JPH10105205A - Material requirements counting method and system therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To shorten the processing time of material requirements planing(MRP) calculation and also to fast cope with various fluctuation by gathering data about items in each item and maintaining them as a corresponding item data dell. SOLUTION: This system generates an item cell 11 for each item and stores data such as a part table about items, inventory, back order in each data area of the item cell. The cell 11 stores addresses which store a item number, a master item counter 13, a low level code, requirements data 20, item data, part table data 23, inventory data, back order data, lot data and a next item cell. When impartiality of the number of calculable items among MRP calculating parts occurs, the impartiality is canceled by shifting the item cell data which is gathered in each item of a calculable item. Therefore negative load can be shifted without applying large overhead and calculation that is equivalent to conventionally successive MRP calculation can be acquired in a short time.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and a system for calculating a required material amount for obtaining, when a production plan is input, various parts and raw materials required for producing a product in accordance with the production plan and a delivery date. .

[0002]

2. Description of the Related Art When one product such as a motor is designated, one of the methods for automatically calculating the type and the number of parts required for producing the product is a material requirement plan (Material Requirements Planning). Requirements Planning (hereinafter referred to as MRP) is known.

[0003] The MRP and the conventional MRP calculation method are described in, for example, "500 illustrated MRP terms" (Nikkan Kogyo Shimbun, published in 1983). Hereinafter, the conventional MRP calculation method will be briefly described. .

[0004] In the MRP calculation, based on a drafted product-level production plan (Master Production Schedule, hereinafter referred to as MPS), "necessary items (items) for parts, raw materials, etc., when necessary (delivery date) ), Calculate the arrangement plan for purchasing or manufacturing as much as necessary (required). This calculation largely requires the following three pieces of information.

(1) MPS In general, items (products) at the top level in the parts exploded view
Production plan. Standard production plan.

(2) Parts table Data specific to an item called “item data” and products and parts, parts and parts called “product configuration data”
A list including two master data items, such as parts and raw materials, and data indicating the relationship between the items.

The item data includes, for example, the time required for manufacturing or purchasing the part (hereinafter referred to as “lead time”). Also, according to the product composition data, for example, what is a child item necessary to manufacture a certain item, and how many child items are required for one item to be manufactured (the number of child items required for each item) "
) Can be obtained.

[0008] (3) Inventory and backlog For each item, the current stock and the number of products in process, or the delivery date and planned number of items already ordered and scheduled to be delivered in the future (backlog information).

Based on such information, the following five calculations are performed for each item in the MRP calculation.

(1) Calculation of total required amount The required amount data of the item is read, the requested amount is collected into a certain period, and the total required amount for each period is calculated.
Hereinafter, for the sake of simplicity, the period will be one day, and daily planning will be described.

(2) Calculation of Net Requirement Based on the calculated total requirement, allocation is made for inventory and backlog, and a required net requirement is calculated for each day.

(3) Lot Summarization Based on the calculated daily net requirement, the lot is set to the most suitable quantity for the order using the lot size set for the item.

(4) Calculation of Lead Time An order is created by subtracting the lead time from the delivery date of the batched quantity and calculating the order date or start date.

(5) Developing required quantity The created order is developed into lower-level items using the parts table. More specifically, the required amount of each child item is calculated based on the child items and the number of members of the product configuration data, and the required amount is written to the required amount data of each item, with the day before the start date of the order as the request date.

The above calculations (1) to (5) are hereinafter collectively referred to as component development calculation.

In the MRP calculation, for all the products set by the MPS, a component development calculation of all the items required when manufacturing the products is performed. However, each item must be performed after all of the parts development calculations for its own parent item have been completed in order to determine the total required amount.

For example, in the case of a product having a configuration as shown in the parts exploded view of FIG. 19, the item CL includes the item HC and the item RC.
Child item. Therefore, the total required amount of the item CL is the sum of the required amounts from the item HC and the item RC, and the component expansion calculation of the item CL must be performed after the completion of the component expansion calculation of the item HC and the item RC.

In this example, a relatively simple component configuration of only one product is used. In practice, however, calculation is performed in a state where hundreds to thousands of products are composed of tens of thousands of items in a complicated manner. You need to do it. Therefore, in the conventional MRP calculation, a low-level code is introduced, and a component development calculation of each item is controlled using a queue. Although it is common to use a pointer called an activity chain instead of a queue, the control method is essentially the same. Therefore, a conventional method using a queue will be described below.
The RP calculation will be described.

