Delivery plan creation system, delivery plan creation method, and delivery plan creation program

Abstract

To provide an art that reduces a possibility of having difficulty preparing a delivery schedule in the midst of a period of a continuous schedule.SOLUTION: A delivery schedule preparation system is configured that comprises: a residual quantity prediction unit that predicts a transition of a residual quantity of gas accumulated in a plurality of gas containers put at a plurality of delivery destinations; a delivery period setting unit that sets a period in which one or more gas containers are empty, and the residual quantity of the gas is not 0 as a delivery period on the basis of the transition as to each of the plurality of delivery destinations; and a delivery schedule preparation unit that prepares the delivery schedule for delivering the gas container to each delivery destination within the delivery period in each of the plurality of delivery destinations on the basis of an objective function evaluating at least a cost on a movement when getting back to a base via the plurality of delivery destinations from the base as a cost in the schedule period, and a cost when being not delivered to the delivery destination to be delivered within the delivery period starting before an end day of the schedule period and ending after the end day thereof.SELECTED DRAWING: Figure 8

Description

[Technical Field]

[0001] The present invention relates to a delivery plan creation system, a delivery plan creation method, and a delivery plan creation program. [Background technology]

[0002] Conventionally, a known method of using gas is to install a gas container in a store, a private home, or the like, and supply the gas in the gas container to a gas appliance. In such a method, a delivery person delivers the gas container to the store, private home, or the like and replaces it before the remaining gas in the gas container reaches zero. Generally, gas containers are installed in multiple locations, and the replacement timing of the gas containers may differ depending on the installation location, making it difficult to create an efficient delivery plan. Patent Document 1 is known as a technology for creating such a delivery plan for gas containers. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2014-199552 Summary of the Invention [Problem to be solved by the invention]

[0004] In the aforementioned Patent Document 1, a delivery period during which gas can be exchanged is identified by predicting gas usage, and a delivery plan is created to deliver to each customer within that delivery period. The delivery plan is created as a plan for a relatively long planning period, for example, two to six months. The planning period and the delivery period during which delivery to each customer is made are independent of each other. Therefore, there may be a delivery period that starts before the end date of the planning period and ends after the end date of the planning period, i.e., a delivery period that straddles the end date of the planning period. Since delivery plans are generally created by searching for a delivery plan with low cost, if there is a delivery period that straddles the end date of the planning period, a delivery plan that does not deliver during the delivery period that straddles the end date of the planning period is always created. If such a delivery plan is created, the number of delivery destinations to be delivered to after the planning period may increase excessively, which may make delivery difficult. The present invention has been made in consideration of the above-mentioned problems, and aims to provide a technique that reduces the possibility that it becomes difficult to create a delivery plan in one of consecutive planning periods. [Means for solving the problem]

[0005] In order to achieve the above-mentioned object, the delivery plan creation system includes a remaining quantity prediction unit that predicts the trend in the remaining quantity of gas accumulated in multiple gas containers installed at each of multiple delivery destinations; a delivery period setting unit that sets a delivery period as a period during which the remaining quantity of gas is below a threshold value and does not become zero based on the trend for each of the multiple delivery destinations; a planning period setting unit that sets a planning period that is the period for which a delivery plan is created; and a delivery plan creation unit that creates the delivery plan to deliver the gas container to each of the delivery destinations within the delivery period for each of the multiple delivery destinations based on an objective function that evaluates, as costs during the planning period, at least the cost of traveling from a base to the base via the multiple delivery destinations and the cost of not delivering to the delivery destination that should be delivered within the delivery period that starts before the end date of the planning period and ends after the end date.

[0006] The planning period is a period for creating a delivery plan. The delivery period is a period during which gas containers should be replaced for each individual delivery destination. Because the planning period and delivery period are independent of each other, the end date of the planning period does not necessarily coincide with the start and end dates of the delivery period. In other words, the delivery period may extend beyond the end date of the planning period. For delivery destinations for which the delivery plan extends beyond the end date of the planning period, delivery may be made within the planning period or after the planning period has elapsed. This process of not targeting a delivery destination within the planning period but targeting it for delivery after the planning period has elapsed is called delivery postponement. When such postponement occurs, the number of delivery destinations to be delivered to after the postponement increases, and in the long term, postponement may lead to reduced efficiency and more complicated delivery planning problems.

[0007] In a solution algorithm for a delivery planning problem that evaluates costs using an objective function and finds a delivery plan with the lowest cost, costs generally increase as the number of delivery destinations increases. Therefore, if postponement is simply allowed when evaluating the cost of a delivery plan within a planning period, postponement will always occur. Therefore, in the delivery plan creation unit, when a postponement occurs, a cost corresponding to the postponement is added and the cost is evaluated using the objective function. As a result, in the process of creating a delivery plan, it is possible to prevent a delivery destination whose delivery period straddles the end date of the planning period from being postponed. Therefore, it is possible to reduce the possibility that it will be difficult to create a delivery plan for one of the consecutive planning periods. [Brief explanation of the drawings]

[0008] [Figure 1] FIG. 1 is a diagram illustrating a delivery plan creation system. [Figure 2] FIG. 2A shows an example of consumption data, FIG. 2B shows remaining amount data, FIG. 2C shows delivery destination data, and FIG. 2D shows cost data. [Figure 3] 10 is a diagram illustrating the prediction of the transition of the remaining amount and the setting of the delivery period. FIG. [Figure 4] FIG. 10 is a diagram showing the relationship between a planning period and a delivery period. [Figure 5] 10 is a flowchart of a delivery plan creation process. [Figure 6] FIG. 10 is a diagram for explaining updating of a plan. [Figure 7] FIG. 10 is a diagram illustrating an example of a delivery plan. [Figure 8] FIG. 8A is a diagram for explaining the first component of the postponement cost, and FIG. 8B is a diagram for explaining the second component. [Figure 9] 10 is a flowchart of a process for solving a delivery planning problem. [Figure 10] FIG. 10A shows the empty delivery order, and FIG. 10B shows the process of generating the initial solution. [Figure 11] 11A to 11D are diagrams showing change patterns. [Figure 12] 12A and 12B are diagrams showing change patterns. DETAILED DESCRIPTION OF THE INVENTION

[0009] Here, the embodiments of the present invention will be described in the following order. (1) Configuration of delivery plan creation system: (2) Delivery plan creation process: (3) Other embodiments:

[0010] (1) Configuration of delivery plan creation system: Fig. 1 is a diagram showing the configuration of a delivery plan creation system 10. The delivery plan creation system 10 is a system that creates a delivery plan when delivering gas containers to multiple delivery destinations (in Fig. 1, delivery destinations 1 to M: M is an integer of 2 or more). At each delivery destination, there are a gas container 61, a measuring instrument 62, and a gas burner 63.

[0011] The gas container 61 is a container that stores LPG (Liquefied Petroleum Gas). The gas container 61 is connected to a measuring instrument 62 via a pipe. The measuring instrument 62 is connected to a gas burner 63 via a pipe. The gas burner 63 is a device that burns gas to achieve various functions, such as a stove or a water heater. The measuring instrument 62 is a device that measures the amount of gas consumed, and measures the amount of gas consumed, for example, by measuring the amount of gas passing through a pipe. When the gas burner 63 burns gas, the gas stored in the gas container 61 is supplied to the gas burner 63, and the measuring instrument 62 measures the amount of gas consumed. The measuring instrument 62 may be of any type as long as it is capable of measuring the amount of gas consumed. Furthermore, the amount of gas consumed may be measured in various ways.

[0012] The gas burner 63 may be any type of device that consumes gas. Multiple gas burners 63 may be used at one delivery destination. In this embodiment, multiple gas containers 61, i.e., two or more, are installed at each delivery destination. The capacity of the gas containers 61 installed at each delivery destination is not limited, and the number is not limited as long as it is two or more.

[0013] When the gas container 61 in use becomes empty (including almost empty), the gas supply source is switched to an unused gas container 61, and gas use continues. In this embodiment, a service is provided in which full (filled) gas containers 61 are delivered to the delivery destination and empty gas containers 61 are collected before all gas containers 61 become empty.

[0014] In order to provide such a service, the consumption amount measured by the meter 62 at each delivery destination is transmitted to the delivery plan creation system 10. In this embodiment, the meter 62 has a communication function to identify the consumption amount. That is, the meter 62 installed at each delivery destination transmits the gas consumption amount at each delivery destination to the delivery plan creation system 10 at any timing. The transmission timing is not limited, but in this embodiment, the gas consumption amount on a daily basis is transmitted to the delivery plan creation system 10. The mode of communication is not limited, and wired communication, wireless communication, etc. may be used. Also, the measurement results by the meter 62 may be transmitted from a mobile terminal, etc.

[0015] The administrator terminal 50 is a terminal used by an administrator who manages delivery plans. The administrator terminal 50 includes a display unit (not shown), such as a display, and an input unit, such as a keyboard and a mouse. The administrator can operate the input unit to give various instructions to the delivery plan creation system 10. For example, the administrator can record various data in the delivery plan creation system 10, or give instructions to the delivery plan creation system 10 to create a delivery plan. The administrator can also visually check various pieces of information displayed on the display unit. For example, the administrator can visually check delivery plans, etc. created by the delivery plan creation system 10.

[0016] The delivery plan creation system 10 includes a control unit 20 including a CPU, RAM, ROM, etc., a storage medium 30, and a communication unit 40. The communication unit 40 is a device that communicates with each measuring instrument 62 at each delivery destination.

[0017] Various programs and various data are recorded on the storage medium 30. In this embodiment, the storage medium 30 stores consumption data 30a, remaining amount data 30b, delivery destination data 30c, worker data 30d, and cost data 30e.

