JPH11282824A - Simulation device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To reproduce an appropriate facility load that is approximate to an actual production line even when the a retroactive simulation is executed in a production system where a product having the branches in its process procedure is produced. SOLUTION: A branch rule calculation part 4 which calculates a branch rule in a confluence process includes a node searching means 41 which obtains the application, warehousing, branch and confluence processes included in a process procedure as the application, warehousing, branch and confluence nodes respectively, a branch probability matrix calculation means 42 which calculates a branch probability matrix of the branch node from the known branch probability of the branch node, a confluence ratio calculation means 43 which calculates a flow rate of the branch node to a branching destination process and then calculates a confluence ratio of the confluence node of the branching destination and a confluence rule generation means 44 which generates a confluence rule in the confluence process from the confluence ratio calculated by the means 43.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a simulation system for a production system or the like, and more particularly to a simulation system for simulating a production system or the like in a time progressing direction.

[0002]

2. Description of the Related Art Production line simulation is one method of planning in the field of manufacturing. The production system simulation (hereinafter, simulation) is intended to directly simulate the time transition of the location of the work object flowing on the production line and the time transition of the state of the equipment. Specifically, work procedures that determine the flow of work,
A production line model is prepared that grasps the manufacturing equipment, jigs, workers, and work constraints imposed on them for each work. Using the model of the production line, the production control of the actual line is simulated and executed in the computer.
In general, the simulation simulates the behavior of the production system in a work progress direction, that is, a time progress direction, with a past, present, or future situation as an initial condition. From that point, it is called forward simulation.

On the other hand, there is also a backward type simulation method in which the future situation is set as an initial condition, and the behavior of the production system is simulated in the reverse direction of the work progress and the time progress in the reverse direction. Some researches have been conducted in the past.

[0004] For example, as a reference 1, by George J. Mezzewski, pp. 716-720, 1985 Summer Computer Simulation Conference, "Multi-objective optimization of backward simulation and job shop scheduling for zero wait time and minimum production period". (George J. Mej
tsky: “Backward simulation and Multiple-Objective
Optimization of job shop scheduling with zero tard
iness and minimum makespan ", Proc Summer Simulation
Computer Simulation Conf, 1985, Vol.1985, pp.716-72
0)] describes a method of performing a simulation using a simulation model having a reversed process procedure, and inverting the time axis after the simulation to realize a time progress in the reverse direction.

[0005] References 2 and 3 describe the future prospects of the production line published in 1985, by Ichiro Inoue, Jin Kawano and Susumu Fujii, pp. 252 to 257, "Backward Simulation System for Production Management Support (Ichiro Inoue, Hiroshi Kouno
and Susumu Fujii: “A backward simulation system fo
r production management support ", Toward the Factor
y of the future, 1985, pp.252-257)], 1986, Modeling and Simulation in Engineering: Ichiro Inoue,
Hitoshi Kono and Susumu Fujii, Backward Simulation
Concepts, methodologies and tools-(Ichiro Inoue, Hiroshi Ko
uno and Susumu Fujii, Backward simulation-concept,
methodlogy and tool−, Modeling and Simulation in
Engineering, 1986, pp. 325-330)), Preserved, Co
It describes a method of performing a simulation in the reverse direction of the time axis by processing three types of events consisting of nfirmed and Definite.

[0006] Also, as abstracts of the abstracts of the Operations Research Society of Japan Autumn Research Conference, 1991, pp. 58-59, published by 1991, Masahiro Kumegawa, Masahiko Fuyuki, Ichiro Inoue, "Schedule generation by forward / backward simulation". Consideration of Method "and 1995, Japan Industrial Management Association, Vol. 46, No. 2, pp. 144-151, Masahiko Fuyuki and Ichiro Inoue, Backward / forward
"Delivery-oriented production scheduling based on the hybrid simulation method" describes a method that combines forward and backward simulation. Specifically, a job group A having a higher processing priority and a job
b Create two groups of B, first perform a forward simulation on Job A, create a simulation model in which the working time of Job A is a non-working time zone, and then create a working calendar This is a two-stage simulation method in which a backward simulation is performed on Job B and a plan is created under the restrictions described above.

