JPH0692237A - Method and system for solving problem - Google Patents

Abstract

PURPOSE:To automatically solve a complicated problem even with few input items by providing a patterned case base for preserving patterned cases having case applying patterns added to case solving patterns and providing steps of preservation, take-out and proposition of solutions. CONSTITUTION:In the case of preparing a train schedule for example, a case taking-out reasoning engine 1602 of a central processing device 1 selects a patterned case 90 selected according to purpose, strategy, etc., from the knowledge stored in a partial schedule case knowledge table 7 to register the pattern case in a practicable partial schedule pattern table 20. Next, a case developing section 1604 automatically prepares the partial schedules in respecting time bands according to the registered partial schedule patterned case 20 (automatically developing to a case by the use of case solving pattern knowledge 907 or taking out simply the partial schedule case) to connect to the partial schedule. These are displayed on a display device 2 at any time and if necessary requested of correction by the use of a key board 3, mouse 4, etc., to be stored in the schedule table 6.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a schedule for personnel, trains, buses, airplanes, and the like, and more particularly to a similar timetable retrieval system.

[0002]

2. Description of the Related Art Japanese Unexamined Patent Publication (Kokai) No. 4-78953 discloses a method of creating a partial diamond and a method of creating a total diamond by combining partial diamonds. In addition, the 28th railway cybernet paper collection (1991) introduces a partial timetable creation system for the Tokaido Shinkansen in "Tokaido Shinkansen timeline creation system". As described above, the problem of creating a complicated diamond is solved by problem division. That is, a technique for creating a partial diamond first and then combining these to create a daily diamond (the former) and a technique for creating a partial diamond with a low automatic level (the latter) are known.

[0003]

[Problems to be solved by the invention] Creating a timetable is a complicated problem involving not only physical conditions such as routes and vehicles, but also manpower operations such as crew members and vehicle inspectors, and artificial conditions such as sales and political factors. . As a solution to such a complicated problem, a method of dividing into partial problems as in the above-mentioned publication is effective, but a method of utilizing past successful cases is also effective. In particular, when many departments and conditions are involved, such as making a diamond, it is practically difficult to accept a large-scale major change except for the new route construction (making a blank diamond). However, it is often difficult to adapt to the current situation even if the precedents and successful cases are used, and it is necessary to recreate whether or not other parts are other cases or completely adapting to the current situation. In order to satisfy such needs, it is an object of the present invention to make it possible to quickly search for cases that can be easily applied / modified and applicable parts / applicable objects (objectives) thereof. It is fundamentally different from the earlier planning method for problem solving.

The prior plan planning method (Hitachi) has a partial plan or case solution patterns (cases and solution cases) obtained by patternizing the partial plan, but case application patterns (such as application conditions, application keys, and adaptation purposes) ( Abstract element that characterizes the application of cases (necessary for selecting applicable cases). That is, it does not have an information structure (data block), that is, a patternized case, which is roughly divided into case solution patterns (cases and solution cases) and case application patterns as distinct components. It is not easy to select a solution or solution method that suits the purpose or situation (the approach of the inference mechanism therefor). Also, specify patterning cases that can be applied by applying to objectives and strategies for dividing and achieving objectives,
It does not describe a method of selecting case solution patterns (cases or solution cases) that suit the purpose or situation, and does not describe a method of selecting these with ease or necessary accuracy and achieving the purpose. Therefore, with respect to complicated problems such as diagram creation, when trying to raise the automatic level, the program becomes extremely complicated and the processing time becomes enormous, and it becomes impossible to obtain a practical solution. Therefore, since the search range of the system is limited so that a practical solution can be obtained, the number of input items becomes very large and it takes time and labor to create. An object of the present invention is to store not only a successful case and its generation pattern, but also a data block in which conditions and keys for application are added by patterning it in a knowledge base, and use this to narrow the search range and reduce the search range. It is to enable highly sophisticated automatic processing even for complicated problems even if there are few input items from. On the other hand, in Japanese Unexamined Patent Publication No. 4-78953, an attempt is made to achieve a high degree of automation by dividing a complicated problem into partial problems, but the present invention solves the complicated problem by utilizing past successful cases. It's an attempt, and the solution is completely different. The method of the previous application is also effective for creating a new timetable (white paper timetable) such as a new route when creating a timetable, but the present invention that facilitates the application of a suitable timetable in the past becomes effective when a normal timetable is revised. There are many. However, the present invention is considered to be more effective when combined with partial problem division.

[0005]

The above object can be achieved by the following. First, in addition to (1) case solution patterns such as solution cases and cases of problems or subproblems, (2) application conditions of these solution cases and cases, characteristic patterns that are key to application, and conforming purpose (adaptation purpose) A patternization case consisting of case application patterns such as the above is registered as a patternization case base. Next, know-how for finding patterning cases that match the purpose and situation from the patterning case base is accumulated in a computer as a knowledge base, and appropriate patterning cases are searched for according to various purposes and situations using these. Provide a mechanism.

[0006]

[Operation] A successful case is registered in the patterning case base as a patterning case, and the search range is narrowed by using the knowledge base for appropriately extracting the patterning case according to the purpose and the situation, thereby inputting from the user. Even if there are few items, advanced automatic processing is possible.

[0007]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention applied to train schedule making is shown in FIGS. 1 to 25.

