JP2010150020A - Vehicle operation plan creation method and apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a vehicle operation plan creation method and apparatus that draft a vehicle allocation plan that increases a turnover rate of vehicles for picking up and delivering loads at a plurality of scattered delivery bases and to destinations in short distance conveyance and reduces delivery costs. <P>SOLUTION: The method includes a planning data selection processing step of asking input and selection of delivery order data, a data reading processing step of reading the delivery order data, master data necessary for vehicle operation plan creation, and a search scenario indicating a search order and a search method, and an optimum operation plan creation processing step of creating a vehicle operation plan from the delivery order data and the master data according to the search scenario through search processing of maximizing an objective function that sums differences each between vehicle receipts and vehicle operation costs. In the planning data selection processing step and the data reading processing step, the one to which at least a final operation cycle in the previous planning is added is added as a target. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention efficiently transports goods such as products stored in distribution bases (warehouses, factories, distribution centers) in manufacturing / distribution industries to destinations by means of transportation such as trucks and trailers held at the distribution bases. The present invention relates to a vehicle operation planning technology for delivery.

  In recent years, there has been an increasing demand for more efficient logistics and cost reduction. In response to such an increase in momentum, various techniques and devices for automatically creating and assisting in creating a dispatch plan have come to be seen. The transportation schedule creation technology in the logistics system corresponds to the so-called traveling salesman problem and vehicle routing problem, and a technology for creating an efficient traveling plan for customers existing in a certain region has already been developed.

  For example, in the technique disclosed in Patent Document 1, a solution is obtained in three stages, and master management means for managing at least item data, distribution base data, vehicle and garage data, and the master management means An operation route setting means for setting an operation route according to a transportation request based on the data in the item, and the means can be loaded on one vehicle according to the transportation request based on the data A set creation unit that creates a group as one set of operation data, and sets a plurality of sets of operation route candidates that can be connected in series between adjacent sets among a plurality of sets, A route selection unit that selects an operation route that satisfies a transportation request and that minimizes one predetermined evaluation index among a plurality of evaluation indexes among candidate operation routes. It is configured Te, while satisfying various constraints, and provides a predetermined target, for example, dispatch unit capable of quickly formulating minimization of gross vehicle number.

  In addition, the technology disclosed in Patent Document 2 is efficient in keeping with time constraints at the collection and delivery destinations of vehicles in which the cargo transportation from the distribution center to the delivery destination and the cargo transportation from the collection destination to the distribution center are mixed. The purpose is to provide a transportation plan creation method and system for creating a simple transportation schedule. When creating a transportation schedule, the location information of the distribution center and the collection and delivery destination base is input, the amount of luggage to be transported, The package information including the time restriction of the collection and delivery destination and the arrival time is input, and the transport vehicle information including the load capacity limit of the transport vehicle is input. In addition, when creating a transportation schedule, a table of the maximum time for arrival delays at each traveling site of each vehicle and the maximum amount of collectable and delivered luggage that can be loaded is created, and the transportation schedule consists of traveling sites for multiple vehicles. Whether or not a baggage collection / delivery destination base can be added is determined based on the table, and a schedule for a transportation vehicle is created using the respective information under the condition that the time constraint is observed.

  Furthermore, in the technique disclosed in Patent Document 3, vehicle allocation and stacking according to the purpose to be prioritized is performed, and the objective function is set to the maximum using the difference between the income obtained by the vehicle and the cost of the vehicle as an objective function. It tries to become. For this reason, it is possible to reduce the empty vehicle travel and improve the vehicle rotation rate corresponding to a plurality of delivery bases, and it is possible to create a vehicle operation plan that can be delivered efficiently and with minimal cost. In addition to proposing multiple objective function optimization processing methods, and introducing a search scenario that specifies the selection of optimization processing methods and the search order of the selected optimization processing methods, it is possible to flexibly adapt to changes in the target of vehicle operation plans. In addition, there is an effect of shortening the plan creation time with respect to an increase in search space caused by dealing with a plurality of delivery bases.

Reference documents including the non-patent documents referred to in the above-mentioned patent documents and “the best mode for carrying out the invention” are described below.
JP 2000-36093 A JP 2003-285930 A JP 2005-122679 A “Combinatorial optimization: focusing on meta-strategy”, Yasunori Yanagiura, Toshihide Ibaraki, Asakura Shoten, 2001

  Delivery of heavy goods such as steel products and heavy machinery, especially delivery bases, warehouses, etc. for short-distance delivery where there are multiple destinations, factories, distribution centers, and consigned warehouses within 200-300 km It is important to reduce costs to efficiently carry out the operation cycle, which is the collection and delivery to the customers, to reduce the number of vehicles and the number of empty vehicles. A simple example of the above two vehicles is delivery as shown in FIG. In FIG. 32, both the vehicle 1 and the vehicle 2 perform the collection work at the delivery base, the vehicle 1 delivers to the delivery destinations A and B, then the collection work at the warehouse, the delivery to the delivery destination C, and returns to the delivery base. . The vehicle 2 performs delivery to delivery destinations D and E, then returns to the delivery base, collects again, delivers to delivery destination F, and returns to the delivery base. In the figure, solid line arrows indicate delivery with loads, and broken line arrows indicate no load, that is, travel in an empty state after unloading.

