JPH09244706A - Method and device for plan correction - Google Patents

Abstract

PROBLEM TO BE SOLVED: To correct a plan so that the correction is suppressed locally by finding an optimum solution which meets essential restrictions as to one partial plan to the utmost and further finding an optimum solution which meets target restriction based upon the condition that a previous plan is not corrected as much as possible when the essential restrictions are met. SOLUTION: A partial plan correction part 60 when reallocating remaining operation styles to remaining resources in response to a change of the operation style of a resource at a specific date performs the reallocation of the date by referring to restriction conditions 34 and a rule base 35 so that the cost in case of a violation of the restriction conditions is minimum, and generates a corrected plan 32. A correction path search part 40 decides whether or not the essential restriction conditions are met and repeats the process of the partial correction part 60 when not. If a sufficient solution can not be obtained even by making a search as to a specific number of succeeding days, that is reported to a plan adjustment part 80, which supports the relaxation of the restrictions.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for correcting a plan created in advance, and more particularly to a plan correction method and apparatus for reducing the correction amount as much as possible.

[0002]

2. Description of the Related Art A method of allocating resources to work using an electronic computer, that is, a so-called scheduling method has been studied for a long time, and various scheduling systems have been put to practical use. However, most of the practically used scheduling systems are pre-plans that are created prior to the operation of the schedule, and if corrections are required during the operation of the schedule, they must be manually corrected or rescheduled. It is the current situation. When correction is performed by re-planning in this way, a plan that drastically modifies the pre-plan is created, which may result in extra operating costs and confusion for people involved in the operation of the plan. is there.

As a technique relating to plan modification, there is, for example, Japanese Patent Laid-Open No. 4-168570. This publication only describes providing a plan modification function on a computer.
In addition, "Solving Large-Scale Co
nstraat Satisfaction and
Scheduling Problems Usin
ga Heuristic Repeat Meth
od "(Proceedings of the Ei
ghth National Conference
on Artificial Intelligenc
e, pp. 17-24, 1990), the plan correction system corrects a conflicting point and continues to correct the conflicting point until there is no new conflicting point caused by this correction. . However, in this method, it is sometimes difficult to suppress the correction locally, because the correction of a certain portion may cause a new conflicting point and a potentially large correction amount.

In the plan modification, there is no guarantee that there is a plan that satisfies a given constraint with a small modification amount, and it is necessary to appropriately relax the initially defined constraint. As a technique of using constraint relaxation for creating a plan, for example, Japanese Patent Laid-Open No. 4-819
No. 68 publication can be cited. This technique identifies a part that violates the constraint and relaxes the low-priority constraint for that part. However, it does not search for the location of constraint violation that requires the least constraint relaxation. Therefore, it may be difficult to suppress the correction locally.

[0005]

As described above, there is no known plan correction technique for suppressing the correction amount as small as possible.

The present invention has been made in view of such circumstances, and an object of the present invention is to provide a plan correction method and apparatus capable of suppressing correction as little as possible.

[0007]

According to the present invention, a predetermined time is divided into a plurality of continuous time zones, and a plurality of resources for each of a plurality of resources that can be worked in parallel and a plurality of operations are assigned to each of the time zones. When a plan to which one of the resources is assigned is set, an indispensable constraint condition to be observed when assigning the operation form to the resource is set, and the operation form of at least one resource belonging to a specific time zone of this plan is set. Of the resource to be changed in response to the change of the resource and the remaining resource except the resource after the change, at least one of the time zones adjacent to this specific time zone The present invention is characterized by a plan correction method and apparatus for satisfying a constraint condition existing therebetween and performing reallocation for this specific time zone so as to minimize the total cost when the constraint condition is violated.

According to the present invention, a plan (partial plan) for one time zone is set as a unit, and an optimum solution that satisfies the necessary constraint condition for this partial plan is obtained. In addition, if a constraint condition that does not modify the preliminary plan as much as possible for the partial plan is introduced as a target constraint and a cost is also set for the target constraint, when there are multiple optimal solutions that satisfy the required constraint conditions, Among them, the optimal solution that does not modify the operation allocation in the partial plan can be obtained.

Further, when the optimal solution obtained for a specific partial plan is reassigned so as to violate the essential constraint condition, this partial plan is modified according to this optimal solution, and the partial plans adjacent to this partial plan are corrected. The above reallocation processing is repeated for one of the above. That is, a search for finding a partial plan that satisfies the essential constraint condition is propagated to the adjacent partial plan.