[0007] Also, as a document 6, in Japanese Patent Application Laid-Open No. 8-314526, "Manufacturing Management System", forward type and backward type simulations are performed, and the error between the results and the reference manufacturing management data is calculated. It describes a method of evaluating the result and adopting a result with a small error as a plan.

[0008] Reference 7 is published in the United States in 1992.
J. Boukachour: "Forward and backward planning in object-oriented simulation" (J. Boukachour: "Forward And B")
ackward Planning Based OnObject-Oriented Simulatio
n, Proc Am Control Conf 1992, vol 4, pp.3218-3219) ''
Describes implementation of forward and backward simulation functions using an object-oriented approach. Specifically, simulation events are defined as abstraction classes, and forward-type events and backward-type events are prepared as derived classes, and derived classes are created for each of forward-type simulation and backward-type simulation, and functions unique to each simulation are created. Implemented. In the backward-type simulation described in the same document, as in other documents, a method in which the process procedure is reversed to that in the normal forward-type simulation and a method in which time is converted so as to be the reverse of the time progress direction are used. ing.

[0009] As Ref. 8, OMRON TEC published in 1993
HNICS, Vol. 33, No. 1, No. 105, pp. 98-103, Akimasa Horisawa and Masaki Tanaka, "Development of a Production Process Simulator Using Time Petri Nets and Implementation of a Simulator with Flexible Operation" It describes a modeling method for reversing the process procedure and a backward-type production simulation method using a time Petri net (TPN).

[0010] Also, as a reference 9, Edward F., 1993, Winter Simulation Conference, pp. 930-938.
Watson, Deborah J. Mediross, Randall P. Sadoboski, "Generating a Parts Input Plan by Backward Simulation" (Edward F. Watson, Deborah J. Medeiros,
Randall P. Sadowski: "Generating component release
plans with backward simulation "Proceedings of the
1993 Winter Simulation Conference, vol.1993, pp.930
-938)] describes a three-stage backward simulation that takes into account the assembly and production line.
More specifically, the first step is to perform a backward simulation based on the delivery date, and the second step is to perform a forward simulation by combining the initial work in progress and the input plan obtained in the first step. Describes a method of creating a work instruction list by performing a detailed simulation including a jig and a worker. In addition, conventionally, as a production planning method for assembly production lines,
After discussing the shortcomings of the MRP method, the necessity of backward simulation is described, and the conventional backward simulation technology is described.

Further, Japanese Patent Application Laid-Open No. 8-161394
(Patent No. 2671839) In the “simulation device”, in order to perform a time reverse simulation that maintains consistency with a work result in a normal time direction simulation, the work target having the latest occurrence time is the opposite to the conventional method. In addition to controlling the work progress by selecting the late-occurrence event, a method of dispatching a lot indicating an unstarted work object staying in the station in a time reverse direction is described.

[0012]

For a product manufactured in a production system, a sequence of a series of work steps in manufacturing corresponding to each product, that is, a process procedure is determined. In general, each work process (process) that constitutes a process procedure has a one-to-one relationship (straight line) with subsequent processes,
It has either a to-many relationship (branch, disassembly) or a many-to-one relationship (merging, assembling).

For example, in the manufacturing process of a product in a semiconductor diffusion process for forming a circuit pattern on a silicon wafer, there is a problem in defect inspection of a product in order to maintain high quality and yield in manufacturing. Rework (rework) is performed on the existing product, in which a specific manufacturing process is performed again.

In the manufacturing process of a semiconductor product, in addition to a manufacturing process procedure that does not include a rework procedure, a branch from an inspection process to a specific process procedure accompanying rework, and a transition from a specific process procedure to an original process procedure. Is expressed in the form of a merger.

In the production simulation, the rework of the product is modeled as a stochastically occurring event, and the normal process procedure or rework is performed based on the probability of passing the inspection in the inspection process or the probability of performing the rework. Generally, a technique of modeling so as to branch to any of the processing procedures is adopted.