FIG. 1 is an overall configuration diagram of the present invention. Since it is abstract, a concrete example of the diagram making system has been described in order to facilitate understanding, but it is not limited to the diagram making. Also, an example in which a timetable is divided into partial timetables to create a timetable has been described. However, considering the case where the number of divisions is one, the present invention is applied to the whole train timetable creation without dividing, that is, not a partial timetable. It is self-evident that it is also an example. In FIG. 1, a partial diagram case knowledge table 7 which is a specific example of the patternization case base 7 is a travel knowledge 40 that stores travel patterns of trains, such as traveling time between stations, as shown in FIG.
Train type knowledge 50 that stores the train type and traveling pattern name such as limited express and normal, display time width area knowledge (pattern cycle knowledge) 60 that stores the display time width (cycle) of the partial timetable, display time width area knowledge name, Partial diamond train information knowledge 80 consisting of the number of trains running in the partial diamond and the detailed data knowledge of the train, the partial diamond train information name, and the partial diamond showing a pattern example of the partial diamond pattern with good compatibility when connecting the partial diamonds Partial diamond case solution pattern knowledge (that is, case solution pattern) 906 that stores the connection information knowledge 85 (which has been described so far in FIG. 2) and partial diamond case application pattern knowledge (that is, case application) that stores arrival times of each train. Pattern) 90
7 is a database having elements of partial diamond patterning case knowledge 90 which is a specific example of the patterning case knowledge 90 of 7. The partial diamond case knowledge base 8 is a partial diamond case knowledge table (that is, a patternized case base) 7.
From the strategy for searching the partial diamond patterning case (that is, applicable patterning case) 90 from among the frames, the frame that describes the purpose that is the achievement goal of the strategy and its subordinate purpose, and the rule for searching the patternization case. This is a specific example of the knowledge base 8 for patterning case extraction. The above knowledge is stored in the storage device 5. Details of these knowledges will be described later.

A train timetable (generally a target object (including various schedules, plans, specifications)) using these patterning case base 7 and patterning case retrieval knowledge base 8
Is created as follows. First, based on the knowledge stored in the partial diagram case knowledge table (generally, patterning case base) 7 shown in FIG. 1 and FIG.
The case retrieval inference engine 1602 selects the (partial diamond) patterning case 90 selected according to the purpose or strategy and registers it in the execution partial diamond pattern table 20 (that is, the applicable patterning case is taken out). next,
The case expansion unit 1604 automatically creates a partial diamond for each time zone based on the registered partial diamond pattern case (extracted pattern case) 20 (that is, (partial diamond) automatically develops a case using the case solution pattern knowledge 907). Alternatively, the partial diamond cases are simply taken out), and the partial diamonds are connected (that is, individual partial cases are combined). The partial diamond creation result, the partial diamond connection result, etc. (that is, the automatically expanded case) are displayed on the display device 2 at any time, and the user
If correction is required, the keyboard 3 and mouse 4 can be used to make a correction request. According to this request, the central processing unit 1 corrects the timetable (that is, the case), and the final timetable (that is, the case where the user is generally satisfied) is created. This train schedule is stored in the schedule table (developed pattern case) 6.

A specific example of a diamond or a partial diamond will be described below. However, similar to the above, except for these words (for example, "partial diamond" enclosed in parentheses), it becomes a general embodiment, and the following follows the above. .

As shown in FIGS. 16 to 19, when the central processing unit 1 selects a patterning case base 7 according to an objective strategy, here, a (partial diamond) patterning case is selected from the partial diamond case knowledge table 7. , (Partial diamond) case extraction Using the case extraction purpose strategy knowledge (data) 81 (FIG. 16) in the knowledge base 8, a (partial diamond) pattern case 90 corresponding to the purpose is selected. In the (partial diamond) pattern case 90, the system develops or simply takes out the (partial diamond) case solving pattern 906 to create a partial diamond. After the solution is generated, that is, the partial diamond is created, the user performs a correction work using the keyboard 3 and the mouse 4 as needed to create a satisfactory solution (partial diamond in this case). Correction registration unit 1606 of central processing unit 1
Is an obtained solution, that is, a case, in this case (partial timetable) The case application pattern knowledge is automatically created for the specified items such as train arrival time from the created partial timetable, and otherwise interactively. And (partial diamond) example solution pattern are combined to create a new (partial diamond) pattern case,
As a new patterning case, it is registered in the patterning case base 7, that is, the partial diamond case knowledge table 7.

FIG. 2 shows an execution part diamond pattern table 2.
Indicates the content of 0. The time zone number storage area 201 stores the number of time zones (= n) dividing one day. Area 20
2 stores the knowledge of the partial diagram patterning case to be executed for each time zone. The number of partial diamond pattern case storage area 203 stores the number of partial diamond patterns to be executed for each time period. Partial diamond pattern case name storage area 20
In 4, the names of partial diamond patterning case knowledge are stored for the number of execution patterns (= m).

FIG. 3 is a (partial diamond) patterning case knowledge 9
An example of the configuration diagram of 0 is shown. The (partial diamond) patterning case knowledge 90 includes the (partial diamond) case solution pattern knowledge 906,
It is composed of two pieces of partial diamond case example application pattern knowledge 907. Similarly, the (partial diamond) case solution pattern knowledge 906 is specifically composed of the partial diamond train information knowledge 80 and the partial diamond connection information knowledge 85. The partial timetable train information knowledge 80 is the knowledge area 60 between display width areas (pattern cycle knowledge).
And train type knowledge 50. The train type knowledge 50 is composed of traveling knowledge 40 and the like. Like this (partial diamond)
The case solution pattern knowledge is patternized information that becomes a solution (here, a partial diamond) or a solution (here, a partial diamond to be obtained) by expanding this and modifying it if necessary. On the other hand, (partial diamond) case application pattern knowledge 9
Reference numeral 07 is characteristic data that includes these input data, but is a key for mainly searching for (applying) cases of output data obtained after creating a case (partial diagram in this example) and applying it. In the present system, the knowledge for applying this case is distinguished from the solution and the case solution pattern knowledge 906 that is the pattern of the solution,
As a case application pattern 907, a suitable case is efficiently searched by referring to this.

