JPH04130955A - Device for managing simulation model - Google Patents

Abstract

PURPOSE:To manage plural simulation models in each graphic structure and to simplify the reuse of the models by collectively managing the attribute and procedure of each function of an element in a graph to express the model of a discrete event system. CONSTITUTION:When a user inputs a command for executing simulation from input devices 204, 205, a control device 201 reads out the data of a relating model from a model data base 103, constructs the model, executes simulation, and displays the simulated result on a display device 203. When necessary, the simulation execution result is stored in an existing data base 104. The model data base 103 can be used by schema-converting data in the existing data base such as a relational data base and production management data or the like can be utilized as the attribute of a model constitutional element. Each model can be expressed by collectively expressing its attribute and procedure and utilizing a class including elements belonging to the model. The classes can be expressed as hierarchical structure and the lower class can succeed the attribute and procedure of the upper class.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention is directed to a discrete event system using a computer.
It relates to IT for managing ration models.

[Conventional technology]

The flow of products in a production system, changes in the status of resources such as machines, and the flow of control signals can be viewed as discrete events that occur over time.

Furthermore, the flow of people and goods in an office can be considered a similar discrete event system.

When evaluating the performance of these systems, conventionally FORT
A compiler language such as RAN or a GPS (general purpose) language developed based on it.
The system was modeled by programming using a simulation language such as Simulation System. When conducting a simini-region, a module for collecting and analyzing the parameters and experimental results necessary for the simulation experiment was embedded in the program code and compiled, and then the simini-region experiment was performed. Therefore, there is a problem that the user must be familiar with programming, which limits the number of users. Furthermore, with this method, it is necessary to recompile the model every time a simulation experiment is performed, making it difficult to reuse the model.

In response to this, a method has recently been proposed in which a front-end processor for the above-mentioned simulation language is developed and used to perform modeling and experiments to evaluate system performance. This method performs modeling by describing a network diagram using commands in a simulation language as icons, and there is also an example of performing system performance analysis by describing a simulation model interactively on a computer. It's coming out.

[Problem to be solved by the invention]

However, the above method has the following problems.

(1) Separate program code must be written to collect the parameters and results necessary for the simulation experiment. Therefore, it is necessary to be familiar with programming and simulation, and a learning period is required before the general public can use it, so the effectiveness of simulation cannot be fully utilized.

(2) When performing large-scale simulations by combining several models, it is difficult to achieve consistency between the models.

In addition, it takes time to develop a large-scale simulation model.

(3) In order to use simulation results from an existing database, it is necessary to program according to the format of the database to be used.

(4) Models are simply stored as files on a computer, and if multiple models are on one computer, the relationships between the models cannot be grasped.

In order to solve the problems of the prior art described above, the present invention separates a simulation model expressed in a graph structure into model elements and relationships between elements, collectively manages attributes and procedures, and also manages the same The aim is to provide users with an easy-to-use simulation environment by managing multiple simulation models in the system.

[Means to solve the problem]

In order to achieve the above object, the simulation model management device of the present invention includes a means for creating a simulation model expressed in a graph structure, and a means for classifying model elements of the simulation model by function and identifying the attributes and unique characteristics of the elements. A means for collectively managing procedures, a means for managing relationships between model elements, a means for collectively managing attributes and procedures common to the entire model, and exchanging model data with an external database. It is characterized by having an interface means for.

[Effect]

11th! [ is a block diagram showing a general configuration of the present invention. A model to be simulated is input by the model editor 102, and information is passed to the control unit 101. The control unit 101 stores this information in the model database 1.
Store in 03. When a command to perform a simulation experiment is input from the command input unit 105, the model database 103 and, if necessary, the existing database 1
04, a simulation experiment is performed. The staining experiment is controlled by the control section 101, and the experimental results are displayed on the display section 106.

In the present invention, in order to enable reuse of simulation models, we use the concept of classes in object-oriented languages and frame theory to classify elements in a graph by function, and group attributes and procedures into the classes. and have it held. This class can have a hierarchical structure, where lower classes have the properties (attributes and procedures) of higher classes.
is inheritable. For example, smalltalk-g
In object-oriented languages such as Q, properties can be inherited simply by specifying hierarchical relationships. The simulation model is managed in the database that can represent this class.

The network diagram input by the model editor 102 is automatically divided into elements and converted into class representations. Furthermore, when reading graph data from an existing database such as a relational database, this data is converted into a class representation in the control unit 101.

〔Example〕

Embodiments of the present invention will be described in detail below based on the drawings.

FIG. 2 is an overall configuration diagram of a workstation for implementing the present invention. In this figure, 201 is a control device, 202 is a main memory, 203 is a display device such as a bitmap display, 204 is a data input device such as a keyboard, 205 is a pointing input device such as a mouse, and 206 is a magnetic It is an auxiliary storage device such as a disk.