In particular, in the semiconductor diffusion line, the frequency of reworking is from several percent to several tens of percents, and in addition to the manufacturing process which becomes a bottleneck in the production capacity of the production line in the reworking process procedure. Including the steps used, it is indispensable to incorporate the reworking procedure into the simulation model in order to keep the prediction accuracy required for the simulation high.

When performing such a simulation, it is necessary to statistically determine the probability of occurrence of rework in the inspection process, but such information is subject to production control because of its importance in quality control. Therefore, statistics are always taken, and it is often relatively easy to obtain as branch probability data when performing a simulation.

On the other hand, it is difficult to obtain statistical data on the merging ratio from each merging destination in the merging step because it is not necessary to obtain statistics in terms of manufacturing control. It is not necessary when performing a simulation.

However, when the backward simulation of such a production system is performed, the work process for manufacturing the product is performed in a reverse order to the normal procedure. That is, since the process steps are reversed, the branching step and the merging step in the forward simulation are respectively called the merging step and the merging step in the backward simulation.
It becomes a branching step, and the branch probability in the step that becomes the branching step in the backward simulation is the merging ratio in the merging step in the forward simulation.

In order to determine the merging ratio in the merging step, it is necessary to statistically determine the merging ratio.
Since the merging ratio is not important for production control, it was necessary to collect a large amount of merging ratio data for only backward simulation, process it statistically, and prepare it as a simulation model. These merging ratio data require a large amount of statistical processing at a time, and require a great deal of trouble to accurately follow the changing merging ratio. Therefore, it has been a problem in performing the backward type simulation to automatically obtain the merging ratio without explicitly preparing the simulation user.

In order to solve such a problem, the above-described prior art does not consider any event that occurs stochastically.
For this reason, simulations that take into account stochastic process branching associated with the rework have not been performed. As a result, it is impossible to maintain a high prediction accuracy by simulation because the amount of work of the product performed in the bottleneck manufacturing facility is not taken into account by a maximum of several tens of percent in the simulation.

Accordingly, the present invention provides a method for performing backward simulation, particularly when performing backward simulation for a production system that manufactures a product having a process procedure having a process branch whose branch destination is determined stochastically. It is an object of the present invention to provide a backward simulation means having a high prediction accuracy after performing a simulation without explicitly preparing a merging ratio in a merging step.

[0023]

SUMMARY OF THE INVENTION In order to solve such a problem, the present invention relates to a simulation apparatus for simulating each work object performed on a production line or the like in a direction opposite to a time progress direction. The following configuration is provided.

(1) An input / output unit that provides information necessary for performing a simulation and exchanges necessary information with an external device that transmits a simulation result, and a branch rule for calculating a branch rule in a merging process A calculation unit, a backward simulation execution unit that performs a backward simulation, and a storage unit that constantly stores and manages information required by the simulation device, and a branch rule calculation unit that calculates a branch rule in the merging process includes: Node search means for obtaining the input step, the receiving step, the branching step, and the merging step in the procedure as the input node, the receiving node, the branching node, and the merging node, respectively, and the branching probability of the branching node from the known branching probability of the branching node. The branch probability matrix calculation means for obtaining the matrix and the branch probability of the branch node are also calculated. Means for calculating a flow rate from a branch node to a branch destination process, and then calculating a merge ratio at the merge node of the branch, and a merge rule generation for generating a merge rule in the merge process from the calculated merge ratio Means comprising:

(2) In the simulation apparatus having the configuration shown in (1), a branch rule calculation unit for calculating a branch rule in a merging step includes a step in which a branch destination is definitely determined in the step procedure. Instead of defining a plurality of branch destination steps in the branch step as one of the branch destination steps of the branch step, the original process procedure is developed into a plurality of process procedures corresponding to the number of branch destination steps, and A configuration including a definite branching process procedure developing unit that associates the developed process procedure with the work target.

[0026]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing a first embodiment of the simulation apparatus according to the present invention. Referring to FIG. 1, a simulation apparatus 1 according to an embodiment of the present invention includes an input / output unit 3 that provides information necessary for performing a simulation and exchanges necessary information with an external device 7 that transmits a simulation result. A branch rule calculation unit 4 for calculating a branch rule in the merging process, a backward simulation execution unit 5 for performing a backward simulation, and a storage unit 6 for constantly storing and managing information required by the simulation apparatus 1. Have been.