  However, since the techniques disclosed in Patent Document 1 and Patent Document 2 are intended to create optimal delivery routes for a plurality of delivery destinations, a plurality of delivery bases including the warehouse as described above are used. There are no considerations.

  Further, in the technique disclosed in Patent Document 3, the above problem has been solved, but when a new vehicle operation plan is created, a search process is performed only on the delivery order data for the new plan and only the vehicle. In order to optimize, if it pays attention to a specific vehicle, it may be a useless run. FIG. 33 is a diagram illustrating an example of an operation plan for one vehicle.

  This example shows the movement of one vehicle. The series of movements developed in the previous plan (from delivery base A (vehicle base) to delivery destination A (1), from delivery destination A to delivery base B ( 2) A series of movements (distribution base A (vehicle base) from delivery base B to delivery base B (3) and return from delivery base B to delivery base A (vehicle base) (4)) ) To delivery base B (5), delivery base B to delivery destination C (6), and return from delivery destination C to delivery base A (vehicle base) (7)).

  Similarly to FIG. 32, the solid line arrows indicate delivery with loads, and the broken line arrows indicate no load, that is, unloading and traveling in an empty state. (For example, the distance from the delivery base B to the delivery base A (vehicle base) (4) is about three times longer than the distance from the delivery base B to the delivery base B (3)). Return travel (4) from delivery destination B to delivery base A (vehicle base) in the final operation cycle ((3) to (4)) at the time of the previous plan, and the initial operation cycle ((5) to ( It can be seen that the travel (5) from the delivery base A (vehicle base) to the delivery base B in 7)) is a wasteful travel because both are empty.

  As described above, in the case where both of the adjacent routes in the opposite direction become empty, the travel plan is useless. Therefore, for example, in the example shown in FIG. 33, if the travel route shown in FIG.

  That is, instead of the route from the delivery destination B shown in FIG. 33 to the delivery location A (vehicle location) (4), the delivery location B approaches the delivery location B as shown in FIG. 34 (4). If you take the route of loading, delivery base A (vehicle base) (5), delivery to delivery destination C (6), and return to delivery base A (vehicle base) (7), you will run wastefully in an empty vehicle. And can be delivered at the shortest distance.

  The present invention solves the above-mentioned problems, collects loads from a plurality of scattered delivery bases, and develops a vehicle allocation plan that improves the vehicle turnover rate and reduces the delivery cost in short-distance delivery delivered to a delivery destination. An object is to provide a vehicle operation plan creation method and apparatus.

  The invention according to claim 1 prompts the input and selection of data relating to a package to be carried, that is, delivery order data, in a method of preparing an operation plan for a vehicle that performs collection work at a plurality of delivery bases and delivers the vehicle to a plurality of delivery destinations. Plan creation data selection processing step, data delivery processing step for reading the delivery order data, master data necessary for vehicle operation plan creation and search scenario indicating search order and search method, vehicle revenue obtained by each vehicle, Based on the search scenario from the delivery order data and the master data, using a search process that maximizes the objective function, the sum of the difference between the vehicle operating costs required when operating each vehicle Optimal operation plan creation process to create a vehicle operation plan, and vehicle operation plan output process to output the created vehicle operation plan The a, in the planning data for the selected process and the data read processing step, a fleet planning method characterized by adding as a target plus at least the last operation cycle of the previous plan.

  Further, the invention according to claim 2 is an apparatus for collecting goods at a plurality of delivery bases and creating an operation plan for a vehicle to be delivered to a plurality of delivery destinations. Prompt plan creation data selection processing means, data delivery processing means for reading the delivery order data, master data necessary for creating the vehicle operation plan and search scenario indicating the search order and search method, vehicle revenue obtained by each vehicle, Based on the search scenario based on the search scenario from the delivery order data and the master data, using a search process that maximizes the objective function, which is the sum of the differences from the vehicle operation costs when operating each vehicle , Optimal operation plan creation processing means for creating a vehicle operation plan, and vehicle operation plan output for outputting the created vehicle operation plan And a stage, wherein in the planning data selection processing means and said data reading processing unit, a vehicle operation planning apparatus characterized by adding as a target plus at least the last operation cycle of the previous plan.