In this way, the search is sequentially propagated to the adjacent partial plans until the optimum solution satisfying the essential constraint condition is obtained. Propagate the search in the positive direction of the time axis,
If an optimum solution that satisfies the essential constraint cannot be obtained even if a search is performed for a predetermined number of continuous time zones, the search is further propagated in the negative direction and sequentially propagated to the adjacent partial plans.

[0011]

BEST MODE FOR CARRYING OUT THE INVENTION

(1) Introduction Hereinafter, an embodiment in which the plan correction method of the present invention is applied to a correction problem of a vehicle operation plan will be described with reference to the drawings.

FIG. 2 is called a vehicle operation table and is a diagram showing a situation in which the vehicle is operated during the time zone of one day. In the figure, the horizontal axis indicates time and the vertical axis indicates an operation number as an operation identifier. For example, the operation method or operation mode of the operation number 0 indicates that the operation is to leave the station A after 5 AM, inspect the train halfway and arrive at the station B after 23:00 (operation) The form is simply referred to as operation).

FIG. 3 is a diagram showing a vehicle operation plan. This plan is a plan that is first created prior to the operation of the schedule, and will be referred to as a pre-plan hereinafter. In the figure, the horizontal axis indicates the date of one month, and the vertical axis indicates the vehicle number as the vehicle identifier. The number placed at a position determined by the date and the vehicle number is the operation number, and any operation number 0 to 22 in the vehicle operation table is assigned. In this example, the operation is assigned to the resource of the vehicle.

The following constraints a to d are indispensable conditions for preparing a vehicle operation plan, and the pre-plan shown in FIG. 3 satisfies these indispensable constraints. Constraint a: The departure station of each vehicle on the same day as the terminal station of the previous day is the same Constraint b: Each vehicle is inspected within 3 days Constraint c: Short-term and long-term predetermined other than the inspection of constraint b There is an inspection plan, and each vehicle should not drive on the day corresponding to these inspections. Constraint d: The vehicle must be in the station to be inspected by the specified time on the day before the inspection in constraint c.

The following constraints e to g are constraints to be considered when modifying a plan. Constraint e: Replacement of a spare vehicle with the operation of the next day for repairing the next day for a failed vehicle Constraint f: Minimize the number of modifications in a day Constraint g: Plan to restore the target plan with the correction in the shortest possible time The constraint e is an essential constraint condition, and the constraints f and g are target constraint conditions. The target plan is a plan that is a pre-plan or a modified plan and is a target for return after modification.

The following constraint conditions of constraints h to i are constraints to be considered when modifying a plan. Constraint h: The end station of each vehicle on the day is the same as the departure station on the next sunrise Constraint i: Return to the pre-plan within 6 days When modifying the plan, constraint a and constraint h are mandatory constraint conditions , Constraint i is a target constraint condition.

While the vehicle operation plan shown in FIG. 3 is being operated, a situation may arise in which the advance plan must be modified due to, for example, a vehicle failure. At this time, due to the constraint e, the failed vehicle is first allocated to the preliminary operation on the next day for repair (it is reallocation when viewed from the advance plan). The plan correction problem handled in the present embodiment is to obtain a correction plan in which the advance plan is corrected from the day after the failure so that the above constraint condition is satisfied and the target plan can be restored within a few days from the failure date. is there.

According to the example of FIG. 3, the daily operation plan is 21
23 types of operations are assigned to one vehicle. When the operation plan of a certain day is changed, the number of allocation methods for allocating the operation to each vehicle when the constraint condition is not taken into consideration, that is, the number of operation plans that can be corrected on the next day is generally plural. When one of the possible operation plans of the next day is selected, the number of possible operation plans of the second day after the selected operation plan is further plural. In this way, when one operation plan on a certain day is used as a starting point, a set of operation plans for the next day derived from this, and a group of operation plans for the next two days derived,
The whole of ... Composes a search tree with each operation plan as one node. The present embodiment is not a method of searching all the routes one by one from the trunk of the search tree toward the branches, and obtains an optimal solution that satisfies the constraint conditions as much as possible for the operation plan for one day. As a result, if the required constraint conditions applied to the pre-plan are violated, the optimum solution is obtained for the operation plan for the next day. Use the method of going. Hereinafter, such a method of obtaining a solution will be referred to as a route search. The operation plan for one day may be called a partial plan. The partial plans on each date after the date of returning to the target plan are as in the target plan.