A branching rule calculating unit 4 for calculating a branching rule in the merging step includes an input node, a receiving step, a branching step and a merging step in the process procedure as input nodes,
A node search means 41 for obtaining as a storage node, a branch node, and a merging node; a branch probability matrix calculating means 42 for obtaining a branch probability matrix of a branch node from a known branch probability of the branch node; and a branch based on the branch probability of the branch node After calculating the flow rate from the node to the branch destination process, a merge ratio calculating means 43 for calculating a merge ratio of the branch at the merge node, and a merge rule generating means for generating a merge rule in the merge process from the calculated merge ratio 44.

The operation of the embodiment apparatus having such a configuration will be described with reference to FIGS. 1 and 2. FIG. 2 is a flowchart showing a processing flow of the embodiment of the simulation apparatus 1. In the simulation apparatus 1, when the backward simulation is started, the simulation apparatus 1 generates a simulation model request signal (S1) via the input / output unit 3, and the external apparatus 7 sends each of the simulation model request signals (S1) as information necessary for performing the simulation. Simulation model data of a production line including information on a process procedure of a product is obtained (S1). The simulation device 1 generates a simulation model storage request signal (S2) and stores the obtained simulation model data in the storage unit 6 (F
101).

Next, the simulation apparatus 1 uses the node search means 41 to generate a node search request signal (S3).
Is generated, and in the process procedure of each product from the process procedure data managed in the storage unit 6 by the branch rule calculation unit 4,
The process corresponding to the input process, the receiving process, the branching process and the merging process is searched. FIG. 3 is a diagram showing one process procedure and a process of searching for the process procedure and extracting a node. The search process of the node search means 41 will be described with reference to the figure. The process procedure is traced from the input step, and the input step is set as the input node (N1). Thereafter, after examining the presence or absence of branching and merging in the process, if there is a branch in the process, it is determined as a branch node (N4, N5), and if there is a merger in the process, it is determined as a merging node (N2, N3), and there is no branch destination. For example, as a storage node (N6), the node search means 41 generates a node information storage request signal (S4) and records each node in the storage unit 6. Steps that have one branch destination, do not merge, and are neither input nodes nor storage nodes are not registered as nodes. In the process of extracting a branch / merge node by following the process according to the process procedure in this way, obtaining the first appearing node by following the process procedure set for each of the branch destination and the merge destination. Thus, the connection relation of the nodes is obtained while preserving the connection relation between the steps in the process procedure. Further, in the branch node, after obtaining a branch probability rji from the branch node i to each branch destination node j from the branch probability information of the process procedure, a storage request signal (S4) is generated and the information is stored in the storage unit. Record in 6. The above processing is performed for all the process procedures (F102,
F103, F104).

Next, in the simulation apparatus 1, the branch probability matrix calculation means 42 of the branch rule calculation unit 4 generates a node information request signal (S 5) and enumerates each process procedure stored in the storage unit 6. The obtained nodes are obtained, and each node is numbered in order, for example, from 1 to perform ordering. Next, the branch probability matrix calculation means 42
After generating a node information request signal (S5) and preparing a square matrix Rn of the number n of nodes enumerated, information for obtaining a branch probability at each branch node is obtained from the storage unit 6, and from the node i to the node j. Is set as the value of the element rji of the corresponding matrix (Equation 1).

[0031]

(Equation 1)

At this time, there is no branch to a plurality of processes, and the input node and the merging node for which the subsequent process is uniquely determined are as follows:
The branch probability to the node corresponding to the branch destination is set to 100%,
The branch probability to other nodes is treated as 0% and set as the value of the corresponding element in the matrix. For the receiving node, the branch destination to all nodes is set as 0%. The branch probability matrix calculation means 42 generates a branch probability matrix storage request signal (S6), and stores the square matrix Rn obtained for each process procedure in the storage unit 6 as a branch probability matrix for each process according to the above procedure ( F1
06).

