected fuel cartridge, and the location and date and time of collection are recorded. The wireless communication terminal 12 then transmits the uplink wireless signal to the wireless base station 5, thereby transmitting the collection information to the server 3.

[0017] The cartridge storage box 13 is provided in the loading platform of the delivery vehicle 2, and is capable of storing a predetermined number of fuel cartridges 20. To this end, the cartridge storage box 13 is provided with a predetermined number of recessed storage sections 13a corresponding to the outer shape of the fuel cartridges 20. In this embodiment, as described above, since the fuel cartridge 20 has a cylindrical shape, the storage sections 13a are formed as cylindrical holes. Each storage section 13a stores one fuel cartridge 20.

[0018] The wireless reader 14 communicates with the RFID tag 21 provided on the fuel cartridge 20 accommodated in each of the accommodation sections 13a of the cartridge accommodation box 13 to read the identification information of the fuel cartridge 20. To this end, the wireless reader 14 has an antenna provided at a position facing the RFID tag 21 of the fuel cartridge 20 accommodated in the accommodation section 13a, and a communication circuit for wirelessly communicating with the RFID tag 21 via the antenna. For example, when the RFID tag 21 is provided at the end of the fuel cartridge 20, the antenna of the wireless reader 14 is provided on the bottom surface of the accommodation section 13a so as to face the RFID tag 21 of the accommodated fuel cartridge 20. Then, when the wireless reader 14 outputs a reading wireless signal via the antenna, the RFID tag 21 generates a response signal based on the power obtained by receiving the wireless signal, and outputs the generated response signal. The RFID tag 21 includes the identification information of the fuel cartridge 20 stored in the memory of the RFID tag 21 in the response signal. The wireless reader 14 receives the response signal via an antenna, and thereby reads the identification information of the fuel cartridge 20. The wireless reader 14 then outputs the read identification information to the management device 15.

[0019] The management device 15 has a processor and a memory, and executes a process related to the management of the fuel cartridge. For example, when the management device 15 receives identification information of the fuel cartridge 20 stored in the cartridge storage box 13 from the wireless reader 14, the date and time of receiving the identification information is set as the collection date and time of the fuel cartridge specified by the identification information. Every time the management device 15 receives identification information of the fuel cartridge 20 from the wireless reader 14, the management device 15 generates collection information including the received identification information, the collection date and time, and the location of the delivery vehicle 2. Then, the management device 15 stores the collection information in its own memory. Then, at a predetermined timing, the management device 15 transmits the collection information stored in the memory to the server 3 via the wireless communication terminal 12. Then, the management device 15 erases the collection information transmitted to the server 3 from the memory. The predetermined timing can be, for example, the timing when the current time becomes a preset transmission time, or the timing when the delivery vehicle 2 reaches a predetermined point. The predetermined location may be, for example, a location where there is a facility where fuel to be filled into fuel cartridges is stored and where used fuel cartridges can be removed from the delivery vehicle 2 and replacement fuel cartridges filled with hydrogen can be loaded onto the delivery vehicle 2. Hereinafter, such a facility will be referred to as a storage facility.

[0020] Furthermore, the management device 15 causes the display device 16 to display information indicating the delivery route, each delivery point, and the number of deliveries, which information has been received from the server 3 via the wireless communication terminal 12.

[0021] The display device 16 is provided in the driver's cab of the delivery vehicle 2, and displays information received from the management device 15 indicating the delivery route, each delivery point, and the number of deliveries.

[0022] The delivery vehicle 2 may be an autonomous vehicle. In this case, the management device 15 outputs information indicating the delivery route to a control device (not shown) that controls the driving of the delivery vehicle 2. The control device then controls the driving of the delivery vehicle 2 along the delivery route indicated in the received information, thereby moving the delivery vehicle 2 sequentially to each delivery point on the delivery route.

[0023] Next, the server 3, which is an example of a delivery point selection device, will be described. 3 is a hardware configuration diagram of a server 3, which is an example of a delivery point selection device. The server 3 has a communication interface 31, a storage device 32, a memory 33, and a processor 34. The communication interface 31, the storage device 32, and the memory 33 are connected to the processor 34 via signal lines. The server 3 may further have input devices (not shown) such as a keyboard and a mouse, and a display device (not shown) such as a liquid crystal display.