[0018] The consumption data 30a is information showing the change in consumption at each delivery destination, and in this embodiment, shows the actual change in consumption up to the day before the current day. For example, as shown in FIG. 2A, the consumption data 30a is defined by associating information showing daily consumption with identification information (1 to M, etc.) of the delivery destination. Note that the unit of consumption may be any unit, and for example, cubic meters or the like can be used.

[0019] The remaining amount data 30b is information indicating the change in the remaining amount of gas at each delivery destination. In this embodiment, the remaining amount data 30b indicates the change in the remaining amount of gas up to the day before the current day. For example, as shown in FIG. 2B, the remaining amount data 30b is defined by associating information indicating the remaining amount for each day with the identification information of the delivery destination. In this embodiment, the remaining amount of gas is the total amount of gas stored in the gas containers 61 installed at each delivery destination. Therefore, the numerical value of the remaining amount data 30b is the sum of the remaining amounts of gas stored in each of the multiple gas containers 61. When an empty gas container 61 is replaced with a full gas container 61 during delivery, the remaining amount for that day increases. The presence or absence of a delivery may be determined using various methods. For example, when a delivery vehicle delivering gas containers 61 completes its day's work, the remaining amount at the delivery destination where the delivery vehicle made the delivery may increase by a value equivalent to the number of replaced gas containers 61 multiplied by the remaining amount of the full gas container 61. The unit of the remaining amount may be any unit, such as cubic meters.

[0020] The delivery destination data 30c is information relating to each of delivery destinations 1 to M. In this embodiment, the identification information of the delivery destination includes information indicating the location of the delivery destination (address, coordinates, etc.), the type and number of gas containers 61 installed at the delivery destination, the time required to replace the gas containers 61, and the first component of the postponement cost. FIG. 2C is a diagram showing an example of the delivery destination data 30c. In this embodiment, the type of gas container is indicated by a weight such as 20 kg or 50 kg, and the maximum amount of gas that can be filled for each type is determined. Therefore, once the type is identified, the remaining amount of gas that will increase upon replacement is determined.

[0021] The time required to replace the gas container 61 indicates the length of time required at each delivery destination to replace the gas container 61. For example, the time required from arrival at the delivery destination to departure after replacing the gas container 61 is identified and defined by statistics or the like. Since the time required to replace the gas container 61 may differ for each delivery destination, in this embodiment, the time required to replace the gas container 61 is determined for each delivery destination. Note that in this embodiment, an objective function is defined to minimize the total time required for delivery, and time is evaluated as cost. Since the time required to replace the gas container 61 is evaluated as cost, it is defined in units of time (minutes in FIG. 2C ).

[0022] The first component of the postponement cost is a value used to calculate the cost to be added to the objective function when postponing delivery to a destination outside the planning period, rather than within the planning period, and will be described in detail later. The first component of the postponement cost is also determined for each destination.

[0023] The worker data 30d is information about a worker who provides a service to replace the gas container 61. In this embodiment, this service is realized by the worker driving a delivery vehicle with a full gas container 61 loaded on the loading platform, leaving a base, visiting multiple delivery destinations, replacing the empty gas container 61 at each delivery destination, and returning to the base. In this embodiment, it is assumed that there are multiple workers.

[0024] In this embodiment, the working hours of each worker are determined in advance, and the allowable overtime hours are also determined in advance as the time during which work can be done outside of working hours. In addition, in this embodiment, the worker's holidays are also determined in advance. In this embodiment, it is assumed that no work is performed on holidays. In this embodiment, it is assumed that the working hours and allowable overtime hours of all workers are the same for all days, and that the holidays of all workers are also the same. However, these times and holidays may differ for each worker, or may differ from day to day (for example, by day of the week). In this embodiment, information indicating the working hours, allowable overtime hours, and holidays is included in the worker data 30d.

[0025] In this embodiment, one worker drives one delivery vehicle. Therefore, the number of workers corresponds to the number of delivery vehicles that can depart from the base. Note that the worker data 30d may include data that associates each worker's identification information (such as name or number) with the delivery vehicle's identification information (such as registration number).

[0026] The cost data 30e is information indicating the cost required to travel between two locations. In this embodiment, the cost indicates the travel time required to travel between the two locations. In this embodiment, the locations are bases or delivery destinations. The cost is determined in advance for all two locations that can be selected from all locations, and is defined as the cost data 30e.

[0027] For example, if the points for which costs are defined are a base and delivery destinations 1 to M, the costs are defined as shown in Figure 2D. That is, the base or any of delivery destinations 1 to M is set as the departure point, and the base or any of delivery destinations 1 to M is set as the arrival point, and travel times are determined for all combinations of points that can be used when traveling from the departure point to the arrival point, and these are taken as costs. In Figure 2D, the parts marked with "-" correspond to combinations where the departure point and arrival point are the same, and are therefore not defined.

[0028] Travel time may be defined in various ways. In this embodiment, a route between two locations is assumed, and the travel time for traveling along this route is determined based on statistics, etc. Of course, travel time may also be defined for each day of the week or time period. In this embodiment, even if the combination of departure and arrival locations is the same, travel time may vary depending on which location is the departure location. Therefore, travel time is also defined for cases where the departure location is specified. For example, for two locations, delivery destination 1 and delivery destination M, both the cost when delivery destination 1 is the departure location and delivery destination M is the arrival location and the cost when delivery destination M is the departure location and delivery destination 1 is the arrival location are defined, and the cost when delivery destination M is the departure location and delivery destination 1 is the arrival location are defined. However, if the difference between the two is small, the cost may be defined without distinguishing between the departure location and the arrival location. The formats of the consumption data 30a, remaining amount data 30b, delivery destination data 30c, worker data 30d, and cost data 30e described above are not limited to the above examples. For example, each data does not have to be recorded in the format shown in Figures 2A to 2D, and various formats can be adopted.

[0029] The control unit 20 executes various programs stored in the storage medium 30 or ROM. The control unit 20 can execute a delivery plan creation program as an example of such a program. When the delivery plan creation program is executed, the control unit 20 functions as a remaining amount prediction unit 20a, a delivery period setting unit 20b, a plan period setting unit 20c, and a delivery plan creation unit 20d.

[0030] The remaining amount prediction unit 20a has a function of predicting the transition of the remaining amount of gas stored in a plurality of gas containers 61 installed at each of a plurality of delivery destinations. To realize this function, the control unit 20 uses the function of the remaining amount prediction unit 20a to communicate with the measuring device 62 at each delivery destination at a predetermined timing and acquires the daily gas consumption amount at each delivery destination. Once the daily gas consumption amount at each delivery destination is acquired, the control unit 20 associates each consumption amount with the identification information of each delivery destination and records it as consumption amount data 30a in the storage medium 30. As a result, the daily consumption amount is recorded as shown in FIG. 2A.

[0031] Furthermore, the control unit 20 acquires the remaining amount of gas at each delivery destination using the function of the remaining amount prediction unit 20a. That is, the control unit 20 acquires the remaining amount up to the previous day for each delivery destination by referring to the remaining amount data 30b, and updates the remaining amount by subtracting the consumption amount for that day from the remaining amount. For example, in the example shown in FIGS. 2A and 2B, the consumption amount for delivery destination 1 on January 2, 2022 is 0.3 cubic meters, and the remaining amount of gas for delivery destination 1 on January 1, 2022 is 48 cubic meters. In this case, the control unit 20 acquires the remaining amount of gas for delivery destination 1 on January 2, 2022 as 47.7 cubic meters (=48-0.3). The control unit 20 performs such calculations for each delivery destination.

[0032] In this manner, by accumulating the consumption data 30a and the remaining amount data 30b, the transition of the remaining amount of gas in the past is identified, and the control unit 20 predicts the transition of the remaining amount of gas after the present, i.e., in the future, using the function of the remaining amount prediction unit 20a. Prediction of the transition of the remaining amount can be performed by various methods. In this embodiment, the control unit 20 predicts the transition of the remaining amount by predicting the future daily consumption.

[0033] That is, the control unit 20 refers to the remaining amount data 30b and predicts future daily consumption amounts from the statistical values ​​of the remaining amounts for each delivery destination for each past day. The statistical values ​​may be obtained in various ways, such as predicting daily consumption amounts based on the average of all past days, predicting daily consumption amounts based on the average of past days of the week, or various other ways may be adopted. Here, the explanation will continue using an example in which consumption amounts for each day of the week are predicted.

[0034] Once the consumption for each day of the week is predicted, the control unit 20 uses the function of the remaining amount prediction unit 20a to predict the continuous change in the remaining amount from now on. For example, if the day after the present is Saturday, the remaining amount for Saturday is predicted by subtracting the consumption amount for Saturday from the current remaining amount. Also, the remaining amount for Sunday is predicted by subtracting the consumption amount for Sunday from the predicted remaining amount for Saturday. By performing such predictions for each day after the present, it is possible to predict the change in the remaining amount for each day until the remaining amount of gas becomes zero. Figure 3 is a diagram showing an example of the change in the predicted remaining amount. In Figure 3, the horizontal axis is time (number of days elapsed) and the vertical axis is the remaining amount, and it shows the predicted value of the change in the remaining amount at a certain delivery destination. In Figure 3, the change in the remaining amount after the prediction start date is change R1.

[0035] The delivery period setting unit 20b has a function of setting a delivery period as a period during which the remaining amount of gas is below a threshold but does not reach zero, based on the transition for each of multiple delivery destinations. The delivery period is a period during which the gas container 61 must be replaced at each delivery destination. When the gas container 61 is replaced during the delivery period, the remaining amount of gas increases, but as time passes, the remaining amount decreases, and a delivery period in which the gas container 61 must be replaced again arrives. If multiple delivery periods are included within the planning period, which is the period for which a delivery plan is created, it is necessary to identify the multiple delivery periods within the planning period. Therefore, the control unit 20 acquires the length of the planning period. Specifically, the planning period setting unit 20c has a function of setting the planning period, and the length of the planning period is acquired by the planning period setting unit 20c.