FIG. 4 shows the contents of the traveling knowledge 40 for storing the traveling patterns of the train in the partial diamond case knowledge table 7, that is, the patterning case knowledge 7, 401 is an area for storing the number of traveling patterns, and 402 is This is an area that stores travel patterns, and the travel patterns are equal to the number of travel patterns (= R).
Remember. The travel knowledge includes the travel pattern name (403), the station name of each station (404), and the running time between stations of each train type (40
5) is data indicating the stop time (406) at the station.

FIG. 5 shows the contents of the train type knowledge 50 that stores the train type and traveling pattern name in the partial diamond case knowledge table 7, that is, the patterning case knowledge 7.
Is an area for storing the number of train types, 502 is an area for storing the train types, and the number of train types is equal to the number of train types (= K).
Remember. Reference numeral 503 is a train type name, and 504 is a traveling pattern name of this train type, which is a traveling pattern name (403) defined in FIG. Therefore, for example, express A is express A
It means traveling in a traveling pattern. Here, the train type refers to the type of train such as limited express, express, and ordinary.

FIG. 6 shows the contents of the display time width area knowledge 60 for storing the display time width area of the partial timetable and the continuation time interval among the knowledge stored in the partial timetable case knowledge table 7, and 601 indicates the display time width. Area that stores the number of domain knowledge,
An area 602 stores display time width area knowledge,
The display time width area is stored for the number of knowledge areas. Reference numeral 602 includes a display time width area knowledge name, a display time width area, and a continuation time interval. Here, the display time width area refers to a display time width area which is a pattern cycle (time) of a partial diamond (703) which is a pattern diamond within a specific time as shown in FIG. 7 (701).
By the way, in the partial diamond, as shown in FIG. 7, when the diamond streak is out of the display time width area and cut at the right end, the remaining streak is displayed from the left end and reaches the destination station. In addition, the continuation interval (702) refers to a driving head that must be kept as a minimum when the train departs continuously.

FIG. 8 shows the content of the partial diamond train information knowledge 80 for storing train information in the patterning case knowledge 7 stored in the partial diamond case knowledge table 7, and 801 stores the number of partial diamond train information. Area, 802 is an area for storing partial train information, and is equivalent to the number of train information (=
P) Remember. 803 is an area for storing a partial diamond train information name, 804 is an area for storing a display time width area knowledge name defined in FIG. 6, 805 is an area for storing the number of trains (= T) traveling in the partial diamond, Reference numeral 806 is an area for storing train data traveling in the partial timetable, and includes an initial train time, a train type name, a starting station, a destination station, and the like. The train type name used here is defined in FIG. As for the first departure time, it is possible to specify the first departure sequence instead.

FIG. 9 shows the contents of the partial diamond case solution pattern knowledge and the partial diamond case application pattern knowledge in the partial diamond pattern case knowledge 90 of the pattern case knowledge 7 stored in the partial diamond case knowledge table 7. . Reference numeral 901 denotes an area for storing the number of registered cases, and reference numeral 902 denotes an area for storing (partial diamond) pattern cases, and stores (partial diamond) pattern cases for the number of registrations (= B). 903
Stores the pattern name of the (partial diamond) pattern case.
Reference numeral 904 stores the date of creation of this patternized case, and reference numeral 905 stores the number of times the pattern case has been extracted as a similar case. 906 stores a (partial diamond) case solution pattern (knowledge). Reference numeral 907 stores a (partial diamond) case application pattern (knowledge). In 906, the names of the partial timetable train information defined in FIG. 8, and the preceding connection partner candidate and the latter connection partner candidate are shown to be compatible with each other when the partial diamonds are connected to the partial diamonds in the preceding or following time zones. Store a set of good partial diamond pattern case names. In 907, the matching purpose is a queue for storing the matching purpose of the present patterning case, that is, the purpose which is considered to be easily achieved by using the present patterning case. The row type pattern indicates the application condition of this patterning case. That is, it shows the number of trains for each train type in the partial timetables generated by this patterning example. Train hours and below are all features extracted from the created partial diamonds. Train hour is the total operating time of the train. Arrival time is the time required from the starting station to the destination station. The maximum parking time is the longest time the train has stopped at the station. The maximum number of shunts refers to the maximum number of times other trains pass while the train is shunting at the station.

FIG. 10 shows traveling knowledge 40 and train type knowledge 5.
The simplest examples of 0, display time width area knowledge 60, and partial diamond train information knowledge 80 are shown.

FIG. 11 is a partial diagram created by the knowledge (data) of FIG. Reference numeral 1100 indicates a display area, 1101 indicates a time axis, and 1103 indicates a station name and a station distance axis. Reference numeral 1102 indicates the created train streak (operating locus).

FIG. 12 shows an overall flow for creating a day's timetable. According to the present invention, in determining the vehicle allocation pattern of the block 1201, a suitable partial diamond pattern ((partial diamond) patterning case, (partial diamond) case solving pattern)
Can be taken out and the partial timetable can be efficiently created.