Further, the total flow rate from the source node j to each node i is defined as a combined flow rate xi, and the total flow rate to all the nodes is defined as Xn (Equation 2).

[0034]

(Equation 2)

Next, the simulation device 1 generates a branch probability matrix information request signal (S 7) by the merging ratio calculation means 43 of the branch rule calculation unit 4, and
A branch probability matrix Rn stored for all process steps is obtained. Here, assuming that the production line is in a steady state, the branch probability rji at the branch node i
Is a steady state value, RnXn = Xn, and Xn can be further summarized using the n-th order square unit matrix In (Equation 3).

[0036]

(Equation 3)

When Rn is given, the merging ratio calculating means 4
3 obtains a combined flow rate xi, which is a sum of flow rates from the source node j to each node i, and a combined flow rate Xn to all nodes by solving the simultaneous equations of (Equation 3).

Next, the merging ratio calculation means 43 obtains the sum of the merging flow rates from all the nodes to the node i using the previously obtained Xn and the known Rn (Equation 4).

[0039]

(Equation 4)

Next, the merging ratio calculating means 43 obtains a ratio cij between the flow rate from the node j to the node i and the sum of the merging flow rates to the node i (Equation 5).

[0041]

(Equation 5)

Finally, the merging ratio calculation means 43 summarizes the merging ratios at all merging nodes into a merging ratio matrix C (Formula 6), and then generates a merged flow rate / merging ratio matrix storage request signal (S8). The combined flow rate Xn and the combined ratio matrix Cn obtained for each process procedure are stored in the storage unit 6 (F107).

[0043]

(Equation 6)

Next, the simulation device 1 generates a merged flow rate / merge ratio matrix request signal (S 9) by the merge rule generation means 44, and stores the merge ratio matrix C stored in the storage unit 6 for all process procedures.
n. The merging rule generation means 44 generates, as a branching rule in the merging step, a table in which the merging ratios from the merging source process are compiled for each merging step based on the merging ratio matrix, and a merging process merging rule storage request signal (S10 ) Is generated and the branching rule in each merging step of each step procedure is recorded in the storage unit 6 (F108).

After the merging ratio in the merging step of all the process steps is determined by the above procedure (F105, F106, F106).
107, F108, F109), the simulation apparatus 1
A backward process procedure / merging rule request signal (S11) is generated, a backward process procedure and a merging rule are obtained, and a backward simulation is started (F110). A result output signal (S13, S14) is generated, and a simulation result is output to an external device via the input unit 3. As described above, when there is a step that stochastically branches in the process procedure, by calculating the merging ratio in the merging step according to the above procedure, it has conventionally required a great deal of labor and time only by preparing data. The simulation can be performed immediately after automatically determining the merging ratio in the merging step.

With this configuration, there is a branch in the process procedure, and particularly, in a situation where the flow rate to the branch destination of the lot to be processed is so large that it cannot be ignored compared to the flow rate of the entire process procedure. When performing a simulation, an appropriate equipment load close to the actual production line is reproduced, and a simulation with higher reproducibility can be performed. This makes it possible to achieve convergence of the solution with a small number of repetitions when the backward simulation and the forward simulation, which are actual application examples of the backward simulation, are repeatedly performed. Therefore, efficient production planning can be performed.

Next, FIG. 4 is a block diagram showing a second embodiment of the simulation apparatus according to the present invention. Referring to FIG. 6, a simulation device 2 according to a second embodiment of the present invention includes a branch rule calculation unit that calculates a branch rule in a merging step in the configuration of the simulation device 1 according to the first embodiment of the present invention. 4. In a process procedure having a step in which a branch destination is definitely determined in the process procedure, instead of defining a plurality of branch destination steps in the branch step as one of the branch destination steps of the branch step, an original This is different from the first embodiment in that a definite branching step developing means 411 for expanding the stepping procedure into a plurality of stepping procedures corresponding to the number of branching destination steps and associating the stepping procedure developed to each work object is added.