[0024] The communication interface 31 is an example of a communication unit, and has an interface circuit for connecting the server 3 to the communication network 4. The communication interface 31 is configured to be able to communicate with the delivery vehicle 2 via the communication network 4 and the wireless base station 5. That is, the communication interface 31 transmits information indicating the delivery route received from the processor 34 to the delivery vehicle 2 via the communication network 4 and the wireless base station 5. The communication interface 31 also passes collection information received from the delivery vehicle 2 via the wireless base station 5 and the communication network 4 to the processor 34.

[0025] The storage device 32 is an example of a storage unit, and includes, for example, a hard disk device or an optical recording medium and an access device therefor. The storage device 32 stores various data and information used in the delivery point selection process. For example, the storage device 32 stores individual collection information received from the delivery vehicle 2, location information of each scheduled delivery point, and collection history information for each scheduled delivery point. Furthermore, the storage device 32 may store, for each scheduled delivery point, event information such as the type, date and time, and location of an event that has been held or is scheduled to be held at or around the scheduled delivery point. Furthermore, the storage device 32 may store, for each scheduled delivery point, environmental information of the scheduled delivery point. Furthermore, the storage device 32 may store a computer program for executing the delivery point selection process, which is executed on the processor 34.

[0026] The memory 33 is another example of a storage unit, and includes, for example, a non-volatile semiconductor memory and a volatile semiconductor memory. The memory 33 temporarily stores various data generated during the execution of the delivery point selection process.

[0027] The processor 34 is an example of a control unit and has one or more CPUs (Central Processing Units) and their peripheral circuits. The processor 34 may further have other arithmetic circuits such as a logical arithmetic unit or a numerical arithmetic unit. The processor 34 executes the delivery point selection process.

[0028] 4 is a functional block diagram of the processor 34 related to the delivery point selection process. The processor 34 has a history update unit 41, a delivery point selection unit 42, and a delivery route setting unit 43. Each of these units in the processor 34 is a functional module realized by, for example, a computer program running on the processor 34. Alternatively, each of these units in the processor 34 may be a dedicated arithmetic circuit provided in the processor 34.

[0029] Each time the server 3 receives collection information from the delivery vehicle 2, the history update unit 41 updates the collection history information representing the date and time when used fuel cartridges were collected and the number of cartridges collected during a specified period in the past for each scheduled delivery point based on the collection information.

[0030] The history update unit 41 refers to the location of the delivery vehicle 2 when the fuel cartridge 20 was collected, which is included in the collection information, and identifies the scheduled delivery point closest to that location. The history update unit 41 then adds the collection date and time of the used fuel cartridge 20, which is included in the collection information, to the collection history information of the identified scheduled delivery point, and increments the number of cartridges collected by one. If a collection date and time within a predetermined time (for example, several minutes to several tens of minutes) from the collection date and time indicated in the collection information has already been registered in the collection history information, the history update unit 41 may increment the number of cartridges collected at the already registered collection date and time by one. This updates the collection history information of the identified scheduled delivery point. If the location of the delivery vehicle 2 indicated in the collection information is away from any scheduled delivery point by a predetermined distance or more, the history update unit 41 may add the location of the delivery vehicle 2 indicated in the collection information as a new scheduled delivery point. The history update unit 41 may then newly generate collection history information for the added scheduled delivery point. However, if the location of the delivery vehicle 2 indicated in the collection information is the location of a storage facility, the collection information is assumed to have been generated when the replacement fuel cartridge was stored. Therefore, the history update unit 41 does not add the location of the delivery vehicle 2 indicated in the collection information as a new scheduled delivery point.

[0031] Furthermore, the history update unit 41 may refer to the event information and determine whether or not an event was held within a predetermined distance (e.g., several hundred meters to 1 km) from the specified scheduled delivery point within a predetermined time (e.g., about 1 hour to 6 hours) before or after the collection date and time. If such an event was held, the history update unit 41 may include the type of the event held in the collection history information in association with the collection date and time.

[0032] Furthermore, the history update unit 41 may include information on the environment of the identified scheduled delivery point at the collection date and time of the fuel cartridge 20, which is included in the collection information, (e.g., weather or temperature, etc.) in the collection history information in association with the collection date and time. In this case, the history update unit 41 may obtain information on the environment of the scheduled delivery point via the communication network 4 from another device that manages environmental information.

[0033] The delivery point selection unit 42 predicts, for each scheduled delivery point, the probability of a fuel cartridge being recalled at the scheduled delivery time when a replacement fuel cartridge 20 is scheduled to be delivered, and the number of cartridges to be replaced. The scheduled delivery time is set, for example, at a fixed cycle. For example, the scheduled delivery time is set to the same time period every day or every few days, the same day of the week every week (for example, Monday and Thursday), or the same date and time every month (for example, the 10th, 20th, and 30th). Alternatively, the scheduled delivery time may be set irregularly. In this case, the scheduled delivery time may be set by the administrator inputting the scheduled delivery time via the input device of the server 3.