  According to the present invention, vehicle allocation and stacking according to the purpose to be prioritized is performed, and the objective function is maximized with the difference between the revenue obtained by the vehicle and the cost of the vehicle as the objective function. Corresponding to a plurality of delivery bases, it is possible to reduce empty vehicle travel and improve vehicle rotation rate, and to create a vehicle operation plan that can be delivered efficiently, with minimal cost and in reality. In addition to proposing multiple objective function optimization processing methods, and introducing a search scenario that specifies the selection of optimization processing methods and the search order of the selected optimization processing methods, it is possible to flexibly adapt to changes in the target of vehicle operation plans. In addition, there is an effect of shortening the plan creation time with respect to an increase in search space caused by dealing with a plurality of delivery bases.

  In addition, when creating a new vehicle operation plan, search processing is performed by adding at least the last operation cycle at the time of the previous plan to the data such as delivery order data and vehicles for the new plan. As a result, further optimization can be realized.

  The best mode for carrying out the present invention will be described below with reference to the drawings. FIG. 2 is a diagram showing an example of the apparatus configuration of the present invention. In the figure, 1 is a vehicle operation plan creation computer, 2 is a master management server, 3 is an ordering system, 4 is a printer, and 5 is a computer network. The apparatuses 1 and 4 are connected to the computer network 5. The vehicle operation plan creation computer 1 is a personal computer or workstation provided with data reading means, optimum vehicle operation plan creation means, vehicle operation plan output means, a keyboard, a mouse, and a display. In this configuration example, only one unit is illustrated, but a plurality of units may be configured corresponding to each type of product to be delivered, branch offices, delivery bases, and the like.

  The master management server 2 is a computer for storing and managing masters necessary for creating a vehicle operation plan. Necessary masters are transmitted to the vehicle operation plan creation computer 1 in response to a request from the data reading means in the vehicle operation plan creation computer 1. Also, change or modify masters.

  The ordering / order receiving system 3 transmits order data in response to a request from the data reading means in the vehicle operation plan creation computer 1. Also, the vehicle operation plan transmitted from the vehicle operation plan output means in the vehicle operation plan creation computer 1 is received, and processing such as billing processing, securing orders to be shipped, shipping instructions to warehouses and instructions to shipping companies, etc. Do. The ordering / ordering system 3 is an independent external system in the present apparatus example, but may be an internal system such as the vehicle operation plan creation computer 1 and the master management server 2. The same applies to the master management server 2, the following printer 4, and the like.

  The printer 4 prints the vehicle operation plan created by the vehicle operation plan creation computer 1. It is also used for the output of the ordering / ordering system 3 and the master management server 2.

  The processing / procedure performed inside the vehicle operation plan creation computer 1 is shown in FIG. In the plan creation data selection process 10, data relating to a package to be carried (hereinafter referred to as delivery order data) is selected by inputting a scheduled delivery date of the package to be carried.

  In the data reading process 11, the delivery order data selected by the plan creation data selection process 10, the master data necessary for creating the vehicle operation plan, and the search scenario indicating the search order and the search method are read, and the next step optimum operation plan is read. Data to be input to the creation process 12 is created.

  In the optimum operation plan creation process 12, an optimum vehicle operation plan is created from the input data created in the data reading process 11. Details of the processing here will be described later.

  In the vehicle operation plan display process 13, the operation plan created in the optimum operation plan creation process 12 is displayed on a display or the like. Further, if necessary, the printer 4 shown in FIG.

  In the vehicle operation plan transmission process 14, a process for transmitting the created vehicle operation plan to the ordering / order receiving system 3 in FIG. 2 is performed. The above is the outline of the processing in the vehicle operation plan creation computer 1.

  Hereinafter, the delivery order data and the master data necessary for the vehicle operation plan will be described in detail. First, the delivery order data is specifically requested and received from the ordering system 3 in FIG. 2 as the data selected by the plan creation data selection processing 10. The delivery order data is data relating to the package to be delivered, and an example of items of the delivery order data is shown in FIG. The items of the delivery order data include the order ID number of the package to be delivered, the delivery destination code for determining the delivery destination, the product code for determining the type of package to be delivered, and the product type for determining the product type. Information such as a code, a base code indicating a delivery base, a time designation indicating a delivery time, a product weight, and a size are included.

  Next, the master data necessary for the vehicle operation plan is received by requesting the master data necessary for creating the vehicle operation plan from the master management server 2 in FIG. The master data includes, for example, the following various types.

  FIG. 4 shows an example of data items called vehicle masters. This vehicle master contains information on vehicle specifications that can be used, such as the vehicle number that distinguishes each vehicle, the shipping company classification to which it belongs, the base code indicating the delivery base to which it belongs, the maximum load weight, the cargo bed length, the cargo bed width, etc. It contains information about individual vehicles, such as when vehicles are available and costs (fixed costs, fuel costs) required to use them. This information is used when calculating the cost when creating the delivery plan.

  FIG. 5 shows an example of an item of data called a distance master. This distance master includes data on the distance from the delivery base to the delivery destination or the distance between the delivery destinations and the moving speed. These data are used for calculating the delivery time when planning the delivery.