[0036] In this embodiment, the planning period is a period of time that includes multiple delivery periods at at least one delivery destination. The length of the planning period may be determined by various methods, and in this embodiment, it is determined in advance by the administrator operating the administrator terminal 50. Once the length of the planning period is determined by the administrator, the control unit 20 sets the period from the current day to that length as the planning period and considers it to be the period for which a delivery plan is created. The lengths of the planning period and delivery period are not limited, but may be, for example, six months, the delivery period may be, for example, five days, and delivery periods may occur approximately once a month.

[0037] In this embodiment, the delivery period is a period during which the remaining amount of gas is equal to or less than the threshold value but does not reach zero. Furthermore, in this embodiment, the delivery period is a period during which one or more gas containers are empty but the remaining amount of gas does not reach zero. Therefore, the threshold value is specified as the remaining amount when one or more gas containers 61 are empty, and in this embodiment, the control unit 20 determines the threshold value for each delivery destination based on the delivery destination data 30c. That is, as shown in FIG. 2C, the type and number of gas containers 61 differ for each delivery destination, so the control unit 20 determines the threshold value based on the type and number.

[0038] The threshold value may be determined as the remaining amount when one or more gas containers 61 are empty, and may be one or more gas containers 61 that are empty and therefore subject to replacement. However, here, an example is assumed in which the threshold value is determined as the remaining amount when only one usable gas container 61 remains. For example, for delivery destinations 1 and 2 shown in FIG. 2C , the control unit 20 sets the threshold value to a value at which the remaining amount is less than the full amount of a 50 kg gas container 61. For delivery destination M shown in FIG. 2C , the control unit 20 sets the threshold value to a value at which the remaining amount is less than the full amount of a 20 kg gas container 61. When setting the threshold value to ensure that the gas containers 61 are emptied, it is preferable to set the threshold value Th with a margin provided so that the remaining amount of usable gas containers 61 is less than the value corresponding to the remaining amount of one gas container 61.

[0039] Furthermore, the gas container 61 needs to be replaced before the remaining amount of gas reaches 0. Therefore, in this embodiment, a lower limit value is set in advance for the remaining amount of gas. To prevent the remaining amount of gas in the gas container 61 from reaching 0, it is preferable to set the lower limit value Tm with a margin provided so that the remaining amount is greater than 0.

[0040] Based on the predicted transition of the remaining gas quantity for each delivery destination, the control unit 20 acquires the date on which the remaining quantity will be equal to or less than the threshold value Th as the start date of the delivery period, and the date on which the remaining quantity will be equal to or less than the lower limit value Tm as the end date of the delivery period. Specifically, the control unit 20 acquires the first delivery period after the present based on the predicted transition of the remaining quantity after the present. In FIG. 3, the threshold value Th and the lower limit value Tm are indicated by dashed lines. In this example, the control unit 20 acquires the start date Ds1 of the delivery period and the end date De1 of the delivery period based on the transition R1 of the remaining quantity after the predicted start date.

[0041] Next, the control unit 20, using the function of the remaining amount prediction unit 20a, assumes that the remaining amount of gas will increase by the accumulated amount in the gas container 61 to be replaced on any day within the first delivery period, and thereafter, the remaining amount of gas will decrease based on the predicted change from the increased remaining amount, and predicts the change in the remaining amount after replacement. In Figure 3, the dashed-dotted arrow indicates the change in the remaining amount when an empty gas container 61 is replaced on replacement date Dr1 within the first delivery period. The increase in the remaining amount indicated by the dashed-dotted arrow is "the number of replaced gas containers 61 multiplied by the remaining amount in the full gas container 61."

[0042] In this embodiment, the threshold value Th for each delivery destination is a fixed value, and the number of gas containers 61 replaced at the time of replacement is constant. That is, in this embodiment, an alternating replacement method is adopted in which a predetermined number of empty gas containers 61 are replaced with full gas containers 61 at the time of replacement, and used and unused gas containers 61 are not replaced. In this case, the increase in the remaining amount due to replacement of the gas container 61 is always constant. For example, even if the replacement date is different from Dr1 shown in FIG. 3, the increase in the remaining amount does not change.

[0043] Therefore, the transition of the remaining amount after replacement does not depend on the replacement date of the gas container 61. In Figure 3, the transition of the remaining amount after replacement is shown by transition R2. As shown in Figure 3, transition R2 is a graph in which transition R1 is shifted upward by the amount of gas increase at the time of replacement. Therefore, if it is assumed that on any day within the delivery period, the remaining amount of gas increases by the amount of gas stored in the gas container 61 to be replaced, and thereafter the remaining amount of gas decreases based on the transition predicted from the increased remaining amount, the transition of the remaining amount can be predicted by shifting by the amount of increase.

[0044] Therefore, the control unit 20 determines the first delivery period based on the transition R1 of the remaining amount from the present onward, the threshold value Th, and the lower limit value Tm. Furthermore, the control unit 20 obtains the second delivery period based on the transition R2 obtained by shifting the transition R1 upward once by an amount corresponding to the increase amount due to the replacement, and the threshold value Th and the lower limit value Tm. That is, the control unit 20 obtains the second delivery period based on the transition R2 obtained by shifting the transition R1 upward (N-1) times. N The Nth delivery period is obtained based on the threshold value Th and the lower limit value Tm, where N is an integer equal to or greater than 1. The control unit 20 performs this process to obtain the delivery period for each delivery destination included in the planned period.

[0045] FIG. 3 shows start dates Ds1, Ds2, and Ds3 and end dates De1, De2, and De3 for a total of three delivery periods. As described above, in this embodiment, by adopting the alternating exchange method, delivery periods within a planning period of any length can be determined in advance before the process of solving the delivery planning problem begins, and the start and end dates can be set to fixed values. FIG. 4 is a diagram comparing an example of a planning period and delivery periods for multiple delivery destinations. The horizontal axis in FIG. 4 represents time, and different delivery destinations are lined up vertically. Furthermore, solid arrows represent delivery periods, and delivery periods for the same delivery destination are lined up horizontally at the same vertical position.

[0046] The dashed arrows are examples of planned periods. The end date of the planned period does not necessarily coincide with the start and end dates of the delivery period. For example, in the example shown in FIG. 4, the start date Ds of the delivery period T2 at the end of the planned period for delivery destination 2 is 42 is a day before the end date Pe of the planning period. 42 is a day after the end date Pe of the planning period. A delivery period in this kind of relationship is called a delivery period that straddles the end date of the planning period. For example, the third delivery period T3 for destination 3 shown in Figure 4 is also a delivery period that straddles the end date of the planning period. With the alternating exchange method, the start and end dates of delivery periods can be fixed before the process of solving the delivery planning problem begins, including delivery periods that straddle the end date of the planning period.

[0047] On the other hand, if a full replacement method is adopted in which all gas containers 61 are replaced, the start and end dates of the delivery period cannot be fixed before the process of solving the delivery planning problem begins. In other words, with the full replacement method, all gas containers 61 are filled to capacity with each replacement, and the remaining amount increases to the maximum remaining amount for each delivery destination. In Figure 3, the maximum remaining amount is shown as Gmax. If all gas containers 61 are replaced on replacement date Dr1 within the first delivery period, the remaining amount will return to Gmax, and it is thought that from that day onwards the remaining amount of gas will decrease according to the predicted trend.

[0048] In Figure 3, the change in the remaining amount Rall is shown by a two-dot chain line. This change Rall is the change when the item is replaced on the replacement date Dr1. If the replacement date is different, the date on which the item returns to Gmax will also be different, and the change will also be different. Therefore, the second delivery period cannot be determined unless the replacement date for the first delivery period is determined. The replacement date is a variable determined by setting a tentative solution during the process of solving the delivery planning problem and searching for a solution that reduces cost. Therefore, in the total replacement method, the delivery period cannot be determined before the process of solving the delivery planning problem begins. On the other hand, in this embodiment, the alternating replacement method is adopted, so the start and end dates of the delivery period can be determined before the process of solving the delivery planning problem begins and can be fixed during the process of solving the delivery planning problem.

[0049] The delivery plan creation unit 20d has a function of creating a delivery plan for replacing empty gas containers at each of a plurality of delivery destinations within the delivery period for each of the delivery destinations during a planning period. In this embodiment, the delivery plan is created by solving a delivery plan problem. The delivery plan problem is a problem in which an objective function for evaluating cost is defined and a solution is determined so as to minimize cost within various constraints, and various algorithms can be used.

[0050] When searching for a solution with a low cost, the control unit 20 varies the variables. Generally, the more variables there are, the more complex the delivery planning problem becomes. If there are variables in a relationship where one variable determines another variable, the delivery planning problem becomes extremely complex, and if there are too many such variables, it becomes extremely difficult to solve the delivery planning problem.

[0051] When the gas container 61 is replaced using the total replacement method, as described above, the second delivery period cannot be determined unless the replacement date for the first delivery period is determined. The same applies to the third and subsequent delivery periods; in order to determine these delivery periods, the previous replacement dates must already be determined. Therefore, when solving a delivery planning problem using the total replacement method, the second delivery period depends on the first replacement date, and the third replacement date, which is determined within the third delivery period, depends on the first and second replacement dates. In this way, the (N+1)th replacement date depends on the first to Nth replacement dates. Furthermore, because the cost depends on all of these replacement dates, the objective function of the delivery planning problem takes on a very complex form. For this reason, it becomes extremely difficult to derive a solution to the delivery planning problem, especially when there are a large number of delivery destinations or a large number of delivery periods included in the planning period.