(2) First Embodiment FIG. 1 is a block diagram showing the arrangement of an information processing apparatus according to the first embodiment. The advance plan 31 is a table for storing the pre-correction advance plan as shown in FIG. The modification plan 32 is composed of one or a plurality of tables, and each table is a modification plan including a constraint violation or a final modification plan having no constraint violation. The partial plan correction unit 60 refers to the advance plan 31 and the correction plan 32 of the given table name to find an optimal operation plan that satisfies the constraint conditions as much as possible for the partial plan of the given date, and the partial operation is given by the result. It is a processing unit that corrects the modified plan 32. Corrected route search unit 40
Determines whether or not the partial plan calculated by referring to the modified plan 32 corrected by the partial plan correction unit 60 can return to the target plan from the next day. If there is a possibility that it will be possible to return to the target plan if the partial plan cannot be recovered and the next day's partial plan is modified, the route search is performed according to the algorithm of driving the partial plan modification unit 60 for the partial plan of the next day. It is a processing unit.
The display device 36 is a device that is connected to the processing device 30 and displays a final modification plan or a modification plan proposal in which constraints are relaxed. The input device 37 is a device connected to the processing device 30 and for inputting instructions and data. Plan adjustment unit 80
Displays the final modification plan on the display device 36 according to the report or request from the modification route search unit 40, or displays the modification plan draft created by adjusting the constraint relaxation for the modification plan including the constraint violation. It is a processing unit displayed above.
The corrected route search unit 40, the partial plan correction unit 60, and the plan adjustment unit 80 can be realized by executing a program stored in the storage device of the processing device 30. The vehicle operation table 33 is a table that stores an operation table as shown in FIG. The constraint conditions 34 are files that store the constraint conditions when the vehicle operation plan is created and modified as described above. The rule base 35 includes the corrected route search unit 4
0, the partial plan correction unit 60 and the plan adjustment unit 80 are files that store rules when the determination processing is performed. Advance plan 31, revised plan 32, vehicle operation table 33, constraint conditions 34
The rule base 35 is a table or a file stored in the storage device of the processing device 30.

FIGS. 4a and 4b show a corrected route search unit 40.
3 is a flowchart showing the flow of the processing of FIG. The corrected route search unit 40 copies the advance plan 31 as a corrected plan 32 according to an instruction from the input device 37 via the plan adjustment unit 80, assigns a table name to the copied corrected plan 32, and the next day of the vehicle that has failed according to the constraint e. The modification plan 32 is modified so that the operation of is replaced with the spare vehicle (step 41).
Next, as a target plan, a primary route search is performed with the table name of the advance plan 32, the table name of the modification plan 32, the first modification date as the next day of the failure, and the direction of search propagation as the forward direction (the rising direction of the date) as parameters. (Step 42). Details of the route search are shown in Figure 4b. First, the corrected route search unit 4
For 0, the partial plan modification unit 60 is called with the table name of the modified plan 32, the modification date, the propagation direction, and the required operation allocation as parameters, and the partial plan is modified to satisfy the constraint condition as much as possible, and the result is given. It is stored in the correction plan 32 having the table name (step 51). The essential operation allocation here is to allocate a spare operation to the failed vehicle. In addition, it is specified when the allocation of vehicle operation set in the target plan is mandatory. Next, referring to the modified partial plan, the restoration constraint violation degree is calculated on condition that only the modification date is modified and the modification does not affect the next day along the direction of search propagation (step 52). . Here, the effect of the modification on the next day does not mean that the partial plan on the modification day does not violate the mandatory constraints of the constraints a to d and the constraint h, and returns to the target plan from the next day. . In calculating the return constraint violation degree, for example, the following rules stored in the rule base 35 are applied.

Rule p: if It is possible to return to the target plan from the next day by the modified partial plan. Then Restoration constraint violation degree = 0 Rule q: if If n more checks can be put in the modified partial plan, then It is possible to return to the target plan from the day of the next. The return constraint violation degree = n × 100 It is also possible to set other rules. The entire restoration constraint violation degree is the sum of restoration constraint violation degrees for all rules. If the return constraint violation degree is 0 as a result of the calculation (YES in step 53), the plan adjusting unit 80 is notified of the end with the table name of the modified plan 32 as a parameter (step 54), and the process ends. To do. When the return constraint violation degree is not 0 (step 53 NO), it is determined whether there is a prospect of returning to the target plan due to the modified partial plan (step 55). For example, constraints b, c, d
If you do not violate, you may return to your goal plan. On the contrary, if the constraints b, c, and d are violated, there is no expectation that the constraint violation will be recovered thereafter, and it is meaningless to proceed the search propagation further. When recovery is possible (step 55YE
S) If the processes of steps 51 to 55 have not been performed a predetermined number of times (step 56 NO), the calculated restoration constraint violation degree is saved and the correction date is set to the next day (step 5
7) Return to step 51 and repeat the above process. The predetermined number of times in step 56 is 5 according to the condition of “return to the pre-plan within 6 days” of the constraint i. When the processes of steps 51 to 55 have been performed a predetermined number of times (YES in step 56), the process goes to step 43, the table name of the target plan, the table name of the modified correction plan 32,
Each correction date and the restoration constraint violation degree are reported to the plan adjustment unit 80. If there is no prospect of returning to the target plan (NO in step 55), a non-recoverable end report is made to the plan adjusting unit 80 using the corrected table name of the modified plan 32 as a parameter (step 59), and the process ends.