FIG. 13 shows a block 120 in FIG.
It is a detailed processing flow of the 1-vehicle allocation pattern determination part.

FIG. 14 is an example of a screen on which a vehicle allocation pattern is set. Here, 1400 shows a list of names of patterning cases registered as patterning cases in the (partial diamond) patterning case knowledge 90, and the pattern name, the creation date, and the number of times of use so far are displayed. B100 of 1401 is a different color, indicating that it is selected. A mouse can be used to scroll up and down the display screen of 1400 with 1402 and 1403, and unnecessary patterns can be selected by selecting a patternization example and then selecting the delete menu 1404 (a different color is selected like B100). You can delete the case. When the patterning case is selected, expanded, modified, and registered, the number of times 905 of use of the selected patterning case is incremented. The initial value of the number of times of use 905, that is, the value when the patternization example is generated is 1
Is. In order to prevent the number of registered patterning cases from increasing and the storage capacity of the storage device becoming full, the number of patterning cases is set to 90.
If 1 exceeds a preset value, the number of times of use is divided by the difference between the current date and the creation date 904 indicated by the computer timer, and the one with a low specification frequency is displayed in a different color (a color different from 1401). , Warn the user to delete. When the user instructs deletion of unnecessary patterning cases according to this warning, the system fills the area where the patterning cases for which deletion is instructed is stored and, at the same time, the patterning case number
Decrement 01 by 1. Reference numeral 1410 is a screen for displaying a vehicle allocation pattern. 1420 is a name display area for each time zone, and 1421 to 1425 are examples of names for each time zone.
1430 is a start time display area for each time zone.
31 to 1435 are examples of the start times of the respective time zones. 14
40 is an end time display area of each time zone, and 1441
To 1445 are examples of end times of the respective time zones. 1450
Is the name display area of the (partial diamond) patterning case, and 1451 to 1455 are the (partial diamond) of each time zone.
Indicates the name of the patternization case. FIG. 14 shows a screen when the correction / change command 1461 is picked with the mouse in order to change the vehicle allocation pattern, and other (partial diamond) pattern case examples 90 are selected from the menu 1471 such as pick and 1451. By picking the pattern name of
Selection of case-based reasoning can be done by picking in menu 1472.

FIG. 15 shows a flow chart of case-based reasoning for retrieving the preferred (partial diamond) patterning case 90.

FIG. 16 shows the flow of processing in case-based reasoning for retrieving a preferred (partial diamond) patterning case 90. In response to the user request (1601), the data extraction inference engine (1602) refers to the partial diamond case extraction knowledge base 8, that is, the patternized case extraction knowledge base 8 to generate the partial diamond case knowledge table 7 or the patternized case base 7. A desired patterning case (1603) is extracted from the inside (block 1502). The automatic creation processing unit (1604) creates a partial diamond by using the partial diamond case solution pattern of the extracted pattern cases (block 1505). The correction registration unit (1606)
At, the user's request for modification (1605) and the request for registration of the modified case (partial diamond) (1607) in the patterned case base 7 are processed (block 150).
4, 1507, 1509). In order to speed up the processing, the data fetching inference engine (1602) can be converted to firmware or hardware using a conventional database search dedicated device. FIG. 17A is a diagram showing an objective itself for extracting a suitable patterning case according to the objective and a strategy method which is a means for achieving the objective. Each objective and strategy is a frame (data and program package , "Marvin Minsky, 'A Framew
ork for Representing Knowledge ', in The Psychology
og Computer Vision, edited by Patrick Henry Winst
on, McGraw-Hill Book Company, New York 1975 ”). 17B and 17C are specific examples of these, and the frame shown in FIG. 17B represents the subordinate purpose of FIG. 17A, that is, high speed and high density diamond creation (1702). FIG. 17C shows the specific contents of the strategy for emphasizing the number of shelters (1703) and the strategy for designating shelters (1704), which are one of the strategies of FIG. The candidate case slots in each frame indicate the storage of candidate (patterned) case (s) applicable to achieving this goal. The name of this (partial diamond) patterning case knowledge is added to this storage area by the system when the system is started up or when the contents of the (partial diamond) patterning case knowledge and the contents of its matching purpose attribute are changed / added. To be done. Of course, the user can interactively set / change. The classes in FIGS. 17B and 17C indicate whether the frame belongs to the target frame or the strategy frame. The name slot of each frame indicates the name of each frame. The candidate case slot indicates the name of the patterning case that is a candidate for achieving the purpose. The target slot in the strategy frame indicates the name of the higher-level target frame.

FIG. 18 shows the detailed flow of block 1501. The contents are as shown in the figure.

FIG. 19 is a diagram illustrating the case retrieval by the case-based reasoning in FIG. 16 using a concrete example.

FIG. 20 is an example of a screen displayed when the case-based reasoning 1472 is selected in FIG.
From this screen, the user decides the strategy to achieve the purpose,
If necessary, the display data is changed, and the system takes out a suitable patterning case according to the input result.

FIG. 21 shows an example of a screen displaying detailed data of the taken out (partial diamond) patterning case 90.

FIG. 22 shows the patterning case knowledge 90 of FIG.
On the screen that displays the partial diagram automatically generated using
The diamond can be modified on this screen.

FIG. 23 is a detailed display screen of the patterning case 90 displayed after the partial diamond is created and registered.