[0052] However, in this embodiment, the alternating exchange method is adopted. As a result, the start and end dates of the delivery periods for multiple delivery destinations and multiple delivery periods can be determined and fixed before the process of solving the delivery plan problem begins. Therefore, this embodiment makes it much easier to solve the delivery plan problem compared to the full exchange method. Furthermore, if the start and end dates of the delivery periods can be fixed, the delivery plan problem can be solved even if the planning period is extended to include multiple delivery periods. Therefore, a delivery plan can be created without excessively shortening the planning period. Furthermore, the longer the planning period, the more likely it is that the delivery plan can be optimized over a long period. In the case of a short planning period, a delivery plan optimized for the short planning period is created, which may result in unreasonable delivery plans after the planning period, and the efficiency of the subsequent delivery plans may decrease. However, this embodiment makes it possible to optimize the delivery plan over a relatively long period, thereby increasing the possibility of creating a delivery plan with high delivery efficiency.

[0053] (2) Delivery plan creation process: Next, the delivery plan creation process will be described in detail. The administrator operates the input unit of the administrator terminal 50 to input information on delivery destination data 30c, worker data 30d, and cost data 30e, and stores the information in the storage medium 30. With these data stored in the storage medium 30, the administrator uses the input unit to instruct the start of the delivery plan creation process. Here, it is assumed that at the start of the process, consumption amount data 30a is not stored, and remaining amount data 30b indicates the maximum remaining amount at all delivery destinations (the state immediately after gas containers 61 have been delivered to all delivery destinations). However, the initial values ​​of consumption amount data 30a and remaining amount data 30b are not limited to these values.

[0054] 5 is a flowchart showing the delivery plan creation process. When the delivery plan creation process starts, the control unit 20 acquires consumption data using the function of the remaining amount prediction unit 20a (step S100). That is, the control unit 20 communicates with the meter 62 at each delivery destination via the communication unit 40 and acquires the daily gas consumption amount for each delivery destination. The acquired consumption amount is associated with the identification information of the delivery destination and stored in the storage medium 30 as consumption amount data 30a.

[0055] Next, the control unit 20 calculates the remaining amount using the function of the remaining amount prediction unit 20a (step S105). That is, the control unit 20 references the consumption amount data 30a and obtains the most recent one-day gas consumption amount for each delivery destination. The control unit 20 also obtains the most recent remaining amount of gas by subtracting the one-day gas consumption amount from the remaining gas amount for the previous day at each delivery destination indicated by the remaining amount data 30b, and updates the remaining amount data 30b with the obtained value.

[0056] Next, the control unit 20 determines whether it is time to create a plan (step S110). In this embodiment, conditions for executing the process of creating a delivery plan are determined in advance, and the control unit 20 determines whether it is time to create a plan based on whether the conditions are met. The conditions for executing the process of creating a delivery plan may be various conditions. For example, the condition may be the passage of a certain period of time, the acquisition of the latest consumption and remaining amounts, the completion of a predetermined number of deliveries to a predetermined number of delivery destinations, or various other conditions. Note that when step S110 is executed for the first time after the delivery plan creation process is started, it is determined that it is time to create a plan.

[0057] In either case, the conditions are set so that the conditions are satisfied within a period shorter than the planned period. If it is determined that it is time to create a plan, a delivery plan is created in the processing from step S115 onwards. Therefore, after starting operation of an already created delivery plan, the control unit 20 updates the delivery plan based on a delivery period set based on the latest prediction of the remaining gas amount before the planned period has elapsed. Figure 6 is a diagram illustrating an example in which the timing for creating a plan occurs every week. Here, it is assumed that a delivery plan for planned period 1 is first created and operation of the delivery plan is started. In this case, for the first week of planned period 1, gas containers 61 are delivered according to the created delivery plan.

[0058] One week after the start of operation of planning period 1, it is time to create a plan, and a delivery plan for planning period 2 is created. In this case, delivery plan 2 is created taking into account the delivery results for one week of planning period 1 and the consumption amount at each delivery destination, and the original delivery plan 1 is updated with the created delivery plan 2. Thereafter, deliveries are made using the updated delivery plan 2. Thereafter, the delivery plan is updated every time it is time to create a plan.

[0059] In this embodiment, a delivery period is set based on a remaining amount prediction to create a delivery plan, so if there is a discrepancy between the predicted remaining amount and the actual remaining amount, the delivery plan may become inappropriate. In particular, when a delivery plan is created for a relatively long planning period, a discrepancy between the predicted remaining amount and the actual remaining amount is likely to occur toward the end of the planning period. However, in this embodiment, the plan is created before the end of the planning period, so a prediction of the remaining amount based on the latest consumption amount is obtained before the end of the planning period, making it less likely to cause a discrepancy.

[0060] Returning to the explanation of Fig. 5, if it is not determined in step S110 that it is time to create a plan, the control unit 20 returns to step S100 and continues to collect consumption amounts, etc. On the other hand, if it is determined in step S110 that it is time to create a plan, the control unit 20 sets a plan period using the function of the plan period setting unit 20c (step S115). That is, the control unit 20 acquires the length of the plan period that is input in advance by the administrator and stored in the storage medium 30, RAM, etc., and sets the plan period.

[0061] Next, the control unit 20 predicts the transition of the remaining gas amount using the function of the delivery period setting unit 20b (step S120). That is, the control unit 20 obtains a predicted value of daily consumption based on the consumption data 30a for each delivery destination, and predicts the transition of daily consumption from the present onward for each delivery destination by sequentially subtracting the consumption from the most recent remaining amount indicated by the remaining amount data 30b. This transition is used to determine the first delivery period.

[0062] Furthermore, the control unit 20 shifts the predicted consumption trend by the amount of the remaining amount that increases due to exchange in the alternating exchange method, and assumes that the same trend will continue, thereby obtaining a trend for determining the second delivery period. By repeating the same shift, the control unit 20 predicts the trend of the remaining amount necessary to determine the delivery period within the planned period. As a result, the trend of the remaining amount as shown in Figure 3 is obtained. The control unit 20 performs this process for each delivery destination.

[0063] Next, the control unit 20 sets a delivery period for each delivery destination using the function of the delivery period setting unit 20b (step S125). That is, the control unit 20 references the delivery destination data 30c and acquires a threshold value Th indicating the upper limit of the remaining amount of gas for each delivery destination based on the type and number of gas containers 61 installed at each delivery destination. The control unit 20 also acquires a lower limit value Tm of the remaining amount of gas for each delivery destination. Then, based on the transition of the remaining amount of gas acquired in step S120, the control unit 20 acquires a period during which the remaining amount of gas is equal to or less than the threshold value Th and equal to or greater than the lower limit value Tm. As a result, a delivery period such as that shown in FIG. 3 is acquired for each delivery destination, and as shown in FIG. 4, multiple delivery periods included in the planned period and delivery periods spanning the end date of the planned period are acquired for multiple delivery destinations.

[0064] Since the delivery period only needs to be fixed before the process for solving the delivery planning problem begins, the delivery period obtained using the threshold value Th and the lower limit value Tm may be corrected before the process for solving the problem begins. Various corrections are possible. For example, if the delivery period includes a holiday, the delivery period may be extended to offset the days on which delivery was effectively not possible due to the holiday. That is, the control unit 20 references the worker data 30d and compares the delivery period obtained using the threshold value Th and the lower limit value Tm with the holiday. If the delivery period includes a holiday, the control unit 20 moves the start date of the delivery period forward to an earlier date.

[0065] The number of days to be advanced is not limited, but may be, for example, the same as the number of holidays included in the delivery period, or may be the number of days obtained by multiplying the number of holidays included in the delivery period by a predetermined coefficient. When extending the delivery period due to holidays or the like, it is preferable to set the threshold value Th so that empty gas containers 61 can be replaced in the alternating replacement method even if the delivery period is extended. That is, it is preferable to set the threshold value Th with a predetermined margin so that the gas container 61 to be replaced is empty even if the delivery period is extended. It is also acceptable for the gas container 61 that should be empty on the delivery date to the delivery destination to not be empty due to the delivery period being extended. However, if such a situation occurs, it is preferable to update the remaining amount data 30b in accordance with the replacement and recreate the delivery plan before the next delivery period for the delivery destination.

[0066] Next, the control unit 20 executes a process for solving the delivery plan problem using the function of the delivery plan creation unit 20d (step S130). In this embodiment, the delivery plan problem is solved by searching for a solution that minimizes the cost evaluated by a predetermined objective function. In this embodiment, the delivery plan is implemented by determining the delivery destinations and the order in which deliveries should be made from departure from a base to return to the base on each day within the planning period.

[0067] Figure 7 is a diagram showing a schematic view of a delivery plan. A delivery plan can be created for each of the start and end dates of a planning period. In Figure 7, the start date of the planning period is set to day 1, and the end date of the planning period is set to day Xmax, and the delivery plans (delivery order to delivery destinations) that can be created for each day are shown from left to right.

[0068] The squares in Figure 7 indicate bases that are the departure and arrival points for delivery. Therefore, the points indicated by squares drawn in different vertical and horizontal positions in Figure 7 are usually the same point, but may be different points. Also, the circles in Figure 7 indicate delivery destinations, and the direction of travel during delivery is indicated by arrows. Therefore, in Figure 7, the delivery order from a base to multiple delivery destinations in order and back to the base is indicated by the vertically arranged squares and circles.

[0069] In this embodiment, a series of processes in which a delivery vehicle departs from a base, makes a delivery, and then returns to the base is called a rotation. In this embodiment, multiple workers are available to work on the same day, and multiple rotations can be performed by each worker working on the same day. Note that although the same worker may be able to perform multiple rotations (i.e., they may be able to perform work that involves arriving at a base and then departing again), an example will be described here in which each worker is responsible for a maximum of one rotation per day. When creating a delivery plan, the destinations to which deliveries should be made and the order in which they should be made for each rotation on each day within the planning period are determined. However, there is an upper limit to the maximum number of rotations that can be performed on the same day. Figure 7 shows an example in which the maximum number of rotations on all days is K. In other words, there can be a maximum of K departures and arrivals on each day.