When control returns from the plan adjusting unit 80, the corrected route search unit 40 saves the corrected correction plan 32 lastly modified, so that it is copied and a new table name is given (step 44). Revised plan 3 which is the next target plan
The secondary route search is performed with the table name of No. 2, the new table name, the first modification date as the last modification date, and the search propagation direction in the reverse direction (downward direction of date) as parameters (step 45). Here the target plan is the last modified revision plan 3
2. The correction route search unit 40 calls the partial plan correction unit 60 with the table name of the new correction plan 32, the correction date, the propagation direction, and the required operation allocation as parameters, and has the table name that specifies the result by correcting the partial plan. It is stored in the correction plan 32 (step 51). If the modification date is the day after the vehicle failure, specify the required operational allocation. Next, the return constraint violation degree is calculated as described above (step 5
2) If the return constraint violation degree is 0 (step 53YE)
S), and an end report with the table name of the modified revision plan 32 as a parameter to the plan adjustment unit 80 (step 5).
4), end the process. If the return constraint violation degree is not 0 (step 53 NO) and the return is possible, steps 51 to 5
When the process of 5 is not performed a predetermined number of times (step 56N
O) and set the modification date in the reverse direction to the next day (step 5).
7) Return to step 51 and repeat the above process. When the processes of steps 51 to 55 are performed a predetermined number of times, that is, when the process of the partial plan on the day following the vehicle failure date is performed (YES in step 56), the process goes to step 46, the table name of the target plan, and the modified plan 32 that has been modified. Table adjusting unit 80, the respective correction dates and the restoration constraint violation degree
Report to

When the control is returned from the plan adjusting unit 80 again, and when the processes of steps 44 to 46 have not been performed the predetermined number of times (NO in step 47), the process returns to step 44 and the correction plan 32 that has been finally corrected is copied. After giving a new table name, the secondary route search of step 45 is performed.
The parameters of this secondary route search are the table name of the modification plan 32 that was the final modification process, which is the target plan, the new table name, the modification date is the last modification date, and the search propagation direction is the final processing direction in step 45. And the opposite direction. In this way, the search propagation direction is negative, positive, negative, ...
When the secondary route search of is performed a predetermined number of times (step 47).
YES), the plan adjustment unit 80 is requested to make an adjustment (step 48).

The partial plan modification unit 60 modifies the partial plan having the given modification date by referring to the modification plan 32 having the given table name, and updates the modification plan 32 having the same table name with the modified partial plan. To do. An example of the correction algorithm is shown below. The problem is that variable c _ij is c _ij = 1: vehicle i is assigned to operation j. c _ij = 0: vehicle i is not assigned to operation j. And x _ij is the cost (or penalty value) when allocating the vehicle i to the operation j,

[0025]

[Equation 1]

C _ij that minimizes the objective function F given by
Is to seek. n is the number of vehicles and m is the number of operations. Here, the cost x _ij is given by the following equation.

[0027]

[Equation 2]

Here, δ _k is the weighting of the cost for the constraint k, and y _kij is y _kij = 1: conflicts with the constraint k when the operation j is assigned to the vehicle i. y _kij = 0: do not conflict. It is. CND is a set of constraints shown below. That is, x _ij is obtained by adding δ _k y _kij for all k in the set CND. The following conditions are imposed on the variables c _ij .

[0029]

(Equation 3)

[0030]

(Equation 4)

That is, i is 1 for c _ij for all j
1 is the sum of the numbers from 1 to m, and 1 is the sum of c _ij from 1 to m for all i.