  FIG. 6 shows an example of data items called a delivery destination master. This delivery destination master contains data related to the delivery destination. Items included in the delivery destination master include a time for unloading the package at the delivery site, a time zone during which the package can be unloaded at the delivery site, and the like. These are also used for calculating the delivery time at the time of delivery planning as in the case of the distance master.

  FIG. 7 shows an example of data items called a fare master. This fare master includes information related to delivery fee data for each product type according to the delivery distance. At the time of creating a delivery plan, these pieces of information are used when calculating the income earned by each vehicle.

  FIG. 8 shows an example of an item of data called a delivery base master. The delivery base master includes data relating to the loading time at each delivery base. These are used to calculate the delivery time when planning the delivery.

  FIG. 9 shows an example of data items called virtual cost masters. The virtual cost master includes data related to a virtual cost, which will be described later. In this example, the item name is associated with the cost value. These data are used to calculate the virtual cost of the objective function at the time of delivery planning. As mentioned above, although an example of the delivery order data and the master data necessary for the vehicle operation plan has been described, it is not limited to this, and should be changed or added as appropriate according to the delivery form.

  As shown in FIG. 11, the vehicle operation cycle (hereinafter also referred to as a trip) is a collection work at a delivery base (a warehouse, etc.), a delivery work to a delivery destination, and a forwarding to a delivery base after the delivery work is completed. A series of operations up to the operation is defined as one operation cycle (one trip), and one or a plurality of operation cycles are planned for each vehicle. FIG. 12 shows a case where a plurality of operation cycles fall within the predetermined working hours, and FIG. 13 shows a case where a plurality of operation cycles exceed the predetermined working hours.

  In the present invention, when a new vehicle operation plan is created, a new plan target delivery order data, data such as vehicles, and at least the final operation cycle at the time of the previous plan are added as targets, which will be described later. Search process to optimize. If the timing of the new plan is late (close to the end time of the previous plan), only the return travel of the last operation cycle can be changed, but if the implementation timing is early, it can be changed to the end of the final operation cycle And Further, if the implementation timing is earlier, the transport route (including the combination of the vehicle and the baggage) of the previous operation cycle can be reviewed to optimize the operation cycle.

  When specifying the review range, the earliest time to be reviewed is specified, the operation cycle and return are specified. In a specific example, the “base code indicating the delivery base to which the vehicle belongs” in the vehicle master is set as the base code of the starting base of the operation cycle of the previous plan of the review range, or in the “time zone in which the vehicle can be used” The start time (earliest time) is replaced with the corresponding time in the operation cycle of the previous plan.

  The above-described optimum operation plan creation process 12 shown in FIG. 1 is based on the formulation of a combination optimal problem that satisfies a predetermined constraint condition and maximizes (or minimizes) a given objective function. . Such combinatorial optimization problems are often solved using a technique called approximate solution. Various methods have been proposed for the approximate solution of the combinatorial optimization problem, and details thereof are described in Non-Patent Document 1, for example.

  In the present invention, the objective function is as shown in equation (1).

各車両毎に車両収入と車両運行コストを求め、全車両合計した金額を目的関数にし、これを最大にしようとするものである。そして、（１）式における車両収入は、（２）式に示すものとなる。

The vehicle revenue and the vehicle operation cost are obtained for each vehicle, and the total amount of all the vehicles is used as an objective function to maximize this. And the vehicle revenue in Formula (1) becomes what is shown in Formula (2).

（２）式の内、配送料金（タリフ）は、いわゆる配送運賃収入であり、一般には積荷の種類と配送拠点からの距離に応じて決まっている。鉄鋼製品であるコイルの場合では、配送先の基準距離比例定額テーブルに従った、コイル重量比例で料金が決まっている（図７参照）。

In the equation (2), the delivery fee (tariff) is so-called delivery fare revenue, and is generally determined according to the type of cargo and the distance from the delivery base. In the case of a coil that is a steel product, the charge is determined in proportion to the coil weight according to the reference distance proportional fixed amount table of the delivery destination (see FIG. 7).

The next virtual income is virtually incorporated into the income as a bonus. There are the following bonuses (see FIG. 9 (1)).
(1) Fuel cost correction bonus: For example, a fixed value for each delivery distance for each used vehicle store is determined and added.
(2) Vehicle fixed cost correction bonus: For example, a fixed amount for each store company of the used vehicle is determined and added.
(3) Delivery charge correction bonus: For example, a correction coefficient for each store company of the used vehicle is multiplied by the delivery charge of the corresponding vehicle.
(4) Delivery destination consolidation bonus: For example, in the order assigned to one trip of one vehicle, the order number -1 of the same delivery destination is multiplied by a fixed amount.
(5) Store company profit bonus: For example, a store company profit weight coefficient for each store company of the used vehicle is multiplied by the profit of the corresponding vehicle, and a store company profit bias is added.
(6) Different distance stacking bonus (negative): For example, the travel time between the nearest delivery destination and the farthest delivery destination among the delivery destinations dispatched to the same vehicle is multiplied by a proportional value (minus value).
(7) Negative bonus for vehicle use: For example, a negative bonus is added depending on the number of vehicles used.
By introducing the virtual income as described above, the vehicle having a large virtual income has a large objective function, and therefore is preferentially selected.