[0070] The objective function can be defined in various ways, but in this embodiment, the following objective function is adopted. (Total working hours of workers) + Mt (Total overtime hours of workers) + (Total postponement costs) Here, the total working hours of a worker is the sum of the time required to work from the base via the delivery destinations to the return to the base for all rotations. In other words, the total working hours of a worker is the sum of the time required for all workers to work for all days. The time required to work from the base via the delivery destinations to the return to the base can be calculated by adding the travel time between points and the work time at each delivery destination.

[0071] In Figure 7, the travel time of the point for rotation 1 on the first day is time t m1 ~t mm , the working time at each delivery destination is time t w0 ~t wm+1 The travel time between locations is the time described in the cost data 30e, and the work time at the delivery destination is the time required to replace the gas container described in the delivery destination data 30c. Note that, here, the work time at the base (for example, the time required for filling, loading, and unloading the gas container 61) is taken into consideration, but if work at the base is not required, the work time may be zero.

[0072] The total overtime hours of a worker is the sum of the overtime hours of each worker for each day for all days. Specifically, the overtime hours are the value obtained by subtracting the working hours from the time required for each worker to perform the work. The total overtime hours of the workers is obtained by obtaining and adding up the overtime hours for all workers for all days. The working hours of the workers are described in the worker data 30d.

[0073] In the objective function, the total overtime hours of the workers is multiplied by a coefficient Mt. The coefficient Mt is a value greater than 1 and is set to minimize overtime hours as much as possible. To more efficiently reduce overtime hours, it is preferable to set the coefficient Mt to a value significantly greater than 1, such as 100. By incorporating the coefficient Mt, the cost associated with travel outside of the worker's working hours can be made greater than the cost associated with travel within the working hours. As a result, even if the total working hours of the workers are the same, there will be a difference in cost between when part of the work within working hours is done during overtime hours and when part of the work is done outside of overtime hours, with the former resulting in a higher cost.

[0074] With this configuration, the objective function can be defined so that a solution with less overtime is sought. Therefore, according to this embodiment, it is possible to create a delivery plan that reduces overtime. Note that, if the overtime hours in this example exceed the allowable overtime hours, the cost may be further multiplied by a coefficient greater than the coefficient Mt and incorporated into the objective function. With this configuration, the objective function can be defined so that a solution with less work exceeding the allowable overtime hours is sought.

[0075] Postponement costs are costs incurred when a delivery plan is created so that delivery is not made during a delivery period that straddles the end date of the planning period to a delivery destination that should be delivered during that delivery period. The final delivery period T2 for delivery destination 2 shown in Figure 4 straddles the end date Pe of the planning period. In this way, for a delivery destination whose delivery plan straddles the end date of the planning period, delivery may be made within the planning period or after the planning period has elapsed. The latter case, in which delivery is not made within the planning period but after the planning period has elapsed, is called postponement of delivery.

[0076] If the postponement cost is not taken into account in an algorithm that minimizes the cost indicated by the objective function, for example, if the objective function of this embodiment does not include the term (sum of postponement costs), then as the number of delivery destinations increases, the travel time to those delivery destinations increases, resulting in an increase in cost. Therefore, when attempting to minimize the cost indicated by the objective function, a plan that includes postponement will always be the solution. However, when postponement is performed, the number of delivery destinations to be delivered to after the postponement increases, and in the long term, postponement may lead to reduced efficiency and the complication of the delivery planning problem. Therefore, creating a delivery plan that always includes postponement cannot be said to be creating an optimal delivery plan.

[0077] Therefore, in this embodiment, a postponement cost is introduced into the objective function. In this embodiment, the postponement cost when postponing to each delivery destination is determined based on a cost value that is reduced when delivery to a delivery destination that should be delivered within the delivery period is not made within the planned period compared to when delivery is made, and a value that is smaller when the number of days in the delivery plan included in the period before the end date of the planned period is large compared to when it is small.

[0078] The cost value that is reduced compared to when delivery to a delivery destination that should be delivered within the delivery period is not made within the planned period is, in other words, the cost value that is reduced compared to when delivery is postponed compared to when delivery is not made. Figure 8A is a diagram for explaining such costs. The left side of Figure 8A schematically shows the delivery order of a delivery plan that returns to the delivery destination from a base via three delivery destinations. In this example, the time required to travel from the first delivery destination P1 to the second delivery destination P2 is 8 minutes, and the time required to travel from the second delivery destination P2 to the third delivery destination P3 is 10 minutes. In addition, the work time at the second delivery destination is 6 minutes. Therefore, the time required to travel from the first delivery destination P1 to the third delivery destination P3 is 24 minutes.

[0079] The right side of Figure 8A shows a schematic diagram of a delivery plan when the second delivery destination P2 is removed from the example on the left. In this example, the time required to travel from the first delivery destination P1 to the third delivery destination P3 is 15 minutes. In the above-mentioned objective function, the total sum of worker working hours is the cost, so in the example shown on the left side of Figure 8A, the cost of the time required to travel from the first delivery destination P1 to the third delivery destination P3 is 24 minutes. On the other hand, in the example shown on the right side of Figure 8A, the cost of the time required to travel from the first delivery destination P1 to the third delivery destination P3 is 15 minutes. Therefore, the difference between the two is 9 minutes.

[0080] In the example on the right, the second delivery destination P2 has been removed from the example on the left. If the removal of the second delivery destination P2 is considered to be removal due to postponement, the example on the right is an example in which delivery to a delivery destination that should be delivered within the delivery period is not made within the planned period. On the other hand, the example on the left is an example in which delivery is made. The cost value in the right case is reduced by 9 minutes compared to the left case. If postponement as in the left case simply reduces the cost value by 9 minutes, as described above, delivery destinations that can be postponed will always be postponed during the process of creating the delivery plan. However, if a cost value of about 9 minutes, which is the cost value reduced by postponement in the case of postponement, is introduced into the objective function as a postponement cost, it is possible to prevent the search for a solution that always results in postponement. In this embodiment, this cost value is called the first component of the postponement cost.

[0081] On the other hand, the number of days in the delivery period included in the planning period varies for each delivery destination, as it depends on the length of the planning period and the amount of gas consumed at the delivery destination. Furthermore, even for the same delivery destination, the number of days in the delivery period included in the planning period may differ if the planning period is different. Therefore, in this embodiment, the first component of the postponement cost is adjusted based on a value that is smaller when the number of days in the delivery plan included in the period before the end date of the planning period is large than when it is small. In the example shown in FIG. 4, the number of days in the delivery plan included in the period before the end date of the planning period is defined, for example, in the delivery period T2 that straddles the end date Pe of the planning period, and is defined as the start date Ds of the delivery period. 42 The number of days from the start date to the end date Pe of the planning period.

[0082] When creating a delivery plan, the greater the number of candidate delivery dates to the delivery destination, the greater the degree of freedom in formulating the delivery plan, making it easier to create an efficient plan. On the other hand, the fewer the number of candidate delivery dates to the delivery destination, the less the degree of freedom in formulating the delivery plan, making it more difficult to create an efficient plan. Therefore, when the delivery period straddles the end date of the planning period, it is preferable that the postponement cost be smaller when the delivery period includes a large number of days than when the delivery period includes a small number of days.

[0083] Therefore, in this embodiment, a second component is defined in advance, which has the property that it takes a larger value when the number of days in the delivery period included in the planning period is large than when the number of days is small. The postponement cost is calculated by multiplying this second component by the first component of the postponement cost. FIG. 8B is a diagram showing an example of the second component C(t). FIG. 8B shows the second component C(t) with the number of days in the planning period t on the horizontal axis and the magnitude of the second component on the vertical axis. In the example shown in FIG. 8, when the number of days in the planning period t is 1, the second component (t) is at its maximum value (>1), and as the number of days in the planning period t increases, the second component C(t) monotonically decreases. Furthermore, when the number of days in the planning period t is equal to or greater than t0, the value of the second component C(t) becomes constant.

[0084] When the postponement cost is defined by multiplying the first component by the second component C(t) having such characteristics, the postponement cost is defined to have a characteristic such that, with the magnitude of the first component as the base magnitude, the cost increases as the number of days in the planning period increases and decreases as the number of days in the planning period decreases. Furthermore, when the number of days in the planning period exceeds a reference value t0, the cost becomes a small, constant value, and the postponement cost is defined so that the cost does not depend on the number of days. Therefore, by creating a delivery plan based on an objective function including such a postponement cost, the postponement cost can be defined so that the longer the delivery period included in the planning period, the less likely postponement will occur, and the fewer days in the delivery period included in the planning period, the more likely postponement will occur. As a result, it is possible to prevent difficulty in creating a delivery plan only within or after the planning period. In other words, it is possible to reduce the possibility that a delivery plan in only one of the consecutive planning periods will be excessively difficult.

[0085] Various methods can be used to determine the first component of the postponement cost. For example, a method can be used in which a delivery plan is created based on an objective function in which the postponement cost is not introduced, and a process is performed for each delivery destination to statistically calculate the cost reduction that would occur if a specific delivery destination were removed from the delivery plan.

[0086] Furthermore, various methods can be employed to determine the second component C(t). For example, a delivery plan is created based on an objective function in which postponement costs are not introduced, and the cost that is reduced by removing a specific delivery destination from the delivery plan is obtained. Then, the value of the cost / first component can be statistically obtained to obtain the second component C(t). When statistically obtaining the value of the cost / first component, for example, t0 is temporarily set, and the second component C(t) is set as a linear function when the second component C(t) is equal to or less than t0, and the second component C(t) is set as a constant when the second component C(t) is equal to or greater than t0. A configuration can be employed in which the linear function and the constant are calculated using the least squares method or the like, and the temporarily set t0 is changed and the same process is repeated, and the result that minimizes the squared error is regarded as the second component C(t).