The operation of the thus configured apparatus of the second embodiment will be described with reference to FIG. 4 and FIG. FIG. 5 is a flowchart showing a processing flow of the embodiment of the simulation device 2. Note that the operation of the simulation apparatus 2 is the same as that of the simulation apparatus 1 of the first embodiment except for the operation related to the node search means 41 and the definite branch step procedure 411 of the branch rule calculation unit 4, so that the description will be given here. Is omitted. The simulation apparatus 2 generates a node search request signal (S3) using the node search means 41 of the branch rule calculation unit 4, and inputs the signal in the process procedure of each product from the process procedure data managed in the storage unit 6. A process corresponding to a process, a receiving process, a branching process and a merging process is searched. Further, the node search means 41 generates a node information storage request signal (S4), and records information of the searched input node, storage node, branch node, and junction node in the storage unit 6. At this time, the branch probability from each branch node to each branch destination node is also obtained from the branch probability data and stored in the storage unit 6 (F102, F103).

The branching rule calculation unit 4 of the simulation apparatus 2 finds a corresponding branching step in the branching node recorded in the storage unit 6 and determines whether a branching destination step in the branching step has a step branching rule that is determined deterministically. Check whether (F
1031). If there is no branch step having a branch rule in which the branch destination step is definitely determined in the process procedure, a node search process is performed for the next process procedure (F1031 No).

When there is a branching step having a branching rule in which the branching destination step is definitely determined (F1031Ye)
s), the simulation apparatus 2 generates a definite branch process procedure development request signal (S15) and a process procedure information request signal (S16) by using the definite branch process procedure development means 411, and The information is obtained, the branch destination step in the branch step is determined to be one of a plurality of branch destination steps, and the process is expanded to the number of branch destination steps (F1032). At this time, the definite branch step procedure expanding means 411 generates a definite branch step development information / storage request signal, and determines each attribute for which a definite branch destination has been determined in each definite branch step by combining the step and the attribute. An attribute list is created and recorded in the storage unit 6, and then the determined branch attribute list is set as attribute information for each developed procedure.

Next, the simulation apparatus 2 obtains a work object using the process procedure before the development from the storage unit 6 by the definite branch process procedure developing means 411, and for each work object, in each deterministic branch process. An attribute for determining a branch destination and a process are combined to create a fixed branch attribute list. Next, from among the developed process procedures, a process procedure having the same definite branch list as the definite branch attribute list of the work object is obtained, and is newly recorded in the storage unit 6 as the process procedure of the work object. (F1033).

By the definite branching step procedure expanding means 411,
After developing the process procedure having the definite branch process into a plurality of process procedures, the simulation apparatus 2 performs the same operation as the simulation apparatus 1 according to the first embodiment of the present invention. Therefore, the description of the operation is omitted here.

With this configuration, even when there is a branching step in which a branch destination is determined deterministically in addition to a branching step in which a branch destination is stochastically determined in a process procedure, the probability branching step is performed in a stochastic branching step. It is possible to obtain a merging ratio in a merging step by branching. Further, even when only a deterministic branching step exists as a branching step in the process procedure, it can be developed into a process procedure without deterministic branching, so that the backward simulation can be easily performed.

[0054]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A first embodiment of the simulation apparatus according to the present invention will be described with reference to the drawings. FIG. 6 shows an example of a process procedure having a stochastic branching process and a simplified process procedure using the node 1, the node 2, the node 3, the node 4, the node 5, and the node 6 extracted by the node searching means 41. FIG. In the figure, the branch probability at node 1 is set to 100% with respect to node 2. Further, 100% is set so that the value of the input node 1 becomes the value of the input node 1 as it is in order to treat that the same value is always input to the node 1. The branch probability at node 2 is 100% relative to node 3, the branch probability at node 3 is 100% relative to node 4, the branch probability at node 4 is 10% relative to node 3, and The branch probability at node 5 is 5% for node 2, 10% for node 3, and 85% for node 6.
The branch at node 6 indicates that it does not exist for any node. The branch probability matrix R6 obtained by the branch probability matrix calculation means 42 is shown (Equation 7).