FIG. 24 is an example of a vehicle allocation pattern display screen after creating a suitable patterning case.

FIG. 25 shows the vehicle allocation pattern of FIG.
It is the figure which displayed each partial diamond within the designated time, and displayed the diamond for one day. 2510 is a display screen, 25
Reference numeral 11 indicates a time axis, and 2512 indicates a station name and a distance axis.

The procedure for creating a train schedule according to the present invention will be described below with reference to FIGS. 12, 13, 15 and 18, and FIGS. 12 to 25.

First, according to the flow chart of FIG.
The outline of the diagram creation will be described. At block 1201, a vehicle allocation pattern for each time zone (a partial diamond pattern case used when creating a partial timetable for each time zone is referred to. The name of this pattern case is registered in the execution partial diamond pattern table 20). To do. Details of the vehicle allocation pattern determination method will be described with reference to the flowchart of FIG. Next, block 120
In step 2, a partial timetable for each time zone is created from the vehicle allocation pattern for each time zone set in block 1201. Block 12
After repeating 02 for the entire time period (1203), block 1
At 204, partial diamonds are connected to create a day's worth of diamonds. If the created result is satisfactory (1205), this is registered as a day's timetable (1206), and if not satisfied, the vehicle allocation pattern is determined and the trial is repeated again.

Next, according to the flow chart of FIG.
A procedure for determining a vehicle allocation pattern (1201) will be described. First, in block 1301, the default value of the vehicle allocation pattern is displayed. The default value may be a previously set value or a value prepared by the system in advance. An example of a screen displaying this default value is 1410 in FIG. When a change is required (1302), in FIG.
Pick the "Modify, Change" 1461 menu. When the "partial timetable creation" 1462 or the "end" 1463 menu is selected, no change is necessary. If you need to change,
The patternization case is changed (blocks 1303 and 130).
4) The name of the changed patterning case is registered (1305) in the executed partial diamond pattern table 20.

As for the change setting, for example, in FIG. 14, when it is desired to change the dispatch pattern of the morning rush, the mouse 4
After picking the "correction / change" menu 1461 with, the area 1452 displaying the morning rush partial diamond execution pattern is picked. Then, this area 1452 becomes a different color, and B100 (1401) of the registered partial diamond table 1400 (a list of pattern case examples registered in the partial diamond pattern case knowledge 90 as a case)
Is a different color. Further, the change option 1470 is displayed, and one of the two options is selected with the mouse 4.
When the partial diamond pattern take-out (1471) is picked, this becomes a different color, and the desired partial diamond pattern name is picked from the list 1400. Then, the picked pattern name is displayed in a different color, and the selected pattern name is further displayed in 1452 to indicate that the selected vehicle pattern has been selected.

An example of picking the case-based reasoning 1472 will be described next with reference to FIGS. 15 to 23. FIG. 15 shows a flowchart of the case-based reasoning method, and FIG. 16 shows a processing flow. In block 1501 of FIG. 15, a suitable partial diamond pattern case is extracted from the partial diamond pattern knowledge table by case-based reasoning. Hereinafter, a method of extracting a suitable partial diamond pattern case will be described.

In FIG. 16, the case retrieval inference engine (1602) retrieves a suitable patterning case (1603) according to the purpose according to a user input (1601).
The diagram showing the objectives and the strategy for achieving the objectives in FIG. 17 is a frame of knowledge representation of the objectives for extracting cases and the strategy method for achieving the objectives. The selection of suitable patterning cases is performed according to this purpose and strategy. Figure 1
As an example shown in 7A, high-speed high-density pattern utilization strategy (1
As one of the purposes of 701), high-speed and high-density diamond creation (1
702). In order to achieve this purpose, there are a strategy for emphasizing the number of shelters (1703) and a strategy for designating shelters (1704). The strategy for emphasizing the number of evacuation lines is a strategy for extracting cases in which the maximum number of evacuation lines in the partial timetable is less than or equal to the specified number of input lines. The evacuation station designation strategy is, for example, a strategy for an express train to be overtaken by an express train at a designated station. FIG. 17B is an example in which the purpose of creating a high-speed and high-density timetable is represented by a frame.
FIG. 17C is an example of a strategy frame, which shows a strategy for emphasizing the number of shelters and a strategy for designating shelters, which are lower strategies. FIG. 18 shows a flow for extracting a preferred example according to the purpose and strategy of FIG.
Further, FIG. 19 shows a flow of processing using a concrete example. First, the user selects a strategy for achieving the purpose in order to retrieve a patterning case that suits the purpose of use (block 180).
1). Then, the case extraction reasoning engine displays the case application pattern of the candidate case registered in the candidate case slot of the strategy frame and the related information on the screen (1802).
In response to this, the user can request an alternative plan and modify the display data (1802). In addition, the strategy can be changed (1803) if necessary. When the user approves the case application pattern and related information displayed on the screen,
The case retrieval inference engine performs knowledge matching operation to retrieve a suitable case (1804). This flow will be described with reference to FIG. 19 using a specific example. It is assumed that the user has selected a strategy for emphasizing the number of shelters to create a high-speed and high-density diamond (purpose). As the case application pattern, the maximum number of three shelters is corrected and input (1601). The case retrieval reasoning engine uses a strategy (17
03) is adopted, and based on the knowledge described in the frame, the user input value and partial diamond case example retrieval are matched with the rule in the knowledge base and the registered case in the partial diagram case knowledge table to extract a suitable case. Here, since it is a high-speed and high-density diamond, the arrival time of the express train is less than 2 hours according to Rule 1. The maximum number of evacuation lines must be 3 or less based on the user input value. The case retrieval inference engine (1602) selects pattern 1 and pattern 2 as candidate cases satisfying these conditions. From the two candidate cases, pattern 1 is selected as a partial diamond patterning case according to rule 2. This is the end of block 1501.