[0034] The delivery point selection unit 42 calculates the probability of fuel cartridge collection occurring at the scheduled delivery time for each scheduled delivery point by inputting the information recorded in the collection history information for that scheduled delivery point and the scheduled delivery time into the predictor. Then, the delivery point selection unit 42 selects a scheduled delivery point whose collection occurrence probability is equal to or greater than a predetermined delivery threshold as a delivery point for that scheduled delivery time.

[0035] The predictor is trained in advance to output the probability of collection occurrence for the input information. The predictor is configured as, for example, a multi-layer perceptron type neural network. In this case, the predictor has an input layer to which information recorded in the collection history information for the scheduled delivery point of interest and the scheduled delivery timing are input, an output layer to which the collection occurrence probability for the scheduled delivery point of interest is output, and a plurality of intermediate layers. The input layer has a plurality of nodes. For example, the interval between the scheduled delivery timing and the collection date and time of the fuel cartridge immediately before, and the interval between the collection date and time of each fuel cartridge and the collection date and time of the fuel cartridge immediately before during a predetermined period in the past are input to each node. Alternatively, the predictor may be configured so that the interval between each collection date and time recorded in the collection history information and the scheduled delivery timing is input to each node. Furthermore, the input layer may be provided with a node to which the number of collections at the collection date and time of each fuel cartridge is input. Furthermore, the input layer may be provided with a node to which information on events at the scheduled delivery timing and the collection date and time of each fuel cartridge is input. In this case, a different numerical value is assigned to each type of event, and a numerical value corresponding to the type of event that has been held or is scheduled to be held is input to the node. Furthermore, the input layer may be provided with a node to which environmental information at the scheduled delivery time and collection date and time of each fuel cartridge is input. In this case, a different numerical value is assigned to each type of environment represented by the environmental information, and a numerical value corresponding to the type of environment at the scheduled delivery time and collection date and time of each fuel cartridge is input to the node.

[0036] The multiple intermediate layers include at least one fully connected layer including multiple nodes that perform fully connected operations on values ​​output from each node in the upstream layer, and one or more activation layers that perform activation operations on outputs from each node in the fully connected layer. Note that the multiple intermediate layers may include a layer that performs product-sum operations on outputs from some nodes in the upstream layer.

[0037] The output layer also outputs the probability of collection by performing a softmax or sigmoid operation on the output from the most downstream intermediate layer. Furthermore, the output layer may be provided with a node that outputs the expected number of fuel cartridges to be collected at the scheduled delivery time, i.e., the expected number of fuel cartridges to be replaced. This node also performs a softmax or sigmoid operation on the output from the most downstream intermediate layer to calculate the probability for each number of fuel cartridges to be replaced, and outputs the number that maximizes the calculated probability as the expected number.

[0038] This predictor is trained in advance by a learning method such as the backpropagation method based on a large amount of training data including collection history information at various locations.

[0039] The predictor is not limited to the above example, and may be any device that is configured to output the probability of collection occurrence by inputting collection history information and scheduled delivery timing for a scheduled delivery point of interest.

[0040] Furthermore, the delivery point selection unit 42 may refer to the collection history information of each scheduled delivery point to identify a scheduled delivery point where the collection occurrence probability is less than the delivery threshold but the elapsed time since the last delivery of a replacement fuel cartridge exceeds a predetermined period.The delivery point selection unit 42 may then set the identified scheduled delivery point as one of the delivery points.This prevents a situation in which a replacement fuel cartridge is not delivered to any scheduled delivery point for a long period of time.

[0041] The delivery point selection unit 42 notifies the delivery route setting unit 43 of the planned delivery timing and each selected delivery point and the expected number of fuel cartridges to be collected at each delivery point.

[0042] The delivery route setting unit 43 sets a delivery route that passes through each delivery point notified by the delivery point selection unit 42. To this end, the delivery route setting unit 43 sets a delivery route that passes through each delivery point according to any method used to search for the shortest route that passes through multiple points. As such a method, the delivery route setting unit 43 can use, for example, a route search method based on a branch and bound method, a route search method based on a genetic algorithm, or a route search method based on a Boltzmann machine.