[0087] The sum of postponement costs is calculated for all postponed delivery destinations, i.e., all delivery destinations for which delivery is not scheduled to occur during a delivery period that overlaps the end date of the planning period. That is, the first component of the postponement cost for each delivery destination is identified based on the delivery destination data 30c. Furthermore, the number of days from the start date of the delivery period to the end date of the planning period is identified, and the value of the second component C(t) is identified based on this number of days. The two are then multiplied to calculate the postponement cost, and the sum of the postponement costs is obtained by adding up all the postponement costs calculated for each delivery destination. Furthermore, the postponement cost can take various forms other than the product of the first and second components, as long as it is a cost to be added to the objective function when delivery is not made within the planning period to a delivery destination that should be delivered within the delivery period.

[0088] The process of solving the delivery planning problem is performed by searching for a solution that minimizes the cost indicated by the objective function within the scope of satisfying the constraints. The constraints can be various conditions, but the constraints in this embodiment include the following conditions: Rotational speed constraints Delivery period restrictions Delivery amount constraints Delivery date constraints

[0089] The turnover constraint is a constraint that there is an upper limit to the number of turnovers per day. This constraint can be introduced, for example, by measuring the number of turnovers per day for a solution candidate, and if there is a day on which the number of turnovers exceeds the maximum value K, performing processing to not consider the candidate as a solution. This turnover constraint makes it possible to create a delivery plan that does not exceed the constraints of resources (such as the number of workers and the number of delivery vehicles) available per day.

[0090] The delivery period constraint is a constraint that each of multiple delivery destinations must be delivered once within one delivery period. This constraint can be implemented, for example, by identifying a delivery date for each delivery destination based on solution candidates and performing a process to check whether the plan is for one delivery within all delivery periods to all delivery destinations. However, there are exceptions to this rule when delivery is postponed. In other words, in a delivery period where the delivery period straddles the end date of the planned period, the number of deliveries may be zero. This delivery period constraint can be used to restrict the creation of delivery plans that do not deliver gas containers 61 to delivery destinations or that result in unnecessary deliveries.

[0091] The delivery volume constraint is a constraint that there is an upper limit to the number of gas containers 61 that can be delivered in one rotation. In other words, since there is an upper limit to the load capacity of a delivery vehicle, there is an upper limit to the number of gas containers 61 that can be transported in one rotation. The delivery volume constraint can be introduced, for example, by identifying the number of gas containers 61 required to complete each rotation based on candidate solutions and performing a process to confirm whether the number of gas containers 61 is equal to or less than the upper limit for all rotations. The delivery period constraint can impose a constraint to prevent the creation of an unrealizable delivery plan that requires a number of gas containers 61 that cannot be transported by the delivery vehicle.

[0092] The delivery date constraint is a constraint that delivery is not performed on holidays. In other words, a constraint is placed on the date on which a delivery plan is created so that workers do not have to work on holidays. The delivery date constraint can be implemented, for example, by identifying the delivery date on which delivery is planned to be performed based on the candidate solutions and performing a process to confirm whether the delivery date matches a holiday. The delivery date constraint can be used to restrict the creation of delivery plans that involve work on holidays.

[0093] In this embodiment, the control unit 20 executes the process of solving the delivery planning problem in step S130 by searching for a solution that reduces the cost indicated by the objective function while satisfying the constraints described above. That is, the control unit 20 maintains predetermined constants before the solution-finding process, and changes the variables by changing the solution during the solution-finding process to search for a solution that reduces the cost.

[0094] In this embodiment, the constants are various values ​​specified in the delivery destination data 30c, worker data 30d, and cost data 30e described above, the planning period set in step S115, the delivery period set in step S125, the maximum number of rotations K, and the number of gas containers 61 that can be delivered in one rotation. In this embodiment, the variables are variable elements in the delivery plan. That is, they are information indicating the order in which deliveries are made to each of the multiple delivery destinations and the date on which the deliveries are made. The delivery order to each delivery destination also includes information indicating which rotation each delivery destination will receive. Therefore, whether there are 0 to K rotations per day, and which rotation delivers to which delivery destination may change.

[0095] 9 is a flowchart showing the process of solving the delivery plan problem. In the process of solving the delivery plan problem, the control unit 20 sets the candidate delivery plan as a tentative solution and searches for a solution with a lower cost while evaluating the cost of the tentative solution using an objective function. When the process of solving the delivery plan problem starts, the control unit 20 generates an initial solution for the tentative solution using the function of the delivery plan creation unit 20d (step S200). That is, the control unit 20 generates the initial solution for the tentative solution by adding each delivery destination to the delivery order departing from and returning to the base on each day within the delivery period so as to satisfy predetermined constraints.

[0096] Specifically, this is implemented by provisionally determining the delivery order of the delivery destinations for each day and each rotation within the planning period, as shown in Figure 7. Note that the delivery plan shown in Figure 7 is an example, and the number of rotations may be less than K on some days. When generating an initial solution, constraints may be taken into account, or some constraints may not be taken into account, but in order to obtain a solution quickly, it is preferable that at least some constraints are satisfied. Here, we will consider an example of generating an initial solution for a provisional solution that satisfies the rotation number constraint and the delivery volume constraint. Of course, delivery period constraints and delivery date constraints may also be imposed.

[0097] Various methods can be used to generate an initial solution for such a provisional solution. For example, an "empty" delivery order is generated in which there are no delivery destinations between bases, and delivery destinations are inserted into the "empty" delivery order so as to satisfy the turnover constraint and the delivery volume constraint. Specifically, the control unit 20 first generates K "empty" delivery orders corresponding to the maximum turnover value K for each day of the delivery period, as shown in FIG. 10A.

[0098] The control unit 20 also references the delivery destination data 30c, acquires the delivery destinations of the delivery targets, randomly extracts and arranges the delivery destinations, and adds each delivery destination to the "empty" delivery order. That is, the control unit 20 processes each delivery destination in the order extracted randomly, and adds each delivery destination to the delivery order one by one. When focusing on the process of determining the delivery order for a certain delivery destination, the control unit 20 evaluates the cost of adding the delivery destination to all possible orders in the delivery order while satisfying the rotation speed constraint and the delivery volume constraint. The cost is evaluated using the objective function described above.

[0099] The number of rotations that can be performed in a day must be equal to or less than the maximum value K, and can be any number between 0 and K. Therefore, the control unit 20 distinguishes daily rotations by identifier. If a rotation exists, it is identified by the value of a variable, such as rotation 1, rotation k, etc. In this embodiment, when increasing the number of rotations, the number of rotations is increased so that identifier k becomes a consecutive number. Therefore, if the delivery orders for identifiers i and (i-1) are both "empty," a delivery destination can be added to the delivery order for identifier (i-1), but a delivery destination cannot be added to the delivery order for identifier i. If there is one or more delivery destinations in the rotation for identifier (i-1), a delivery destination can be added to the delivery order for identifier i. If such additions are performed under the condition that the maximum number of rotations is equal to or less than K, the rotation number constraint condition is satisfied.

[0100] Furthermore, if the delivery order of the delivery destinations is added so that the total number of gas containers 61 exchanged at the delivery destinations in each rotation is less than or equal to the number of gas containers 61 that can be delivered in one rotation, the delivery volume constraint condition will be satisfied.

[0101] The control unit 20 assumes that delivery destinations have been added in all possible orders for adding delivery destinations under these constraints. Figure 10B shows the process of generating an initial solution for a tentative solution, and horizontal black arrows indicate delivery orders in which a delivery destination can be added for the state shown in Figure 10B. Note that delivery destinations cannot be added to delivery orders indicated by horizontal white arrows.

[0102] The control unit 20 evaluates the cost using the above-mentioned objective function for all cases in which a delivery destination is added to each of the delivery orders that can be added. Then, among all cases in which a delivery destination is added to each of the delivery orders that can be added, the control unit 20 adds the delivery destination to the delivery order with the smallest cost. The control unit 20 performs this process for all of the delivery destinations to generate an initial solution for the tentative solution. Such an initial solution can generate an initial solution that is not excessively costly, thereby increasing the possibility of obtaining a solution quickly. Furthermore, since a solution that satisfies at least some of the constraints becomes the initial solution, it is easier to search for a solution in the solution-finding process compared to when the constraints are not considered in the initial solution.

[0103] Once a tentative solution is obtained, the control unit 20 acquires the cost indicated by the objective function in the tentative solution that satisfies the constraints, and repeats the process of determining as the latest tentative solution the solution that reduces the cost indicated by the objective function when the tentative solution is modified using a pattern selected from a plurality of predetermined change patterns.To this end, the control unit 20 first randomly selects one change pattern from a plurality of predetermined change patterns (step S205).

[0104] A change pattern is a method for changing a solution, and six change patterns are defined in this embodiment. The first change pattern is a pattern in which the delivery order is swapped within a delivery order that returns to the base via multiple delivery destinations from a base. FIG. 11A is a diagram for explaining the first change pattern. In FIG. 11A, as in FIG. 7, bases are represented by squares, delivery destinations by circles, and the delivery order is represented by arrows. Also, in FIG. 11A, the delivery order before the change is shown on the left side, and the delivery order after the change is shown on the right side. The above notation method is also used in FIGS. 11B to 12B. However, while the departure point and the arrival point are represented by the same square in FIG. 11A, in other diagrams such as FIG. 11B, the departure point and the arrival point are represented by different squares. As shown in FIG. 11A, the first change pattern is a pattern in which the delivery orders of two delivery destinations (represented by black and gray circles) in one delivery order are swapped with each other.

[0105] The second change pattern is a pattern in which the delivery order from the middle of one of two delivery orders, which return from a base to the base via multiple delivery destinations, to the base is swapped with the delivery order from the middle of the other delivery order to the base. Figure 11B is a diagram for explaining this second change pattern. As shown in Figure 11B, the second change pattern is a pattern in which the delivery orders from the middle of the delivery destination to the end (represented by black and gray circles) of the two delivery orders are swapped with each other.