  Next, the vehicle operation cost in the equation (1) is as shown in the equation (3).

（３）式の内、輸送コストは、輸送に直接かかるコストとして計上されるものであり、次のようなものが上げられる（図３〜図９参照）。
燃料費コスト： 車種クラス毎に走行に要する燃料費である。
走行時間コスト： 配送にかかる時間に関するコストである。
固定費コスト： 車両を利用する上で必要となる費用である。

Of the formula (3), the transportation cost is counted as a cost directly related to transportation, and the following can be raised (see FIGS. 3 to 9).
Fuel cost cost: Fuel cost required for driving for each vehicle class.
Travel time cost: Cost related to delivery time.
Fixed cost: This is the cost required to use the vehicle.

The next virtual cost is virtually incorporated into the cost as a penalty when the constraint is exceeded. Examples of penalties include the following (see FIG. 9 (2)).
Overload penalty: Proportional to the weight exceeding the maximum load for each vehicle class per trip.
Cargo size excess penalty: Each trip is proportional to the cargo length exceeding the cargo bed length for each vehicle class (minimum length for all possible loading methods for that vehicle class).
Time Designated Delay Penalty: The delivery time is proportional to the time delayed from the time before the delivery allowance time at the delivery designated end time.
Vehicle available time excess penalty: The return time at the end of every trip is proportional to the time beyond the available end time for each vehicle.
Maximum number of destinations penalties: If the maximum number of destinations is exceeded for each trip, it is proportional to the total number of destinations.
By introducing the virtual cost (penalty) as described above, it is possible to search even in an infeasible region, and more efficient optimization calculation can be performed. However, a solution with a penalty value of 0 is adopted as the solution.

FIG. 10 is a diagram showing a process flow of the optimum operation plan creation process 12 of FIG. In the data reading at step 20, necessary data is read by the optimum operation plan creation means, and the optimum operation plan creation process is initialized. The data to be read includes the delivery order data, vehicle master data, delivery destination master data, distance master data, fare master data, base master data, virtual cost master data, and a search scenario described later.
Next, in the search scenario interpretation at step 21, the search scenario is interpreted from the read data. A search scenario is a description of initial solution creation, optimization processing by various search units, number of iterations, conditional branching, end conditions, etc., and input and storage strategically according to the target to be optimized in advance by a person Let me.

  The optimization process by various search units includes 1 order (1 load) and 1 operation cycle in 1 vehicle, 1 order in the same delivery destination in 1 operation cycle in 1 vehicle and 1 in 1 vehicle. There are optimization processes based on search units such as operation cycle, one vehicle and one operation cycle within one vehicle, one vehicle that is not identical to one vehicle, and two vehicles that are not identical to one vehicle.

  FIG. 14 shows a program describing an example of a search scenario. The search scenario interpretation 21 interprets this search scenario as a processing flow as shown in FIG. In this example of the search scenario, M initial solutions are created, and the created initial solution is optimized N times while changing the search unit. CARGO in the LOOP in FIG. 14 corresponds to step 31 in FIG. 15 (optimization processing 31 using the load as a search unit), and DESTIN is similarly used in step 32 (optimization processing 32 using the delivery destination as a search unit). TRUCK is step 33 (optimization process 33 with one vehicle and one vehicle as a search unit), TRUCK1by2 is step 34 (optimization process 34 with two vehicles and one vehicle as a search unit), and TRIP is step 35 (operation cycle) The optimization process 35) and TRIP_DIV using the search unit correspond to step 36 (leveling operation cycle 36), respectively.

  In the optimization process 31 using the load as a search unit and the optimization process 32 using the delivery destination as a search unit, the delivery lots are aggregated and adjusted. In the optimization process 33 using 1 vehicle and 1 vehicle as a search unit and the optimization process 34 using 1 vehicle and 2 vehicles as a search unit in FIG. 15, the type of vehicle to be used is adjusted. In the optimization process 35 and the operation cycle leveling 36 using the operation cycle of FIG. 15 as a search unit, the operation cycle is adjusted.

In step 22, the optimization method interpreted in step 21 is executed.
In step 23, the one having the best evaluation function value is selected from those searched by the search scenario, and the optimum operation plan creation process is terminated.
The above is the outline process flow of the optimum operation plan creation process which is the core of the present invention. Now, each search method (steps 31 to 36) will be described in detail using the processing flow of the search scenario in FIG.