Example of partial diamond patterning (block 180
4, and block 1501) After the completion of taking out, the user judges whether or not to adopt this (1502). If not, the operation of 1501 is repeated and the inference is performed again. When the adopted pattern case is modified (1503), the partial diamond case solution pattern can be modified and can be modified (1504).

Next, blocks 1501 to 15
An example of the operation method according to the flow up to 04 will be described. First, in FIG. 14, case-based reasoning 1
When 472 is picked, the case retrieval input screen of FIG. 20 is displayed. The system shows a strategy applicable to achieve a given purpose (high-density diamond creation (2010)).
The target strategy diagram of is taken out and displayed (202)
0). When there are a plurality of strategies, the strategy selected by the system (for example, a strategy with a high priority) is displayed in a different color, but the user may specify a desired strategy with the mouse 4.
At this time, the user may select the purpose 2010, and in this case, the case stored at the head of the candidate case queue of the target frame in FIG. 17B is displayed in 2030. Alternative 20
If you pick 31, the next case in the candidate case queue will be 20
It is displayed on 30. The strategy displayed in FIG. 20 shows the two subordinate strategies shown in FIG. The case extraction inference engine (1602) selects the patternized case specified as the value of the candidate case slot of the selected strategy as a suitable case pattern, and the user data input screen 2030
The pattern selection case selection data (case application pattern and related information) is displayed. When the user selects the approval, that is, the end (2032), the above patterning case is selected. If the user does not approve and picks the alternative (2031), the next element of the slot (queue) of the strategy is selected, and the selection data is displayed in 2030. When the user corrects the display data of 2030, the pattern extraction case satisfying this is extracted from the case extraction inference engine (16
02) infers and takes out (this also applies to the example of FIG. 19). This example will be described according to the user input example used in FIG. First, the strategy for emphasizing the number of shelters in 2021 is selected (it becomes a different color). Then, the user data input screen 20
30 is displayed. As a high-density diamond, a 3-7-3 pattern (3 limited express trains, 7 express trains, 3 normal trains) was shown as the default value. The maximum number of shelters was entered as 3 (correction input). When the input end (2032) is picked with the mouse, the data extraction inference engine starts the inference. Since the inference procedure has already been described (FIG. 19), its explanation is omitted. A detailed result example after the inference is shown in an area 2110 in FIG. The display content is the data content of the partial diamond patterning case, 2111 is the partial diamond patterning case name, 211
2 to 2114 are partial diamond example solution pattern knowledge, 211
Reference numeral 5 is knowledge of partial diamond case example application patterns. Detailed data can be confirmed by picking items 2112 to 2115. Partial timetable To confirm train information knowledge,
Pick 2112. Then, train information data of 2120 is displayed, and further picking 2121 displays train 1 data 2130. At this time, the data can be corrected, and the necessary data can be corrected by picking the partial diamond pattern correction (2116). Further, by picking the partial diamond case application pattern knowledge 2115, the partial diamond case application pattern (2140) created by the patternized case extracted by inference is displayed. Since this data is output data, it cannot be modified.
If the data retrieved by this inference is not desired and is not adopted, by picking retry (2117), it is possible to return to the 20th screen and repeat the inference.

Returning to the flow of FIG. 15 again, description will be made. After extracting a suitable patterning case, the automatic creation processing unit (1604) automatically creates a partial diamond using the partial diamond case solution pattern of the extracted partial diamond pattern cases as input data. In other words, the departure and arrival times of all train stations of each train line are calculated from the start time pattern of the partial diamond example solution pattern knowledge, the starting station, the destination station, and the traveling pattern knowledge, and set in the line information table and collide with other lines. If there is a shortage of time to continue or to continue, the departure and arrival times will be automatically delayed and adjusted to satisfy these, and a partial diamond will be created. Further, if it is necessary to correct the partial timetable (1506), correction processing is performed (1507). And
When the partial diamond is registered as a case in block 1508, the partial diamond case solution pattern 906 of FIG.
Is created by the system or is input by the user, and the value of each attribute of the partial diamond case application pattern 907 is obtained from the partial diamond information table to create a partial diamond pattern case, which is used as the partial diamond pattern case knowledge 90. Register (1509). In this registration, the number of partial diamond patterning cases (901) in FIG. 9 is counted up, and each data of the partial diamond patterning cases is obtained at the position corresponding to the number of cases having the item of partial diamond patterning case knowledge 90. Fill in. It is necessary to save the start and end addresses of each patterning case and case solution pattern as management data of the patterning case, but these are natural and therefore omitted in this embodiment.

The operation method according to the flow from block 1505 to block 1509 in FIG. 15 will be described above with reference to FIGS. 21 and 22. In FIG. 21,
After the partial diamond example solution pattern knowledge is displayed and corrected, when the partial diamond case creation (2151) is picked, the automatic creation processing unit (1604) automatically creates a partial diamond based on the partial diamond case solution pattern. That is, the case solution pattern is developed into cases (1505). The created partial timetable is displayed on the display device 2, an example of which is shown in FIG. When the created result is satisfactory but a small amount of correction is required, the correction 2202 is picked and the correction work is performed (1507).