An example of the values of the set CND of constraints and δ _k stored in the rule base 35 is given below. k = 1: the current departure station of vehicle i is different from the departure station of operation j; δ ₁ = 10000 k = 2: vehicle i requires inspection but operation j does not have inspection; δ ₂ = 10000 k = 3: Vehicle i is a monthly inspection; when the monthly inspection is not in operation; δ ₃ = 10000 k = 4: Vehicle i does not return to the base by the time specified on the day before the monthly inspection; Return to the base by scheduled time When not in operation; δ ₄ = 10,000 k = 5: The end station of the operation of vehicle i determined in the pre-plan is different from the end station of operation j; δ ₅ = 100 k = 6: The operation determined in the pre-plan Operation j is different; δ
₆ = 100 k = 7: There is an inspection in the operation decided in advance plan,
In operation j, there is no check; δ ₇ = 100 In the set CND of constraints, k = 1 to 4 correspond to the mandatory constraint violation about the target plan, and k = 5 to 7 correspond to the target constraint violation about the pre-plan. Equivalent to. The cost in the case of violation of the mandatory constraint is significantly higher than the cost in the case of violation of the target constraint. If there are no mandatory constraint violations, the optimal solution that minimizes the total cost of goal constraint violations is obtained.

When the required operation allocation is specified, the specified vehicle and its operation are excluded from the above optimization processing. For example, regarding the first correction date, the operation allocation of the failed vehicle is inevitably determined because the failed vehicle is assigned the operation of the spare vehicle, and the failed vehicle and the spare vehicle are excluded from the target vehicle i and operation j. In addition, the direction of search propagation is considered according to the essential constraints a and h at the time of plan modification, and for the given modification date, the direction opposite to the direction of search propagation is applied to the vehicle i so that the boundary conditions are matched with the previous day. j must be assigned. For example, when the search propagation direction is the forward direction, at least the departure station must be assigned so as not to violate the constraint with respect to the ending station on the previous day. In addition, when the search propagation direction is negative, at least the terminal station must be assigned so that it does not violate the constraint to the station departing the next day.

As described above, when the return constraint violation degree of the partial plan becomes 0 and the end is reported to the plan adjusting unit 80 (step 54), even if a partial plan in which the return constraint violation degree is not 0 occurs in the middle. Amended plan 32 to be reported
Has no mandatory constraint violations. When reporting a return constraint violation to the plan adjusting unit 80 (steps 43 and 46), there is no essential constraint violation in the other partial plans except the last modified partial plan.

FIG. 5 is a flow chart showing the flow of processing of the plan adjusting unit 80. The plan adjusting unit 80 not only transfers the data input from the input device 37 to the corrected route search unit 40 at the start of the plan correction process, but also performs the process shown in FIG. 5 according to a report or request from the corrected route search unit 40. The plan adjusting unit 80 first branches the process according to the type of report or request (step 81). At the end report, the revision plan with the given table name is read (step 8).
2) The revised plan is edited and displayed on the display device 36 (step 83). When it is impossible to return to the target plan, another solution can be obtained by returning to a certain day that has been corrected so far, according to the backtracking method described in the third embodiment below. When reporting a return constraint violation,
The received table name, each correction date, and the restoration constraint violation degree are accumulated in a table on a storage device (not shown) (step 84), and the process returns to the correction route search unit 40 (step 85).
In the case of an adjustment request, the table that stores the restoration constraint violation degree for each correction date for each correction plan is referred to, and the partial plan with the smallest restoration constraint violation degree is selected for the first correction plan (step 86). Advance planning 3
1 or the modified plan 32 and the reported modified plan 32, the operation attributes required for each vehicle to return to the target plan on the next day are set for the selected partial plan (step 87). Here, the operational attribute refers to a departure station, a termination station, presence / absence of inspection, and the like.

For example, FIG. 6 (a) is a diagram showing the operation history of five days before and after, focusing on the modification date of a certain vehicle. According to this operation history, there is no inspection before and after the modification date. Therefore, in order to return this vehicle to the target plan on the next day, it must be inspected on the day of modification due to constraint b "Each vehicle must be inspected within 3 days". Need to receive. In addition, since the ending station on the day before the correction is B and the departure station on the next day of the correction is B, in order for this vehicle to return to the target plan on the next day due to constraint a and constraint h, this vehicle has the departure station B, It is necessary to operate the terminal station B. In this way, the operation attribute required for this vehicle is the presence / absence of inspection, and the departure station is B and the end station is B. FIG. 6 (b) shows the actual operation of each target vehicle and the necessary operation of each vehicle in parallel.