Step 30 first creates an initial solution on which the search is based. In the case of the present invention, the delivery order data is loaded on the vehicle given by the vehicle master. The initial solution is created through the following four steps, for example.
(1) The delivery order data is grouped into delivery destination units.
(2) Sort the data grouped in units of delivery destinations in order from the base.
(3) Randomly arrange the vehicles given by the vehicle master in the order sorted in (2), and allocate them until the loaded weight is exceeded.
(4) Repeat (3) until the baggage runs out. When there are not enough vehicles, new vehicle data for the vehicle defined in the vehicle master is generated and the packages are loaded.

In the process of creating the initial solution, the process proceeds so as to satisfy the following constraint conditions in the delivery order data and the master data.
Loadability by designation: For example, what is called store designation (see FIG. 3) prohibits loading on vehicles other than the designated store (transporter). In addition, there are vehicle type class designation, packing style (appearance and shape of packed items) designation, and the like.

Unloading / loading restrictions: Depending on the location where the delivered package is unloaded / loaded, there are restrictions such as the time required for unloading / loading or the non-working time zone.
Vehicle allocation priority constraints: There are various priority constraints in the allocation of vehicles among the available vehicle allocations such as the fixed vehicle, the temporary vehicle, and the route dealer described above.
Prohibition of stacking in out-of-service areas: Delivery destinations that cannot be routed between delivery destinations in a combination of delivery destinations that are stacked within one trip (one cycle of transportation) are not allowed. This includes, for example, road conditions, non-communication areas due to geographical restrictions, restrictions due to sales areas of store companies (transporters), and the like.
The farthest delivery distance limit for each trip: A maximum limit is set for the distance that can be delivered for each trip, and stacking exceeding the limit is not allowed.
The above are the main constraint conditions, but there are various other restrictions depending on delivery conditions such as cargo length restrictions.

  In step 31 of FIG. 15 (optimization processing 31 using the load as a search unit), assuming that the load is Job and the operation cycle in the vehicle is Agent, the processing flow shown in FIG. 16 is obtained. First, in step 200, an initial solution corresponding to the case is created.

  In the next step 201, a combination of all jobs and all agents is created, and the combination order is randomly created. This is schematically shown in FIG. If there are M jobs J1 to JM and N agents A1 to AN, M × N job / agent combinations are created and the combinations are arranged randomly.

  Next, a combination of “some job” and “some agent” is sequentially extracted from the top of the random arrangement (step 202), and the replacement process 1 (step 203) is performed. This replacement process is schematically shown in FIG. Move Agent: A1 Job: J1 to Agent: A2. Then, the evaluation function is calculated (step 204), and when the evaluation value is improved (step 205), the following step 206 is not performed and the order is increased by one (step 214), and the random order number is exceeded (step 215). Repeat until.

  If the evaluation value is not improved in step 205, a new replacement process 2 is performed in step 206. In the process shown in FIG. 19, one Job in Agent: A2 is selected at random (here, Job: J2), and moved to Agent: A1 from which Job: J1 originated. Then, the evaluation function is calculated (step 207), and when the evaluation value is improved (step 208), the following step 209 is not performed and the order is increased by one (step 214), and the random order number is exceeded (step 215). Repeat until.

  If the evaluation value is not improved in step 208, further replacement processing 3 is performed in step 209. In the processing shown in FIG. 20, one Job in Agent: A2 is selected at random (here, Job: J3) and moved to Agent: A1 from which Job: J1 originated. Then, the evaluation function is calculated (step 210), and when the evaluation value is improved (step 211), the following step 212 is not performed and the order is increased by one (step 214), and the random order number is exceeded (step 215). Repeat until.

  If the evaluation value is not improved in step 211, it is determined in step 212 whether or not Job remains in Agent: A2, and if it remains, the process returns to step 206. If it does not remain, the evaluation value is good in this search. The job is returned to the Agent before movement (step 211), and the search returns to another combination.

  In step 215, it is determined whether or not a search has been performed for all combinations in the random order number. If there is an improvement in the evaluation value in all cases, further improvement is sought and the process returns to step 203. If all the evaluations have been made but there is no improvement in the evaluation value (step 216), the search ends.

  In step 32 of FIG. 15 (optimization processing 32 with the delivery destination as a search unit), the package collected by the delivery destination is regarded as Job, and the cycle in one vehicle is regarded as Agent, so that the processing flow of FIG. 16 is the same. become.

  The flow of step 33 (optimization processing 33 using one vehicle and one vehicle as a search unit) in FIG. 15 is as shown in FIG. As shown in FIG. 23, the search is between vehicles. In the initial solution reading 300, the solution optimized in the previous optimization process or the solution created in the initial solution creation process is read. In step 301, the evaluation function of the read initial solution is calculated based on the equation (1).

  In the next step 303, combinations of all the vehicles are created, and the combination order is randomly arranged. FIG. 22 schematically shows the above. In the combination of vehicles, since the combination with itself is not created, a random arrangement order of N × N−1 is created.

  Next, a process of exchanging the combined vehicles with each other sequentially from the top of the random arrangement (step 304) is performed (step 305).