Next, when picking up the vehicle allocation pattern registration 2201, if the content of the case solution pattern extracted by inference in FIG. 21 is changed or the partial timetable is modified in FIG. 22, it becomes a new case. Is displayed, and the new pattern name (D100 (22
11)) is input. When end (2212) is picked, the display screen becomes as shown in FIG. 23, and the patterning case knowledge is displayed in 2310. 2311 has a new pattern name D10
0 is displayed. This name can be entered by the user and checked by the system for uniqueness and can be modified by the user. If the same partial train information as the new partial train information is registered in the partial train information knowledge 80, the name is displayed in 2312. If the partial diamond train information of the new pattern diamond (partial diamond) is not registered in the partial diamond train information knowledge 80, that is,
When the modification of 1504 or 1507 changes the partial timetable train information, it is necessary to input new partial timetable train information. 2312 is a different color or a different pattern (e
x. Against the background of the (mesh pattern), the user inputs a new partial train schedule train information name 803 (information indicating the relationship between all the trains forming the partial schedule and trains). As a result, new partial diamond train information is registered in the partial diamond train information knowledge 80. It is also possible to display the content of the partial diamond patterning example adopted in 1502 and the partial diamond train information in another window and input this name from there. The user inputs the pre-connection candidate and the post-connection candidate. The system proposes what is currently connected and the user adds it.
It can be modified. By the above operation, the partial diagram example solution pattern 906 is created. Each attribute of the partial diamond case example application pattern 907 is expanded pattern case 6
Each attribute value is calculated and obtained from the correction result (when there is a correction) in 1507 of FIG. 15 (partial diamond table). This is 23 by picking 2315
It is displayed and confirmed on 20. In this way, patterning cases can be created and registered (1509). The inference ends and the screen returns to the vehicle allocation pattern screen (FIG. 24). FIG. 24 shows that a new pattern D100 has been selected for the morning dispatch pattern (block 1452). Further, D001 is a different color in the registered pattern list (1405).

On the other hand, when the patterning case B159 obtained by inference in FIG. 21 is used without modification,
After the picking up of the vehicle allocation pattern of FIG. 22, the screen moves to FIG. 24 and B159 is displayed at 1452.

In this way, the dispatch pattern is determined (120
1) Proceed to create a partial timetable for the relevant time period. this is,
24 or 14 is a partial diamond creation 1462
Created by picking. However, in this case, unlike FIG. 22, it is displayed on the diagram of one day. This operation is performed for each time period to create a day's timetable. This one-day timetable is shown at 2510 in FIG. With the above work, one day's timetable creation is completed.

[0047]

According to the present invention, (1) it is possible to quickly create a partial diamond based on pattern knowledge. (2) It is possible to select the desired patterning cases with high efficiency based on the case application patterns and objective strategy knowledge, and it is easy to make a partial solution or a comprehensive solution that integrates and adjusts them with a small amount of modification, that is, It is possible to create partial diamonds and whole diamonds. (3) A pattern can be selected according to the purpose. In addition, it is possible to select (partial) patterns that flexibly respond to various requests from users by strategy. (4) Since a newly created case (partial diagram) can be compressed and efficiently stored as a patternized case, a large number of various cases can be easily registered.

The effect is as follows.

[Brief description of drawings]

FIG. 1 is an overall configuration of a train schedule creation system according to the present invention.

FIG. 2 shows an example of contents of an execution part diamond pattern table according to the present invention.

FIG. 3 is an example of a configuration diagram of a partial diamond patterning example according to the present invention.

FIG. 4 is an example of contents of traveling pattern knowledge.

[Fig. 5] Content example of train type knowledge.

FIG. 6 is an example of contents of display time width area knowledge of a partial diamond.

FIG. 7 is an example of a partial diamond.

FIG. 8 is an example of contents of partial timetable train information knowledge.

FIG. 9 is a content example of a partial diamond patterning example.

FIG. 10 is a specific example of partial schedule train information knowledge.

FIG. 11 is an example of a partial diamond created based on the data of FIG.

FIG. 12 is a flowchart showing a schematic procedure of diamond creation.

FIG. 13 is a flowchart showing a procedure for determining a vehicle allocation pattern.

FIG. 14 is a dispatch pattern input screen.

FIG. 15 is a flow chart showing a preferred partial diamond retrieval by case-based reasoning according to the present invention.

FIG. 16 is a diagram showing a processing flow of case-based reasoning according to the present invention.

FIG. 17 is a diagram illustrating a configuration diagram of a purpose-specific case retrieval strategy and a strategy used in a frame.

FIG. 18 is a case-based reasoning flow.

FIG. 19 shows an example of patterning case extraction by case-based inference using a concrete example.

FIG. 20 is a data input screen for inferring case retrieval.

FIG. 21 is an example of a case-based reasoning detailed result screen.

FIG. 22 is a display example of a partial diamond.

FIG. 23 is a patternization example registration screen.

FIG. 24 is a vehicle allocation pattern setting screen after inference is completed.

FIG. 25 is a diagram display screen for one day.