Subsequent to the process of step 87, the actual operation of each vehicle is compared with the operation required for returning to the target plan, and the vehicles whose operation attributes do not match are extracted (step 88). According to the example of FIG. 6B, the operation attributes of the vehicle with vehicle number 4 do not match. Next, with regard to the vehicles whose operation attributes do not match, the vehicle is selected as a candidate when the constraint is relaxed by adopting the necessary operation (step 89). Specifically, the cost when the constraint is relaxed is calculated with reference to the rule base 35 and the vehicle operation table 33. Below are some examples of rules. Rule s: early inspection for operation after departure at 7 o'clock; cost 100 rule t: inspection for spare vehicle; cost 120 For example, operation of vehicle number 4 in FIG. 6 (b) after 7 o'clock If it is a departure of, you can apply the rule s and relax the constraint,
The cost at that time is 100. The total cost of the partial plan is the sum of all the costs of operation with relaxed constraints. Next, if the processes of steps 86 to 89 have not been performed a predetermined number of times (NO in step 90), the process returns to step 86, and the table in which the restoration constraint violation degree is accumulated is referred to,
In the same modified plan, the partial plan having the second smallest return constraint violation degree is selected and the above process is repeated. Steps 86-89
When the above-described processing has been performed a predetermined number of times or the target partial plan has been exhausted (YES in step 90), the partial plan with the minimum cost when the constraint is relaxed by the above processing is corrected to the constraint-relaxed partial plan, and this constraint relaxation is performed. The revised plan including the partial plan thus edited is edited and displayed on the display device 36 (step 91). It is advisable to highlight the operation with relaxed constraints. In addition, the cost due to constraint relaxation is also displayed. In addition, each modified partial plan in the same modified plan is modified into a modified partial plan, and a modified plan including the modified partial plans is displayed on the display device 36, and the operator of the information processing device corrects each modified constraint. You may decide whether to adopt the plan. By doing so, it is possible to check each partial plan for which the constraint is relaxed by a person without depending on the rule registered in the rule base 35. If there is an instruction from the input device 37, the plan adjusting unit 80 returns to step 86 and repeats the processes of steps 86 to 91 for the next correction plan.

According to the above embodiment, one modification plan 3
Among the two, the partial plan with the smallest return constraint violation degree is selected, and the constraint of the partial plan with the minimum cost when the constraint is relaxed is relaxed. In this way, if one partial plan with the lowest cost is selected and the constraint is relaxed, it is possible to return to the target plan from the next day along the direction of constraint propagation. The final revised plan is the revised plan 32 in which the constraints are relaxed up to the partial plan in which the constraints have been relaxed along the direction of constraint propagation, and the target plan from the next day. Is bound at the boundary between and the partial plan for the next day. Then, the final revised plan will be restored to the advance plan 31 within 6 days according to the constraint i.

(3) Second Embodiment In the first embodiment, the operation for one day is assigned to each vehicle. That is, the operation assigned to each vehicle is on a daily basis. If the daily operation is expanded according to the time, it will be subdivided into smaller units. The second embodiment is intended to overcome the constraint violation of the modified plan by allocating the vehicle thus divided into units as a unit.

FIG. 7 is a diagram showing an example of vehicle operation called a mountain diagram. FIG. 7A is a mountain diagram showing details of operation with operation number 0. The horizontal axis of the figure indicates the time, and the vertical axis indicates the station where the vehicle stops. The mountain diagram is drawn as a polygonal line consisting of several line segments, and each line segment indicates at which time the vehicle is running on a line connecting which station to which station. From the figure, it can be seen that the daily operation can be divided in units of line segments. FIG. 7A is a mountain diagram showing the departure station A, the end station B, and the operation with inspection, which is developed.
7 (b) is also a mountain schedule similar to that of FIG. 7 (a), but it is a mountain schedule showing the details of the operation of operation number 4, and the departure station B, the end station C, and the operation without inspection are developed. It is shown.