  Then, when the solution is improved in the calculation of the evaluation function (step 306) (step 307), the process is advanced by one without performing the process of step 308 (step 309), and the process is repeated until the random order number is exceeded.

  If the evaluation value is not improved, the replaced vehicle is returned to the original state (step 308), the order is advanced by one (step 309), and the process is repeated until the random order number is exceeded.

  In step 311, after determining whether or not a search has been performed for all combinations in the random order number (step 310), if there is an improvement in the evaluation value in all cases, further improvement is obtained and the process returns to step 305. If all the evaluation values have not been improved, the final improved solution is output (step 312) and the search is terminated.

  The flow of step 34 in FIG. 15 (optimization process 34 using two vehicles and one vehicle as a search unit) is as shown in FIG. FIG. 26 shows an image of search.

  In reading the initial solution (step 400), the solution optimized by the previous optimization process or the solution created by the initial solution creation process is read. The evaluation function of the read initial solution is calculated based on the equation (1) (step 401).

  In step 402, a combination of all vehicles is created. In step 403, the combination created in step 402 and the vehicle combination are created, and the combination order is randomly arranged. FIG. 25 schematically shows the above. In the combination of vehicles, since the combination with itself is not created, a random arrangement order of N × N−1 × N−2 is created.

  Next, in order from the beginning of the random sequence (step 404), all the loads loaded in the two vehicles are moved by one vehicle, and the load of one vehicle is moved to one of the two vehicles (step 405). . One of the two vehicles is selected at random. FIG. 26 schematically shows the above.

  If the solution is improved in the evaluation function calculation (step 406) (step 407), the process is advanced by one without performing the process of step 408 (step 409), and the process is repeated until the random order number is exceeded.

  If the evaluation value is not improved, the replaced vehicle is returned to the original state (step 408), the order is advanced by one (step 409), and the process is repeated until the random order number is exceeded.

  In step 411, after determining whether or not a search has been performed for all combinations in the random order number (step 410), if there is an improvement in the evaluation value in all cases, the process returns to step 405 for further improvement. If all the evaluation values have not been improved, the final improved solution is output (step 412), and the search is terminated.

  In step 35 of FIG. 15 (optimization processing 35 using the operation cycle as a search unit), the operation cycle is regarded as Job, and the vehicle is regarded as Agent, so that the processing flow is the same as that of FIG.

  Step 36 in FIG. 15 (leveling operation 36) is performed by moving the trip source trip to another vehicle as shown in FIG. The processing flow is shown in FIG.

In step 501, the order of the movement source vehicles is arranged at random (see FIG. 29).
All combinations of trips and vehicles are created, and the order is randomly arranged as shown in FIG. 30 (step 502). Although omitted in the middle, in step 515, if the order number of the movement source vehicles is exceeded, an evaluation function is calculated (step 516), and the search ends.

  In step 37 of FIG. 15, it is determined whether or not the optimization process by a series of search units has been performed a predetermined number of times. If the specified number of times has not been performed, the process returns to step 31 to improve the solution again by optimization processing. If it has been performed a predetermined number of times, the best solution is stored (step 38).

  In step 39, it is determined whether the initial solution has been made a predetermined number of times. If the predetermined number of times has not been created, an initial solution is created and an optimization process is performed. When the initialization has been created a predetermined number of times, the search scenario processing ends.

  As described above, in the present invention, the selection of the optimization processing method and the search order of the selected optimization processing method are referred to as a search scenario, and the search scenario can be changed variously depending on the difference in the planning target. Each optimization process may be a single unit, or any combination thereof, and there is an advantage that it is possible to flexibly cope with a target change in the vehicle operation plan. Furthermore, the search range increases by adding a plurality of delivery bases to the planning target. However, the calculation time can be shortened by selecting the optimal search scenario according to the difference in the planning target.

  FIG. 31 shows an example of the effect of the present invention. Case A is a case where 198 coils that are steel products are delivered, and cases B, C, and D are cases where the number of coils is 278, 229, and 182 respectively. The number of vehicles, the loading rate, and the cost for delivery (normalized by the actual value of Case D) are compared with the actual value and the planned value according to the present invention. Of the number of vehicles, the real number indicates the number of vehicles actually prepared (should), and the total number indicates the number of vehicles including multiple trips. The total number divided by the real number is shown as the rotation rate. This rotation rate is 1.00 unless there are a plurality of trips, and the value increases as the number of trips increases, and the vehicle is used more efficiently. The loading rate indicates an average value of the loading rate of the load on each vehicle including a plurality of trips. In addition, the total of fuel costs, overtime costs, overtime costs, fixed costs, etc. is shown as delivery costs.

  Compared to actual results, the vehicle operation plan according to the present invention can reduce the number of vehicles used for delivery by 1 to 4 vehicles (2.5 on average), and the delivery cost is also 1.75 to 4.07% (on average) 3.24%) There is an effect that it can be reduced.