[0106] The third change pattern is a pattern in which a delivery destination is moved from one of two delivery orders that return to the base via multiple delivery destinations to the base, but the delivery destination is not moved from the other to the one. Figure 11C is a diagram for explaining this third change pattern. As shown in Figure 11C, the third change pattern is a pattern in which a delivery destination (represented by a black circle) in one of two delivery orders is incorporated into the other delivery order, but the other delivery destination is not incorporated into one of the delivery destinations.

[0107] The fourth change pattern is a pattern in which a delivery destination in one of two delivery orders that return to the base via multiple delivery destinations is swapped with a delivery destination in the other order. Figure 11D is a diagram for explaining this fourth change pattern. As shown in Figure 11D, the fourth change pattern is a pattern in which a delivery destination in one of two delivery orders (represented by a black circle) is swapped with a delivery destination in the other order (represented by a gray circle).

[0108] The fifth change pattern is a pattern in which a delivery destination that has a delivery period that starts before the end date of the planning period and ends after the end date of the planning period and is not currently a delivery target is incorporated into the delivery target for that delivery destination. Figure 12A is a diagram for explaining this fifth change pattern. As shown in Figure 12A, the fifth change pattern is a pattern in which a delivery destination (represented by a black circle) that has been postponed in the delivery order for days included in a delivery period that straddles the planning period is incorporated into the delivery target for that delivery destination.

[0109] The sixth change pattern is a pattern in which a delivery destination that is set for a delivery period that starts before the end date of the planning period and ends after the end date of the planning period and is currently a delivery target is excluded from the delivery target. FIG. 12B is a diagram for explaining this sixth change pattern. As shown in FIG. 12B, the sixth change pattern is a pattern in which a delivery destination (represented by a black circle) is excluded from the delivery order for a day included in a delivery period that spans the planning period and postponed to outside the planning period. As described above, by predefining multiple change patterns, the process for searching for a solution can be realized by the simple process of selecting a change pattern. Note that the change patterns are not limited to the six patterns described above, and other change patterns may be used, and the number of change patterns may be greater or less.

[0110] Once a change pattern is selected, the control unit 20 searches for an improved solution (step S210). That is, the control unit 20 changes the current tentative solution according to the change pattern selected in step S205 (or step S250, which will be described later). Various methods can be adopted to change the tentative solution according to the change pattern. For example, the control unit 20 selects one or two delivery orders from the tentative solution according to a predetermined rule to be changed according to the selected change pattern. The control unit 20 also selects one or more delivery destinations to be changed from the delivery destinations included in the selected delivery order according to a predetermined rule. Then, the control unit 20 changes the selected delivery destinations according to the change pattern. As a result, a solution obtained by changing the tentative solution is obtained.

[0111] Next, the control unit 20 determines whether the solution acquired in step S210 is an improved solution (step S215). That is, the control unit 20 calculates a cost using an objective function based on the solution acquired in step S210 and compares it with the cost of the tentative solution. Then, if the cost of the solution acquired in step S210 is smaller than the cost of the tentative solution, the control unit 20 determines that the solution acquired in step S210 is an improved solution.

[0112] If it is determined in step S215 that an improved solution has been acquired, the control unit 20 determines whether the constraints are satisfied (step S220). That is, the control unit 20 determines whether the solution acquired in step S210 satisfies all of the rotation speed constraints, delivery period constraints, delivery volume constraints, and delivery date constraints. If it is determined in step S220 that the constraints are satisfied, the control unit 20 updates the tentative solution with the solution acquired in step S210, i.e., the improved solution. As a result, a solution with a lower cost than before the update has been found. On the other hand, if it is not determined in step S220 that the constraints are satisfied, the control unit 20 skips step S225.

[0113] When step S225 is executed, or when it is not determined in step S220 that the constraint conditions are satisfied, the control unit 20 determines whether or not the termination conditions are satisfied (step S230). In this embodiment, the termination conditions are defined in advance. For example, the termination condition may be that the cost falls below a threshold value. Alternatively, the termination condition may be that a predetermined amount of time has elapsed since the process of solving the delivery plan problem began, or that a predetermined step (e.g., step S210) has been repeated a predetermined number of times since the process of solving the delivery plan problem began. If it is not determined that the termination condition is satisfied, the control unit 20 repeats the processes from step S205 onwards.

[0114] On the other hand, if it is determined in step S215 that the solution is not an improved solution, the control unit 20 determines whether the search using the same change pattern as the change pattern used when searching for an improved solution in step S210 has ended (step S235). That is, the control unit 20 determines whether the search for an improved solution for the same tentative solution has ended for the same change pattern. Specifically, there are multiple possibilities for the delivery order that can be changed using one change pattern. For example, in the case of the first change pattern shown in FIG. 11A, the delivery order to be changed may be changed to a delivery order on a different day or a different rotation. Furthermore, even if the delivery order is the same day or the same rotation, the delivery destination to be changed may be changed. Therefore, if there is room to change the current tentative solution using the same change pattern, the control unit 20 determines that the search using the same change pattern has not ended. If it is determined that the search using the same change pattern has not ended, the control unit 20 repeats the processing from step S210 onwards.

[0115] On the other hand, if it is determined in step S235 that the search for the same change pattern has ended, the control unit 20 determines whether or not the search for all change patterns has ended (step S240). That is, the control unit 20 determines whether or not the solution search for the same tentative solution in step S210 has been performed using all six change patterns.

[0116] If it is determined in step S240 that the search for all change patterns has not been completed, the control unit 20 determines whether or not the termination condition has been satisfied (step S245). The processing in step S245 is the same as the processing in step S230. If it is determined in step S245 that the termination condition has not been satisfied, the control unit 20 selects one of the unsearched change patterns (step S250) and repeats the processing from step S210 onwards. That is, the control unit 20 selects one change pattern that has not been applied to the same tentative solution, and repeats the processing from step S210 onwards.

[0117] On the other hand, if it is determined in step S240 that the search for all change patterns has been completed, it means that no improved solution has been found even if the current tentative solution is changed using the predetermined change pattern. Therefore, the control unit 20 applies a perturbation to the tentative solution (step S255). The perturbation process may be a known process for escaping from a locally optimal solution, and for example, the technology disclosed in "Outline of Meta-Strategies: Shinji Imahori and Mutsunori Yanagiura, Operations Research, 58 (2013) 695-702" can be used.

[0118] Next, the control unit 20 determines whether the termination condition is satisfied (step S260). The process of step S260 is the same as the process of step S230. If it is not determined in step S260 that the termination condition is satisfied, the control unit 20 repeats the processes from step S205 onwards. That is, a new search for an improved solution is performed using the tentative solution updated by applying the perturbation as a target.

[0119] If it is determined in step S230, step S245, or step S260 that the termination condition is satisfied, the control unit 20 identifies the current tentative solution as the solution of the delivery plan (step S270). That is, the control unit 20 considers the tentative solution obtained at the stage when the termination condition is satisfied to be the best solution, and sets the tentative solution as the solution of the delivery plan.

[0120] When a solution for the delivery plan is obtained, the control unit 20 returns to the process shown in FIG. 5 and outputs the delivery plan (step S135). That is, the control unit 20 uses the function of the delivery plan creation unit 20d to send information indicating the delivery plan to the manager terminal 50 via the communication unit 40. The manager terminal 50 displays the information indicating the delivery plan on a display unit (not shown). As a result, the manager can recognize the delivery plan that each worker should implement.

[0121] (3) Other embodiments: The above-described embodiment is merely an example of implementing the present invention, and various other embodiments are possible. For example, the delivery plan creation system may be implemented by multiple devices, or may be configured to be implemented by a cloud server, etc. Furthermore, at least some of the remaining amount prediction unit 20a, delivery period setting unit 20b, plan period setting unit 20c, and delivery plan creation unit 20d may be separated into multiple devices. For example, the remaining amount prediction unit 20a may have a configuration in which the function of collecting consumption and the function of predicting remaining amount are performed by different devices. Of course, some of the components of the above-described embodiment may be omitted, and the order of processing may be changed or omitted. For example, steps S120 and S125 may be performed before step S115. Furthermore, the cost and constraints in the above-described embodiment are merely examples, and items evaluated in terms of cost may be constraints, or items considered as constraints may be evaluated in terms of cost. For example, the second term of the cost (Mt (total overtime hours of workers)) in the above embodiment may be deleted, and a constraint may be introduced to prevent overtime hours from occurring. Also, a constraint may be introduced to prevent work exceeding the allowable overtime hours from occurring.

[0122] The remaining amount prediction unit is only required to be able to predict the change in the remaining amount of gas stored in multiple gas containers installed at each of multiple delivery destinations. In other words, the remaining amount prediction unit is only required to be able to predict the change in the remaining amount of gas for each delivery destination at least within a planned period. The information on which the prediction is based may be, as in the above-described embodiment, statistical values ​​of past changes in the remaining amount of gas in the gas containers, or a history of past remaining amounts in the gas containers. In either case, the remaining amount prediction unit is only required to be able to predict the change in gas within a future planned period based on the past remaining amount of gas. Of course, other than the day of the week, fluctuation factors such as season and temperature may also be taken into account in the statistical values.

[0123] The actual remaining amount of gas in the gas container may be obtained by various methods. For example, the past remaining amount or the current remaining amount of gas in the gas container may be obtained based on the measurement value (meter reading) of a meter connected to the gas container. The method for replacing the gas container at the delivery destination may be a staggered replacement method or a full replacement method in which all of the gas containers are replaced.

[0124] The delivery period setting unit may set a delivery period during which the remaining amount of gas is below a threshold but does not reach zero, based on the transition for each of the multiple delivery destinations. That is, the delivery period setting unit determines that replacement is necessary when the remaining amount of gas is below a predetermined threshold. The threshold may be determined in various ways as long as it is determined in advance so that replacement does not occur excessively and the remaining amount of gas does not reach zero. In the case of an alternating replacement method, the threshold is the remaining amount of gas when one or more predetermined number of gas containers are emptied at the delivery destination. The threshold may be determined for each delivery destination, or may be determined based on the number of installed gas containers, and various ways can be adopted.