[0043] Alternatively, a candidate delivery route for each combination of delivery points may be set and stored in the storage device 32. In this case, the delivery route setting unit 43 may select a candidate delivery route that passes through each delivery point notified by the delivery point selection unit 42 from among the candidate delivery routes stored in the storage device 32 as the delivery route to be actually used.

[0044] In addition, when there are multiple delivery vehicles 2 available for delivery of the fuel cartridge at the scheduled delivery time, the delivery route setting unit 43 may set a delivery route for each of the multiple delivery vehicles 2. In this case, the delivery route setting unit 43 sets a delivery area for each delivery vehicle 2 in which the delivery vehicle 2 is responsible for delivery. Then, the delivery route setting unit 43 refers to the position of each delivery point and each delivery area, and identifies the delivery points included in the delivery area for each delivery area. Then, the delivery route setting unit 43 may set a delivery route for each delivery area by executing the above process for each delivery point included in the delivery area for each delivery area. Alternatively, the delivery route setting unit 43 may assign multiple delivery vehicles 2 to the same delivery route.

[0045] The delivery route setting unit 43 also calculates the total expected number of fuel cartridges to be collected at each delivery point on the delivery route. The delivery route setting unit 43 then adds a predetermined number of spare cartridges to the calculated total expected number, and sets the resulting number as the number of cartridges to be delivered.

[0046] The delivery route setting unit 43 notifies the delivery vehicle 2 of the scheduled delivery timing, the delivery route and each delivery point on the delivery route, and the number of deliveries via the communication interface 31, the communication network 4, and the wireless base station 5. Furthermore, the delivery route setting unit 43 may notify a management computer (not shown) of the storage facility of the scheduled delivery timing, the delivery route and each delivery point on the delivery route, and the number of deliveries.

[0047] FIG. 5 is a diagram for explaining an outline of the delivery point selection process. In the example shown in FIG. 5, ten scheduled delivery points A to J are set. Also, assume that the delivery threshold is 0.7. In the example of FIG. 5, the return occurrence probability calculated for each of these scheduled delivery points at the scheduled delivery timing of interest is 0.9, 0.6, 0.5, 0.8, 0.8, 0.4, 0.2, 0.7, 0.3, 0.5. Therefore, each of the scheduled delivery points A, D, E, and H, whose return occurrence probability is equal to or greater than the delivery threshold, is selected as a delivery point. Then, a delivery route 500 is set so as to pass through the delivery points A, D, E, and H.

[0048] 6 is an operational flowchart for explaining an outline of the delivery point selection process executed by the processor 34. When a predetermined time before a scheduled delivery timing of interest, the processor 34 may execute the delivery point selection process according to the following operational flowchart.

[0049] The delivery point selection unit 42 of the processor 34 calculates, for each scheduled delivery point, the probability of collection occurring at a scheduled delivery time of interest and the expected number of fuel cartridges to be collected based on collection history information for that scheduled delivery point (step S101).Then, the delivery point selection unit 42 selects, from among the scheduled delivery points, a scheduled delivery point whose collection occurrence probability is equal to or greater than a predetermined delivery threshold as a delivery point (step S102).

[0050] The delivery route setting unit 43 of the processor 34 sets a delivery route that passes through each selected delivery point (step S103). The delivery route setting unit 43 determines the number of delivery bottles to be obtained by adding a predetermined number of spare bottles to the total expected number of fuel cartridges to be collected at each selected delivery point (step S104). Then, the delivery route setting unit 43 notifies the delivery vehicle 2 of the scheduled delivery timing of interest, the delivery route, each delivery point on the delivery route, and the number of delivery bottles via the communication interface 31, the communication network 4, and the wireless base station 5 (step S105). Then, the processor 34 ends the delivery point selection process.

[0051] As described above, the delivery point selection device refers to the history of past fuel cartridge deliveries to each scheduled delivery point to calculate the probability of collection occurring at a scheduled delivery time of interest.The delivery point selection device then selects, among each scheduled delivery point, a scheduled delivery point with a collection occurrence probability equal to or greater than a predetermined delivery probability as a delivery point for delivering a replacement fuel cartridge at the scheduled delivery time of interest.As a result, the delivery point selection device can appropriately set the delivery timing of a replacement fuel cartridge for each scheduled delivery point.

[0052] According to a modified example, a worker who rides in delivery vehicle 2 and collects used fuel cartridges at each delivery point may notify server 3 of the number of fuel cartridges collected at each delivery point and the collection date and time as collection information via the worker's portable wireless terminal. In this case, delivery vehicle 2 does not need to be equipped with wireless reader 14.