  In addition, when creating a new vehicle operation plan, search processing should be performed by adding at least the last operation cycle at the time of the previous plan to the delivery plan data, vehicles, etc. for the new plan. Can be optimized. For example, if only the return travel of the final operation cycle is changed, the above-described FIG. 33 can be optimized as shown in FIG. As a similar example, FIG. 35 shows an example in which the number of delivery destinations is increased by one, and only delivery order data and vehicles different from the previous plan target (from delivery destination B to return to delivery base A (vehicle base) (4)) As a search process. On the other hand, when the present invention is applied and the return travel (4) of the final operation cycle is added to the new plan object, long-distance travel such as travel (6) and (7) in FIG. Optimization is achieved by eliminating unnecessary idle driving (see FIG. 36).

It is a figure which shows a vehicle operation plan creation computer processing flow. It is a figure which shows an example of an apparatus structure. It is a figure which shows an example of a delivery order data item. It is a figure which shows an example of a vehicle master item. It is a figure which shows an example of a distance master item. It is a figure which shows an example of a delivery destination master item. It is a figure which shows an example of a fare master item. It is a figure which shows an example of a base master item. It is a figure which shows an example of a virtual revenue and cost master item. It is a figure which shows the processing flow of optimal operation plan creation processing. It is a figure explaining an operation cycle (trip). It is a figure which shows the example which a delivery work complete | finishes within predetermined working hours. It is a figure which shows the example which delivery work does not complete | finish within the predetermined working hours. It is a figure which shows the program which described an example of the search scenario. It is a figure which shows the processing flow of the search scenario of FIG. It is a figure which shows an example of the detailed process flow in the optimization process which used the load as a search unit. It is a figure explaining the random combination order creation of Job and Agent. It is a figure explaining the replacement process 1 of Job. It is a figure explaining the replacement process 2 of Job. It is a figure explaining the replacement process 3 of Job. It is a figure which shows an example of the detailed process flow in the optimization process which used 1 vehicle and 1 vehicle as a search unit. It is a figure which shows creation of the random combination order of vehicles. It is a figure which shows the replacement | exchange process of a vehicle. It is a figure which shows an example of the detailed process flow in the optimization process which used 2 vehicles and 1 vehicle as a search unit. It is a figure which shows creation of the random combination order of 2 vehicles and 1 vehicles. It is a figure which shows the exchange process of 2 vehicles and 1 vehicle. It is a figure which shows the leveling of a driving cycle. It is a figure which shows an example of the detailed process flow in the leveling of an operation cycle. It is a figure which shows creation of the random order of a movement origin vehicle. It is a figure which shows creation of the random combination order of Trip and vehicles. It is a figure which shows an example of the effect of this invention. It is a figure explaining an example of delivery. It is a figure which shows the example of the operation plan of a certain vehicle. It is a figure which shows the example of the operation plan without waste in the example of FIG. It is a figure which shows the other example of the operation plan of a certain vehicle. It is a figure which shows the example of the operation plan which applied this invention to the example of FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Vehicle operation plan creation computer 2 Master management server 3 Order receiving system 4 Printer 5 Computer network 10 Plan creation data selection process 11 Data reading process 12 Optimal operation plan creation process 13 Vehicle operation plan display process 14 Vehicle operation plan transmission process

Claims (2)

In a method for collecting vehicles at multiple delivery bases and creating an operation plan for vehicles delivered to multiple destinations,
A process of selecting planning data that prompts the user to enter and select data relating to the package to be carried, ie delivery order data;
A data reading process step of reading the delivery order data, master data necessary for creating a vehicle operation plan, and a search scenario indicating a search order and a search method;
The sum of the difference between the vehicle revenue obtained by each vehicle and the vehicle operating cost when operating each vehicle is used as an objective function, and the delivery order data and the master are searched using a search process that maximizes the objective function. Based on the search scenario from the data, an optimum operation plan creation processing step of creating a vehicle operation plan,
A vehicle operation plan output process for outputting the created vehicle operation plan,
In the plan creation data selection processing step and the data reading processing step, the vehicle operation plan creation method is characterized by adding at least the last operation cycle at the time of the previous plan as a target. In a device that collects cargo at multiple delivery bases and creates an operation plan for vehicles delivered to multiple delivery destinations,
Selection processing means for planning data for prompting input and selection of data relating to a package to be carried, that is, delivery order data;
A data reading processing means for reading the delivery order data, master data necessary for creating a vehicle operation plan and a search scenario indicating a search order and a search method;
The sum of the difference between the vehicle revenue obtained by each vehicle and the vehicle operating cost when operating each vehicle is used as an objective function, and the delivery order data and the master are searched using a search process that maximizes the objective function. Based on the search scenario from the data, an optimum operation plan creation processing means for creating a vehicle operation plan,
A vehicle operation plan output means for outputting the created vehicle operation plan,
The vehicle operation plan creation apparatus characterized in that the plan creation data selection processing means and the data reading processing means add at least the last operation cycle at the time of the previous planning.

Images