[Explanation of symbols]

1 ... Central processing unit, 7 ... Patterning case base (partial diamond case knowledge table), 8 ... Patterning case retrieval knowledge base, (Partial diamond case retrieval knowledge base), 20
… Patterned case example (execution partial diamond pattern table) taken out, 90… (partial diamond) pattern case knowledge, 9
06 ... Case solution pattern, 907 ... Case application pattern.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Kyoko Takahashi 12 at 890 Kashimada, Saiwai-ku, Kawasaki-shi, Kanagawa Hitachi Information Systems Development Division, Ltd. (72) Haruhisa Niki 890 Kashimada, Sai-ku, Kawasaki-shi, Kanagawa No. 12 Inside Hitachi Information Systems Development Division

Claims (15)

[Claims] 1. A computer system for solving a problem,
For example solution patterns such as solution cases in which the solution creation method and the key elements of the solution creation are patterned, or cases that are not patterned as it is, the application conditions of this case, the characteristic patterns that are the key of application, and this case A case called the patterning case base that saves the patterning cases to which case application patterns such as the conforming purpose that are the matching purposes are stored, the step of storing the patterning cases in this, and the patterning cases for solving the problem A problem solving method that includes the steps of extracting from the patterning case base using the case application pattern as a key, and the step of promptly proposing a high-quality solution using the extracted patterning case. 2. When the above problem is divided into a plurality of subproblems, each subproblem is solved, and then individual solutions are integrated to solve the problem, the solution itself of the subproblem, a solution creating method, or a solution Patterning cases in which the application conditions of this case, the characteristic pattern that is the key of application, and the case application pattern such as the conformance purpose are added to the solution case in which the key elements of creation are patterned or the case that is not patterned To save the pattern to the pattern case base, and when solving the partial problem, apply the pattern case to solve the problem as a case application. Use the pattern as a key to extract from the pattern case base, and use the retrieved pattern case. The problem solving method according to claim 1, comprising the step of rapidly proposing a high-quality solution. 3. A step of saving a pattern-based case, which is a patternized case suitable for the purpose, in the patternized case base for the purpose of solving the above-mentioned problems or sub-problems, and data or knowledge for extracting the case 3. A method comprising the steps of setting a strategy for achieving an objective and accumulating it in a computer system, and using this to select and propose a patterning case suitable for the objective. How to solve the problem described in the section. 4. A step of accumulating in a computer a knowledge base of know-how for finding a patterning case or a case suitable for a purpose or a situation from the patterning case base, and using the know-how according to various purposes or situations. 4. A problem solving method according to any one of claims 1 to 3, which has a part for extracting an appropriate patterning case and a part for expanding to a case, and enables the system to search for and propose a suitable patterning case. 5. The method according to claim 1, further comprising the step of accumulating the alternative strategy or patterning case in a computer system and the step of proposing the alternative patterning case or alternative plan automatically or at the user's request. The problem solving method according to any one of items 1 to 4. 6. When the above system searches and proposes a patternized case or a case, a step of extracting a case application pattern, which is a key of the search, or a part thereof or related information thereof, as case extraction information. The method has a step of simultaneously displaying and a step of suggesting a case or a patternized case corresponding to the corrected case extraction information when the user corrects this, by the system using the knowledge base, purpose and strategy. The problem solving method according to any one of items 1 to 5. 7. The method according to any one of claims 1 to 4, further comprising a step of registering a correction result of the patterning case as a new patterning case.
How to solve the problem described in the section. 8. A step of counting the frequency of use of patterning cases and the number of registered patterning cases to prevent the storage device from becoming full due to the accumulation of patterning cases, and the number of patterning cases is a constant value. 5. If the number exceeds the specified number, the less frequently used one is displayed, and the unnecessary patterning case is deleted by the user's approval.
How to solve the problem described in the section. 9. The above problem is an example of a combination of basic data and procedures for creating partial diamonds and partial diamonds, which are constituent elements of the above diamonds, in a computer system for creating timetables of trains, buses, aircrafts, etc. Claims consisting of a pattern and a characteristic pattern that is a key to the application of partial diamonds as a case application pattern, and storing these in a patternized case base, and a step of rapidly proposing a high-quality diamond using The problem solving method described in 1 to 8. 10. The problem solving method according to any one of claims 1 to 6, wherein when the partial diamonds are developed, combined and adjusted, the transportation diamonds for one day and one month are provided. 11. The patterning example in the partial diamond includes at least one of a traveling pattern, a train type, a train departure time, a train type creation order, a train priority, and a driving head, in addition to the solution itself. The problem solving method according to items 1 to 6. 12. A step of retrieving a diamond to be created, a similar diamond similar to the partial diamond, or a patterning case thereof from a patterning case base when the diamond is created,
The problem solving method according to any one of claims 1 to 6, wherein the target diamond can be created by the step of correcting this. 13. The know-how for finding out the similar diamonds and patterning examples thereof is stored as a knowledge base such as a strategy for achieving purpose of diamond creation, and by utilizing this, a diamond suitable for various purposes and situations and its Steps for automatically proposing patterning cases, steps for proposing alternative partial diamond patterning cases and alternatives, and also displaying case retrieval information at the time of these proposals, and when the user corrects this, corrects it using the knowledge base 7. The problem solving method according to claim 3, further comprising a step of automatically proposing a partial diamond patterning case corresponding to. 14. A step of registering a newly created diamond as a new patterning case in a patterning case base after taking out a similar diamond, and a less frequently used one automatically or by user's approval. 10. The problem solving method according to claim 8, further comprising a step of deleting. 15. In a computer system incorporating the problem solving method according to any one of claims 1 to 13, a firmware or a dedicated device for extracting a patternized case from a patternized case base with a case application pattern as a key is provided for the purpose of speeding up. Problem solving system characterized by