[0053] According to another modified example, the predictor may be configured to output the timing at which the probability of collection is highest for each scheduled delivery point by inputting collection history information for that scheduled delivery point. In this case, the predictor may be configured as a multilayer perceptron type neural network, as in the above embodiment. The input layer is input with the interval between the collection date and time of each fuel cartridge and the collection date and time of the fuel cartridge immediately before that in a predetermined period in the past. Also, as in the above embodiment, the input layer may be provided with a node to which the number of collected cartridges at the collection date and time of each fuel cartridge is input. Furthermore, the input layer may be provided with a node to which information on events at the scheduled delivery timing and the collection date and time of each fuel cartridge is input. The output layer outputs the timing at which the probability of collection is highest by performing a softmax operation or a sigmoid operation on the output from the most downstream intermediate layer. As in the above embodiment, the predictor may be trained in advance by a learning method such as an error backpropagation method based on a large amount of teacher data including collection history information at various points.

[0054] According to this modification, the delivery point selection unit 42 can predict the timing when the probability of return is highest for each scheduled delivery point. Therefore, the delivery route setting unit 43 can set a delivery route for each scheduled delivery point so that the delivery vehicle 2 passes through the scheduled delivery point at the timing when the probability of return is highest at the scheduled delivery point.

[0055] In addition, a computer program for realizing the functions of the processor 34 of the server 3 according to the above embodiment or variant example may be provided in a form recorded on a computer-readable portable recording medium, such as a semiconductor memory, a magnetic recording medium, or an optical recording medium.

[0056] As described above, those skilled in the art can make various modifications to the embodiments within the scope of the present invention. [Explanation of symbols]

[0057] 1. Delivery System 2 Delivery vehicles 3 Server 4. Communication Network 5. Radio base station 11 GPS receiver 12 Wireless communication terminals 13 Cartridge storage box 14 Wireless Reader 15 Management device 16 Display device 20 Fuel Cartridge 21 RFID Tags 31 Communication Interface 32 Storage Device 33 Memory 34 processors 41 History Update Section 42 Delivery point selection section 43 Delivery route setting section

Claims

1. a storage unit that stores collection history information indicating the date and time when a used fuel cartridge was collected during a specified period of time in the past for each of a plurality of scheduled delivery points to which a fuel cartridge filled with a specified fuel is scheduled to be delivered; a delivery point selection unit that calculates the probability of collection of the planned delivery point at the scheduled delivery time by inputting the collection history information and the scheduled delivery time of the planned delivery point into a predictor that has been trained in advance to calculate a collection occurrence probability in which collection of the used fuel cartridge will occur for each of the multiple planned delivery points, and selects, from among the multiple planned delivery points, a planned delivery point whose collection occurrence probability is equal to or greater than a predetermined threshold as a point to which the fuel cartridge will be delivered at the scheduled delivery time; having For each of the multiple scheduled delivery points, the predictor calculates the collection occurrence probability for the scheduled delivery point by inputting into the predictor an interval between each of the individual dates and times when the used fuel cartridges were collected during the specified period, which is included in the collection history information for the scheduled delivery point, and the scheduled delivery timing, and the predictor is trained in advance according to a predetermined supervised learning method so as to calculate the collection occurrence probability for an input of an interval between each of the individual dates and times when the used fuel cartridges were collected, which is included in the past collection history information for each of the multiple points, and the scheduled delivery timing. Delivery point selection device.

2. a storage unit that stores collection history information indicating collection dates and times when used fuel cartridges were collected during a specified period of time in the past for each of a plurality of scheduled delivery points to which a fuel cartridge filled with a specified fuel is scheduled to be delivered; a delivery point selection unit that determines the timing for each of the plurality of scheduled delivery points by inputting an interval between each collection date and time during the predetermined period and the collection date and time immediately preceding each collection date and time, the interval being included in the collection history information for the scheduled delivery point, into a predictor that has been trained in advance to output the timing at which the probability of collection of used fuel cartridges occurring will be highest for each of the plurality of scheduled delivery points; a delivery route setting unit that sets a delivery route for each of the plurality of scheduled delivery points so as to deliver the fuel cartridge at the timing of the corresponding scheduled delivery point; having the predictor is trained in advance according to a predetermined supervised learning method so as to output the timing in response to an input of an interval between each of the individual dates and times at which the used fuel cartridges were collected, which are included in the past collection history information at each of a plurality of locations, and the collection date and time immediately preceding each of the individual collection dates and times; Delivery point selection device.

Images