Here, the correspondence with FIG. 1 is shown. The control unit 101 corresponds to a central processing unit 200 consisting of a control device 201 and a main storage device 202. The model editor 102 and the command input unit 105 correspond to the data input device 204 and the pointing input device 205. The databases 103 and 104 correspond to the auxiliary storage device 206, and the display unit 106 corresponds to the display device 203.

FIG. 3 shows the flow of control in this embodiment. The user edits and creates a model using the model editor 102 (see FIG. 1) using the data input device 204 and pointing input device 205 while looking at the display device 203. After each operation is completed, the control device 201 classifies the nodes representing the model components and the arcs representing the relationships between the nodes by type and assigns identification codes to the nodes. Thereafter, the information added to the nodes and arcs is converted into an internal schema that defines the structure for storing data in files, and is stored in the model database 103 (see FIG. 1) provided in the auxiliary storage device 206. Additionally, attributes and procedures common to the entire model, such as a graph structure representing the relationship between nodes and arcs, are stored in the model database 103.

When the user inputs a simulation execution command from the input device 204 or 205, the control device 201 reads related model data from the model database 103, constructs the model, and executes simulation.
The results are displayed on the display device 203. In addition, if necessary, the simulation execution results can be stored in the existing database 104.
Store in.

The model database 103 can be used by schema-converting data in an existing database such as a relational database, and production management data and the like can be used as attributes of model components.

Each model collectively represents attributes and procedures, and is expressed using classes that have elements belonging thereto. Furthermore, classes can have a hierarchical structure, and lower classes can inherit the attributes and procedures of higher classes. The meanings of attributes and procedures will be explained later.

FIG. 4 shows an example of a simulation model of a production line described using the present invention. In this model, the processes in the production line, the workers required to execute this process, resources such as buffers, and parts are represented by nodes 401 to 410, and their interrelationships are represented by arcs 411 to 419. A graphical representation 420 of the simulation model is generated.

That is, 401 is a node representing a part (6 rolls), 402 is a node representing stepped processing in processing machine MC-^, which requires 3 minutes per roll, and 403
is a node representing buffer B, 404 is a node representing worker A, 405 is a node representing a part (frame), 4
06 is a node representing the work of roll assembly that takes 1 minute, 407 is a node representing worker B, and 408 is a part (
409 is a node representing buffer C, 410 is a node representing a transport operation that takes 5 minutes, and 411 to 419 are arcs representing relationships between nodes 401 to 410.

It is assumed that the processing in the production line shown in FIG. 4 has the following flow.

Input 6 rolls and process them one by one into the machine MC-^
Perform stepped processing and place in a buffer.

If there is a roll in buffer B, worker B assembles the roll to the frame using two screws and places it in buffer C. When three frames are stored in buffer C, the transfer takes 5 minutes.

This graph is created by the user selecting an icon indicating a desired node from the menu displayed on the display device 203 using the pointing input device 205, moving it to the desired location, and then positioning and displaying each node. , can be created by connecting each node with an arc using the pointing input device 205.

These flows are expressed by the graph in FIG. 4 and the node attributes and procedures described below.

FIG. 5 shows the relationship between classes in which each element in the graph is classified by function in order to manage the model shown in FIG. 4.

Here, the object class 501 is an extracted class that has attributes and procedures common to all classes, and in this embodiment, as shown in Table 1, it has attributes such as identification code IO and procedures related to it. let Also, if the class contains a graph element, this element is included.

The node class 502 and below are classes that have attributes and procedures of model elements, the arc class 503 is a class that represents relationships between model elements, and the graph class 504
is a class that contains attributes and procedures for the entire model.

Node class 502. arc class 503 and graph class 504 Table 2 shows an example of the structure of - Shown in Table 4.

Note that a node set in a graph class is represented by a set or a sequence whose elements are each node in the graph. The same applies to arc sets.

Also, below the node class 502, there is a component class 5.
05, resource class 506, process class 50
7, and each has attributes and procedures as shown in Tables 5 to 7.

Furthermore, Table 8 is below the resource class 506.

There is a worker class 508 and a buffer class 509 as shown in Table 9.

As an example, model data of node 404 belonging to worker class 508 is shown in FIG.
The model data of node 402 belonging to
! It is shown in I0. As shown in these model data, according to the class concept, lower classes are identified by the identification code ld
It can be seen that they automatically inherit the attributes of higher classes such as ``Kaaraku''.

The class hierarchy shown in FIG. 5 and the elements within the classes are stored for each simulation model in the model database 103 on the auxiliary storage device 206, and each class is managed by the control device 201. At this time, if procedures and attributes for collecting results are prepared for each model element and graph class 504, the results can be collected at the time of stain correction.

The model database 103 stores the class hierarchy for each simulation model. In this embodiment, a new simulation model can be easily created by using a plurality of simulation models stored in the model database 103. In other words, it has the ability to fuse multiple existing independent simulation models.