The second embodiment has the same system configuration as that of the first embodiment except that it has a file for storing mountain schedules for each operation stored in the vehicle operation table 33. The plan adjustment unit 80 receives a request for adjustment from the corrected route search unit 40 and attempts to relax the constraint, but considers a case where the constraint a or the constraint h cannot be observed and the constraint relaxation is difficult. For example, the actual operation and the necessary operation targeted for two vehicles are shown in FIG. Required operation of vehicle number 5 is shown in Fig. 7.
The operation of operation number 0 shown in (a), and the necessary operation of vehicle number 6 is the operation of operation number 4 shown in FIG. 7 (b).
However, the target operation is different for each terminal station of the vehicle, so there is no such operation. In this case, the operation of operation number 0 and operation number 4 are both time 14:00 to 1
Since it is stored in station C between 5:00, both operations are divided between times 14:00 and 15:00 and operation number 0
When the first half of the operation of No. 5 and the second half of the operation of No. 4 are connected, the operation targeted by vehicle No. 5 is satisfied. When the first half of the operation of operation number 4 and the second half of the operation of operation number 0 are connected, the operation targeted by vehicle number 6 is satisfied and the constraint violation is resolved.

As shown in FIG. 8, the plan adjusting unit 80 displays information about vehicles whose operation attributes do not match on the display device 36 and allows the operator to instruct necessary operation through the input device 37. Therefore, the file storing the mountain diagram is searched, the mountain diagram of the designated operation number is read and displayed on the display device 36. In accordance with an instruction from the operator, an operation number is given to each operation obtained by dividing the daily operation and the correction plan corrected is displayed on the display device 36.

In the second embodiment, the operation for one day is assigned to the vehicle as much as possible, and the operation for the vehicle is narrowed down by the conditions of the minimum constraint violation degree and the minimum cost for relaxing the constraint. For operation in which the constraint cannot be relaxed, the data of mountain schedule as shown in FIG. 7 is applied to overcome the constraint violation. In this way, the search space can be reduced by hierarchically narrowing down the target.

(4) Third Embodiment In this embodiment, a backtracking method is further added to the route search method of the first embodiment. According to the constraint i, there is a target constraint of “returning to the advance plan within 6 days”, and in the first embodiment, the predetermined number of times of step 55 is set to 5 in accordance with this constraint. Then, when the predetermined number of times is reached in step 55, the process goes to step 43 or 46 to report the violation of the return constraint to the plan adjusting unit 80. The present embodiment changes the cost in the case of violation of the target constraint of the constraint set CND between step 42 and step 43 and between step 45 and step 46, and copies it in order to save the modification plan 32 created at the end. Then, a new table name is given and the processing of step 42 or step 45 is repeated.
That is, the modification date returns to the day after the failure or the last modification date, and the primary route search or the secondary route search is performed for the same search propagation direction. By performing the backtracking in this way, a solution that satisfies the essential constraints may be obtained. When the constraint track cannot be obtained even by the backtrack,
Going to step 43 or step 46, the plan adjusting unit 80 is notified of the correction plan 32 obtained by the original route search and the correction plan 32 obtained by the backtrack. When there are multiple correction routes for the backtrack, the same correction plan 3 is used for each computer that constitutes the parallel computer.
2 is given, and route search processing (step 42 or 4
5) can be performed in parallel.

(5) Other Applications In the above embodiment, the problem of correcting the vehicle operation plan has been described as an example. However, in general, a plurality of time zones (1 hour, 1 day, 1 week, etc.) that continue for a predetermined time period are described. It can be thought of as a problem of modifying a plan that divides into one and allocates one of a plurality of resources that can work in parallel to one of the operation modes of the resource to one of the time zones and to other applications. It is easy to apply. For example, if a nurse is regarded as a resource and its work form is regarded as an operational form, it can be applied to a nurse work plan in a hospital. At this time, the conditions of the mandatory constraint and the target constraint may be set according to the circumstances of the hospital.

[0046]

According to the present invention, an optimum solution satisfying an essential constraint is calculated for one partial plan as much as possible in a unit of a partial plan, and when the essential constraint is satisfied, the advance plan is not modified as much as possible. By finding an optimal solution that satisfies the target constraint, it is possible to modify the plan locally.

[Brief description of drawings]

FIG. 1 is a configuration diagram of an information processing apparatus according to an embodiment.

FIG. 2 is a diagram showing a vehicle operation table of the embodiment.

FIG. 3 is a diagram showing a vehicle operation plan of the embodiment.

FIG. 4a is a flowchart showing a processing flow of a corrected route search unit 40 according to the embodiment.

FIG. 4b is a flowchart showing a process flow following FIG. 4a.

FIG. 5 is a flowchart showing a processing flow of a plan adjusting unit 80 according to the embodiment.

FIG. 6 is a diagram showing an operation history, a target vehicle operation, and a necessary operation according to the embodiment.

FIG. 7 is a diagram showing a mountain diamond according to the embodiment.

FIG. 8 is a diagram illustrating a target vehicle operation and a necessary operation according to the embodiment.