[0055]

(Equation 7)

Next, the total sum flow xi of the flow from the merge source node j to each node i and the merge flow X6 to all the nodes, obtained by the merge ratio calculating means 43 of the branching rule calculator 4, are shown ( Equation 8). X6 can be obtained by solving simultaneous equations. Note that any means may be used for solving the simultaneous equations.

[0057]

(Equation 8)

Next, the ratio cij between the flow rates from the node j to the node i to the node i and the sum of the flow rates to the node i calculated by the merging ratio calculating means 43 and the previously calculated X6 and R6 are combined into a merging ratio matrix C6. (Equation 9).

[0059]

(Equation 9)

FIG. 7 is a diagram showing a process procedure having a stochastic branching step at the time of performing the backward simulation corresponding to the merging ratio matrix C6 together with the merging ratio in the merging step. The branching rule at the time of the backward simulation in the merging step is generated in the form of a table of a merging ratio from the merging source node to each merging node. For example, the merging rule of the merging step (node 3) in FIG.
Backward simulation that generates a merge rule based on the merge ratios of 1 to 9%, node 2 to 81%, and node 4 to 10% in a table of one merge source and merge ratio, and reverses the process procedure. Is used as a branching rule.

The merging ratio in the merging step can be obtained based on the branching probability in the stochastic branching step given as described above, and backward simulation can be performed in consideration of the process procedure having the stochastic branching step. .

Next, a second embodiment of the simulation apparatus of the present invention will be described with reference to the drawings. FIG. 8 shows a process procedure having a branching rule in which the branch destination is determined deterministically.
It is a figure showing an example. In the figure, the step of performing deterministic branch is 2
In step 2, the branch destination is determined by the values A and B of the attribute R1 of the work object.

FIG. 9 is a diagram showing one embodiment of the expansion processing of the step procedure shown in FIG. As shown in FIG. 9, when the definite branching process procedure developing means 411 develops the process procedure for the process 2,
The procedure shown in FIG. 8 is developed into two procedures having no deterministic branch. Regarding the step procedure that branches to step 3 in step 2 which is one of the two step procedures after development,
An attribute list of R1 = A $ is given. In addition, an attribute list of {2, R1 = B} is given to the step procedure that branches to step 4 in step 2, which is the other expanded step procedure.

Next, the simulation apparatus of the present invention obtains, from the storage unit 6, information on the work object using the process procedure before the development by the definite branch process procedure development means 411.
By combining the process procedure for each work object from the input process, as shown in FIG. 9, the combination information of the attribute and the process for determining the branch destination in the definite branch process (attribute list)
Get. Based on this attribute list information, a process procedure after development having the same attribute list is searched for each work object, and the corresponding process procedure is set for each work object and stored in the storage unit 6. I do.

As described above, by developing a process procedure having a deterministic branch into a process procedure having no deterministic branch and processing the same, it is possible to completely consider a branch based on a deterministic branch in a backward simulation. Is possible.

[0066]

As described above, according to the present invention, when a backward simulation is performed on a product whose process procedure includes a step of performing a stochastic branch, a merging step necessary for the backward simulation is performed. Is automatically calculated by the simulation apparatus of the present invention by calculation, so that the backward simulation can be performed without the user preparing the merge ratio, and the rework process in the semiconductor diffusion line can be performed. In addition, a stochastic process procedure branch can be taken into a simulation model.

Since the backward simulation according to the present invention can handle stochastic process branching, more appropriate equipment load can be reproduced even in the backward simulation. Maintaining high simulation prediction accuracy, especially when the amount of work performed on the production equipment that becomes the bottleneck, such as reprocessing of semiconductor diffusion lines, occupies up to several tens of percent of the simulation. Can be. Therefore, according to the backward simulation device of the present invention, it is possible to perform highly accurate required construction period prediction and planning based on high prediction accuracy.

Further, in the backward simulation according to the present invention, even when there is a deterministic branching step determined by the preceding step in the process procedure, the process procedure can be developed into a process procedure without deterministic branching. A backward simulation having a branching step can be easily performed. In addition, in the backward simulation according to the present invention, in addition to the deterministic branching step in the process procedure, even when there is a stochastic branching step in which the branch destination step is stochastically determined, the process procedure has no deterministic branching. By developing the process procedure and calculating the merging ratio in the merging process that has been merged as a result of the stochastic process branching, a simulation can be performed in consideration of the deterministic branch and the stochastic process branching. .