[0125] The delivery period may be a period that allows the gas container to be replaced before the remaining gas level reaches 0, and the end date of the delivery period may be a day before the remaining gas level reaches 0. However, because the change in the remaining gas level is a prediction, it is preferable that the end date of the delivery period be a margin of days before the predicted day when the remaining gas level reaches 0. Such an end date may be defined as the day when the remaining gas level falls below a lower limit. Note that in the above-described embodiment, the delivery period and the planned period are determined in days, but the start and end of the period may also be delimited by more precise units, such as hours or minutes.

[0126] The planning period setting unit is only required to be able to set a planning period, which is the period for which the delivery plan is to be created. In other words, the planning period setting unit sets a period that includes at least the delivery period of the delivery destination of the delivery target as the planning period. The planning period may be a fixed period determined in advance, may vary dynamically, or may be determined by input by the user, etc.

[0127] The delivery plan creation unit is only required to create a delivery plan based on the cost evaluated by the objective function. Since gas container replacement is performed once per delivery period, for delivery destinations that have multiple delivery periods within the planning period, a delivery plan is created so that one delivery is made within each delivery period.

[0128] The cost is the cost of traveling from a base to multiple delivery destinations and returning to the base, and various costs may be adopted in addition to the cost determined based on the travel time between locations as in the above-described embodiment. Furthermore, various costs may be incorporated into the objective function. In the above-described embodiment, the base as the departure point and the base as the final arrival point are the same location, but the two may be different locations.

[0129] Various algorithms for solving delivery planning problems can be employed to create delivery plans. In other words, various aspects can be used for the constraints, the objective function for evaluating a better delivery plan, the cost that is a factor in evaluating the quality of a delivery plan, the variables that vary when the objective function is changed, and the fixed constants. Furthermore, the type of algorithm is not limited to the iterative local search method used in the above-described embodiment. Various algorithms can be employed, including constructive solution methods such as the greedy algorithm, annealing algorithm, genetic algorithm, tabu search, set partitioning, insertion algorithm, Clarke-Wright saving algorithm, and Fisher-Jaikumar generalized allocation algorithm.

[0130] The cost when postponement is performed may be determined based on various indicators. In other words, the cost may be defined so that a situation in which the cost is reduced simply by postponement is prevented and both postponement and non-postponement cases can be included in the delivery plan. Therefore, various definitions other than the above-mentioned definition may be adopted for the cost when postponement is performed. For example, a cost may be introduced that is larger when the number of postponed delivery destinations is large than when the number is small.

[0131] The cost when postponement is performed may be determined by at least one of the following: a cost value that is lower than when delivery is not made within the planned period to a delivery destination that should be delivered within the delivery period, and a value that is lower when the number of days in the delivery plan included in the period before the end date of the planned period is large compared to when it is small. The characteristics of these values ​​do not need to be the same as those in the above-described embodiment. For example, the first component of the postponement cost may be a predetermined value, or the second component may vary in dependence on the number of days in the delivery plan in a curved manner.

[0132] Furthermore, embodiments of the invention may be programs or methods. The above-described systems, programs, and methods may be realized as a single device or multiple devices, and include various aspects. They may also be modified as appropriate, such as being partly software and partly hardware. Furthermore, the invention may also be realized as a recording medium for a program that controls the system. Of course, the recording medium for the software may be a magnetic recording medium or a semiconductor memory, and any recording medium developed in the future may be considered in the same way. [Explanation of symbols]

[0133] 10...Delivery plan creation system, 20...Control unit, 20a...Remaining amount prediction unit, 20b...Delivery period setting unit, 20c...Plan period setting unit, 20d...Delivery plan creation unit, 30...Storage medium, 30a...Consumption amount data, 30b...Remaining amount data, 30c...Delivery destination data, 30d...Worker data, 30e...Cost data, 40...Communication unit, 50...Administrator terminal, 61...Gas container, 62...Measuring instrument, 63...Gas burner

Claims

1. a remaining amount prediction unit that predicts a transition of the remaining amount of gas stored in a plurality of gas containers installed at each of a plurality of delivery destinations; a delivery period setting unit that sets a delivery period during which the remaining amount of gas is equal to or less than a threshold value and does not become zero, based on the transition for each of the plurality of delivery destinations; a planning period setting unit that sets a planning period, which is a target period for creating a delivery plan that indicates the destinations and the order in which deliveries should be made from departure from a base to return to the base on each day within the planning period; a delivery plan creation unit that creates the delivery plan for delivering the gas containers to each of the plurality of delivery destinations within the delivery period for each of the plurality of delivery destinations by solving a delivery plan problem that searches for a solution that minimizes an objective function that includes, as costs during the planning period, at least a cost associated with traveling from the base to the base via the plurality of delivery destinations and a cost that occurs when delivery is not made to the delivery destinations that should be delivered within the delivery period that starts before the end date of the planning period and ends after the end date of the planning period; A delivery plan creation system comprising:

2. The delivery plan creation unit After the created delivery plan is put into operation, the delivery plan is updated based on the delivery period set based on the latest prediction of the remaining amount of gas before the planned period has elapsed. The delivery plan creation system according to claim 1 .

3. The cost when delivery to the delivery destination that should be delivered within the delivery period is not made within the planned period is: A value of the cost that is reduced when delivery to the delivery destination that should be delivered within the delivery period is not made within the planned period compared to when delivery is made; a value that is smaller when the number of days in the delivery plan included in the period before the end date of the planning period is large than when the number of days is small; Identified based on at least one of 3. A delivery plan creation system according to claim 1 or 2.

4. The delivery plan creation unit The delivery plan is created with a constraint that delivery is not performed on holidays; The delivery period setting unit If the delivery period includes a holiday, move the start date of the delivery period forward to an earlier date.

3. A delivery plan creation system according to claim 1 or 2.

5. The delivery plan creation unit Setting the cost of travel outside a predetermined working time range to be greater than the cost of travel within the working time range; 3. A delivery plan creation system according to claim 1 or 2.

6. The delivery plan creation unit Obtaining a cost indicated by the objective function in a tentative solution of the delivery plan that satisfies predetermined constraints; repeating a process of determining, as the latest provisional solution, a solution that reduces the cost indicated by the objective function when the provisional solution is modified using a pattern selected from a plurality of predetermined change patterns; acquiring the tentative solution when a predetermined termination condition is satisfied as a solution of the delivery plan; 3. A delivery plan creation system according to claim 1 or 2.

7. The delivery plan creation unit generating an initial solution of the tentative solution by adding each of the delivery destinations in a predetermined order so as to satisfy the constraints on the order of deliveries departing from and returning to the base on each day within the delivery period; The delivery plan creation system according to claim 6.

8. The change pattern is A pattern in which the delivery order is changed within the delivery order from the base via the plurality of delivery destinations and back to the base; a pattern in which the delivery order from one of two delivery orders returning from the base via the plurality of delivery destinations to the base is swapped with the delivery order from the other of two delivery orders returning from the other to the base; a pattern in which the delivery destination is moved from one of two delivery orders returning to the base via the plurality of delivery destinations to the base, and the delivery destination is not moved from the other to the one; a pattern in which one of two delivery sequences returning from the base via the plurality of delivery destinations is swapped with the other delivery destination; a pattern in which the delivery destinations for which the delivery period is set to start before the end date of the planning period and end after the end date of the planning period, and which are not currently subject to delivery, are included in the delivery target; a pattern of excluding from delivery targets the delivery destinations that are set for the delivery period that starts before the end date of the planning period and ends after the end date of the planning period and that are currently subject to delivery; Contains at least one pattern of The delivery plan creation system according to claim 6.

9. By computer, a remaining amount prediction step of predicting a transition of the remaining amount of gas stored in a plurality of gas containers installed at each of a plurality of delivery destinations; a delivery period setting step of setting a delivery period during which the remaining amount of gas is equal to or less than a threshold value but does not become zero, based on the transition for each of the plurality of delivery destinations; a planning period setting step of setting a planning period, which is a target period for creating a delivery plan indicating the delivery destinations and the order in which deliveries should be made from departure from a base to return to the base on each day within the planning period; a delivery plan creation process for creating the delivery plan for delivering the gas containers to each of the plurality of delivery destinations within the delivery period for each of the plurality of delivery destinations by solving a delivery plan problem that searches for a solution that minimizes an objective function that includes, as costs during the planning period, at least a cost associated with traveling from the base to the base via the plurality of delivery destinations and a cost that occurs when delivery is not made to the delivery destinations that should be delivered within the delivery period that starts before the end date of the planning period and ends after the end date of the planning period; How to create a delivery plan.

10. Computer, a remaining amount prediction unit that predicts a transition of the remaining amount of gas stored in a plurality of gas containers installed at each of a plurality of delivery destinations; a delivery period setting unit that sets a period during which the remaining amount of gas is equal to or less than a threshold value but does not become zero as a delivery period based on the transition for each of the plurality of delivery destinations; a planning period setting unit that sets a planning period, which is a target period for creating a delivery plan that indicates the destinations and the order in which deliveries should be made from departure from a base to return to the base on each day within the planning period; a delivery plan creation unit that creates the delivery plan for delivering the gas containers to each of the plurality of delivery destinations within the delivery period for each of the plurality of delivery destinations by solving a delivery plan problem that searches for a solution that minimizes an objective function that includes, as costs during the planning period, at least a cost associated with traveling from the base to the base via the plurality of delivery destinations and a cost that occurs when delivery is not made to the delivery destinations that should be delivered within the delivery period that starts before the end date of the planning period and ends after the end date of the planning period; A delivery planning program that functions as a

Images