The fusion a capability of the simulation model will be explained below. FIG. 7 shows model A shown in network diagram 701 and model B shown in network diagram 702.
The procedure for this case is shown in the explanatory diagram of FIG. 8. In addition, model A and model B
are model databases 103 or existing databases 10 as independent simulation models.
4, and is read out from this database 103 or 104 and displayed on the display device 203. Immediately after reading, the network diagram 701 shows only the nodes 706, 707 and the arc 712, and the network diagram 702 shows the node 708.
~711 and only arcs 715-717 are shown.

In the example shown in FIG. 7, first, using the pointing input device 205, the network diagram 701 is input to the model.
Create output-to-door 03 and transport node 7.07
and are connected by an arc 713 (step 5101). The central processing unit 2000 control unit 201 reconfigures the model A, adds elements to the output node class and arc class 503 of the lower class of the node class 502, and changes the model database 103 (step 510).
2). Next, an input two-door 04 is created in the network diagram 702 of model B, and connected to the rack node 708 with an arc 714 (step 5103). The model database 103 is also changed in the same way for the model (
Step 5104). Thereafter, a connecting arc 705 is connected between the nodes 703 and 704 of both networks using the pointing input device 205 (step 5105).

The control device 201 reconfigures the graph class 504, creates new copies of the elements in each class, and creates a model that is a fusion of model A and model B (step 3106). Note that if the simulation conditions of model A and model B are different, after performing the work to match them (steps 8107 to 5109),
The output two-door 03 and the input two-door 04 are subclasses of the node class 502 in FIG.
Like the arc and graph classes, it is stored in the model database 103 (step 5110).

Table 10 shows the relationship between the elements of each class in Model A, Model B, and the fusion model.

As mentioned above in Table 10, according to this embodiment, a new simulation model can be created by combining a plurality of independent stain 3 ration models.

That is, it becomes possible to reuse each model by using the simulation model fusion mechanism, and the amount of effort can be significantly reduced compared to the case where a new simulation model is created.

The simulation of the fused simulation model created in this way is executed according to the flowchart shown in FIG. Note that the same applies to independent simulation models.

In the above example, the discrete event system was modeled using a general graph, but when the discrete event system was modeled using a Petri net, the classes could be divided into a Petri net class, a place class, a transition class,
Simulation models can be stored by functionally classifying them into six types: token class and arc class.

Furthermore, by storing results such as changes in the lead time and buffer capacity of each part in the model database during simulation execution, it becomes possible to manage the model and results all at once.

Furthermore, XNS (Xerox Network S
If you use a network such as system),
It becomes possible to have a plurality of distributed model databases 103 shown in FIG. 1, and it becomes possible for a plurality of people to develop models at separate locations. In this case, by reusing models and using network functions, it becomes easier to develop and manage large-scale simulation models.

〔Effect of the invention〕

As described above, according to the present invention, it is possible to collectively save attributes and procedures of a model of a discrete event system for each function of an element in a graph. Therefore, it is possible to manage a plurality of stain 5 ration models with their respective graph structures, and the models can be easily reused. In addition, since models are managed using a database, it is possible for multiple people to develop a model separately and then combine them at the end. Therefore, the simulation model development period can be shortened. Furthermore, since it has an interface with an external database, it becomes possible to convert the schema of the external database into a simulation model and to register the simulation results in the external database. Therefore, the user does not need to be aware of the format of the external database.

[Brief explanation of drawings]

FIG. 1 is a schematic configuration diagram of the present invention, FIG. 2 is a configuration diagram of a workstation for implementing the invention, FIG. 3 is an explanatory diagram showing the flow of control in this embodiment, and FIG. 4 is an embodiment. An explanatory diagram showing a discrete event simulation model in
Fig. 5 is an explanatory diagram showing the class structure of model elements in this embodiment, Fig. 6 is an explanatory diagram showing model data of graph elements in the embodiment, Fig. 7 is an explanatory diagram showing model fusion, and Fig. 8 is an explanatory diagram showing the procedure for model fusion, No. 9
The figure is a flowchart showing processing at the time of staining. 101: Control unit 102: Model editor 1
03: Model database 104: Existing database 105 Command input section 1
06: Display section 200: Central processing unit 201: Control lid 202:
Main storage device 203: Display device 204: Data input device 205: Pointing input device 206: Auxiliary storage device 401 to 410: Node 4
11-419: Graph representation of arc 420 rainbow mini-region model 501:
Object class 502: Node class 503: Arc class 50
4 Nigelaf class 505: Parts class, 506
: Resource class 507: Process class 508 Two worker classes 509: Buffer class 601, 6
02: Node model data 701, 702: Network diagram 703: Output node 704: Human power to door 05
: Connection arc 706~711: Node 712~
717: Ark

Claims (1)

[Claims] 1. A means for creating a simulation model expressed in a graph structure, a means for classifying model elements of the simulation model by function and collectively managing attributes and unique procedures of the elements, and a means for managing relationships between model elements. A simulation model management device comprising: a means for managing attributes and procedures common to the entire model; and an interface means for exchanging model data with an external database. .