[Brief description of the drawings]

FIG. 1 is a block diagram showing a first embodiment of the present invention.

FIG. 2 is a flowchart showing a processing flow of the first embodiment of the present invention.

FIG. 3 is a diagram showing one process procedure and a process of searching for the process procedure and extracting a node;

FIG. 4 is a block diagram showing a second embodiment.

FIG. 5 is a flowchart illustrating a flow of a process according to a second embodiment.

FIG. 6 is a diagram showing an example of a process procedure having a stochastic branching process and an example of a process procedure simplified by nodes extracted by the node searching means 41.

FIG. 7 is a diagram showing a process procedure having a stochastic branching step at the time of performing a backward simulation corresponding to a merging ratio matrix C6 together with a merging ratio in the merging step.

FIG. 8 is a diagram illustrating an example of a process procedure having a process branching rule in which a branch destination is definitely determined.

FIG. 9 is a diagram illustrating an example of a process procedure developing process performed by a definite branching process procedure developing unit 411.

[Explanation of symbols]

REFERENCE SIGNS LIST 1 simulation device 2 simulation device 3 input / output unit 4 branch rule calculation unit 5 backward simulation execution unit 6 storage unit 7 external device 41 node search unit 411 definite branch process procedure expansion unit 42 branch probability matrix calculation unit 43 merge ratio calculation unit 44 Merging rule generation means S1 Simulation model request signal S2 Simulation model storage request signal S3 Node search request signal S4 Node information storage request signal S5 Node information request signal S6 Branch probability matrix storage request signal S7 Branch probability matrix information request signal S8 Merging flow rate / merging Ratio matrix storage request signal S9 Merging flow rate / merging ratio matrix request signal S10 Merging process merging rule storage request signal S11 Reverse process procedure / merging rule request signal S12 Simulator Result storage request signal S13 simulation result output signal (→ input / output unit) S14 simulation result output signal (→ external device) S15 firm branch process procedure deployment request signal S16 process procedure information request signal S17 firm branch process deployment information / storage request signal F101 Simulation model input process F102 Enumeration of all process procedures and loop process F103 Node search process F104 Loop end judgment process F105 Enumeration of all process procedures and loop process F106 Branch probability matrix calculation process F107 Merge ratio calculation process F108 Merge rule generation process F109 loop End determination processing F110 Backward simulation processing F1031 Deterministic branch process determination processing F1032 Process procedure development processing F1033 Correlation processing between work objects and developed process procedures

Claims (2)

[Claims] 1. An input / output unit for providing information necessary for performing a simulation and for transmitting and receiving necessary information to and from an external device for transmitting a simulation result, and a branch rule calculation for calculating a branch rule in a merging process. Unit, a backward simulation execution unit that performs backward simulation, and a storage unit that constantly stores and manages information required by the simulation device, and performs simulation in which the simulation is performed in a time reverse direction opposite to the time progress direction. In the apparatus, the branch rule calculation unit includes: a node search unit that determines an input step, a storage step, a branch step, and a merge step in a process procedure as an input node, a storage node, a branch node, and a merge node, respectively; Branch probability matrix that calculates the branch probability matrix of the branch node from the branch probability of After calculating the flow rate from the branch node to the branch destination process based on the branch probability of the branch node, and calculating the merge ratio at the merge node of the branch branch. A simulation rule generating means for generating a merge rule in a merge step from a merge ratio. 2. The process procedure according to claim 1, wherein the branch rule calculation unit includes a process in which a branch destination is determined determinatively in the process procedure.
Instead of defining a plurality of branch destination processes in the branch process as one of the branch destination processes of the branch process, the original process procedure is developed into a plurality of process procedures corresponding to the number of the branch destination processes, and individual operations are performed. 2. The simulation apparatus according to claim 1, further comprising a definite branching process procedure developing unit that associates the developing process procedure with the object.