[Explanation of Codes] 32 ... Revised Plan, 34 ... Constraints, 40 ...
Corrected route search unit, 60 ... Partial plan correction unit, 80 ...
・ Plan adjustment department

Claims (10)

[Claims] 1. A predetermined time is divided into a plurality of consecutive time zones, and a method of operating the resource (operation mode) is determined for each of the plurality of resources that can work in parallel and each of the time zones. A method in which one of them is allocated is regarded as a plan, and the plan is modified. In the method, the plan in which the resource and the operation form are respectively expressed by an identifier is stored in the first storage device, and the operation is performed in the resource. An essential constraint condition to be observed when allocating a form is stored in the second storage device, and a cost or penalty value in case of violating the constraint condition when allocating the operation form to the resource is stored in the third storage device. , Reallocating the remaining operation modes to the resources that are changed in response to the operation mode change of at least one resource belonging to a specific time zone of the plan and the remaining resources except the changed operation mode Sometimes at least Characterized by satisfying a constraint condition existing between the specific time zone and one of the adjacent time zones, and performing the reallocation for the specific time zone so as to minimize the total cost. How to fix the plan. 2. In addition to the indispensable constraint condition, a target constraint on the condition that the original plan is not changed as much as possible is stored in a second storage device, and a cost when the target constraint is violated is thirdly stored. 2. The method of modifying a plan according to claim 1, wherein the re-allocation is performed so that the total cost is stored in the storage device and the total cost is minimized for the specific time zone. 3. When a reallocation satisfying all the constraints for the specific time zone is completed, a modified plan including the change of the operation mode and the reallocation is displayed on a display device. The method of modifying a plan according to claim 1, wherein 4. When the reassignment results in reassignment that violates the constraint condition, the plan of the particular time zone is modified by the change of the operation mode and the result of the reassignment, and the identification is performed. Performing the reallocation process so as to satisfy the constraint condition with the specific time zone for one of the time zones adjacent to the time zone and modifying the plan for the one of the adjacent time zones. The method for modifying a plan according to claim 1, wherein: 5. The restoration constraint violation degree, which expresses the degree of constraint violation by a numerical value, is calculated when the reassignment results in reassignment that violates the constraint condition as a result. Plan correction method. 6. Predetermination for a plurality of consecutive time zones, such as the reassignment process and plan modification for one of the adjacent time periods, and the reassignment process for one of the adjacent time periods. After repeating the reassignment process for the number of times, the reassignment is violated to the constraint condition, and when the reassignment ends, the reassignment is performed for a time zone that is adjacent in the time direction starting from the last reassigned time zone. Processing is performed to correct the plan for the adjacent time zones in the opposite direction, and the reassignment processing up to the predetermined number of times for a plurality of consecutive time zones in the opposite direction as long as the reassignment violates the constraint condition. 5. The method of modifying a plan according to claim 4, further comprising sequentially repeating the above and the modification of the plan. 7. The plan modification according to claim 6, wherein a restoration constraint violation degree that numerically expresses the degree of constraint violation is calculated for each of the time zones that are terminated by reallocation that violates the constraint conditions. Method. 8. A predetermined time is divided into a plurality of continuous time zones, and a method of operating the resource (operation mode) is determined for each of the plurality of resources that can work in parallel and each of the time zones. In a device that regards one to which one of them is allocated as a plan and corrects the plan, a first storage device that stores the plan that represents the resource and the operation form by an identifier, and the operation that is performed in the resource A second storage device that stores an indispensable constraint condition that must be observed when assigning a form, and a third storage device that stores a cost or penalty value when the constraint form is violated when assigning the operation form to the resource as a rule. The device and the resource to be changed in response to the change of the operation form of at least one resource belonging to a specific time zone in the plan and the remaining operation form except the changed operation form are re-reproduced. Assign At least one of the time slots adjacent to the particular time slot is satisfied, and the reallocation is performed for the particular time slot so as to minimize the total cost. A plan amendment apparatus comprising: 9. The second storage device stores a target constraint on the condition that the original plan is not changed as much as possible in addition to the essential constraint condition, and the third storage device further stores the target constraint. 9. The plan correction device according to claim 8, wherein the cost when the cost is violated is stored, and the processing means performs the reallocation so as to minimize the total of the costs for the specific time zone. 10. The processing means, when the reassignment satisfying all the constraint conditions for the specific time zone is completed, displays the corrected plan including the change of the operation mode and the reassignment. 9. The plan correction device according to claim 8, which is displayed above.