Here, the low-level code will be briefly described. Based on the parts exploded view of a certain product as shown in FIG. 19, a level code can be assigned to each item according to which layer in the parts configuration. Some items can have a plurality of level codes, and the lowest level code among them is the low level code. Here, the term “lower” refers to a direction in which the level proceeds to a more basic item level on the parts exploded view.

Normally, as for the level code, the number of level codes increases as the product level is set to 0 level and the hierarchy proceeds toward the basic items. Therefore, the low level code is a level code having the largest value among the level codes of the item. This also means that, unlike the level code, only one low level code is determined for each item.

The low-level code of each item is set by searching all the items in the parts table before the MRP calculation starts, and is written in the item data as item-specific information.

As described above, by setting the low-level code, it is possible to calculate that all the items belonging to the next level can be calculated if all the parts development calculations for the items above a certain level have been completed. Can be guaranteed.

Therefore, the conventional MRP calculation utilizes a low-level code and determines the order of the component development calculation of each item on a level-by-level basis. Specifically, the process proceeds according to the following procedure.

(1) Reading of MPS For a product-level item in the MPS, the requested amount read from the MPS is written. At this time, the item name is put into the level 0 queue.

(2) Start of Calculation from Queue of Level 0 When reading of all MPSs is completed, one item name is taken out of the queue of Level 0 and the component development calculation of the item is performed. When finished, take out the next item name from the queue. Here, at the time of developing the required amount in the parts development calculation, the child item name is put in the queue of the low level code of the child item. However, if it is already in the queue, it does nothing.

(3) Level-by-Level Expansion When a queue at a certain level becomes empty, one item name is taken out from the queue at the next lower level, and a part expansion calculation of the item is performed. Until it is empty, it keeps getting out of the queue at that level. When expanding the required amount of the parts expansion calculation,
Queue the child item name to the child item's low-level code queue. However, if it is already in the queue, it does nothing.

(4) Completion of MRP calculation Computation is completed when all levels of queues become empty.

As an improvement of the above-described conventional MRP calculation method, there is a "tree structure data multiplex processing method" described in Japanese Patent Application Laid-Open No. 3-226845. In this method, tree structure data in which levels are set like a parts table are processed in parallel using one target level control unit and a plurality of data processing units. The target level control unit is
Until the processing of one level of all the data processing units is completed, the calculation instruction of the next level is not issued. Each data processing unit does not start calculation of the next level until instructed by the target level control unit. As described above, the progress of the parallel processing is controlled in synchronization with each level, and the processing speed is increased by performing the component development calculation of each item by a plurality of data processing units for each level.

[0029]

When a plurality of data processing units simultaneously perform a component development calculation for a plurality of items, it is necessary to distribute the load to each data processing unit. Here, the load of each data processing unit means the total of the calculation load amounts of the items to be calculated by the data processing unit. Assuming that the calculation load of each item is almost the same, the magnitude of the load of each data processing unit can be represented by the number of items to be calculated. Distributing the load to each data processing unit means allocating an item to be calculated to each data processing unit. The following three types of timings of the load distribution processing are considered.

(1) Method of performing load distribution processing at the same time as MRP calculation processing For example, the "tree structure data multiplex processing method" described in JP-A-3-226845 is this method. There is a data holding unit that collectively manages all the items to be calculated. The data holding unit allocates an item to be calculated to a plurality of data processing units, and each data processing unit calculates the assigned item. The data processing unit that has completed the calculation requests the data holding unit, which is collectively managed, to allocate a calculation item. Dynamic load distribution method or master
It is also called a slave method and is a general method for distributing the load in the process of processing.

(2) Method of Performing Load Balancing Processing Before MRP Calculation Processing Before performing MRP calculation, an item to be calculated is distributed to each data processing unit. Then, at the time of MRP calculation, each data processing unit processes only the items distributed to itself. This is called a static load distribution method.

(3) Method Combining (1) and (2) First, before performing the MRP calculation as in (2), an item to be calculated is allocated to each data processing unit. And MRP
At the time of calculation, each data processing unit basically proceeds with the process according to the item assigned to itself. However, when the load balance among the data processing units is lost, the method of changing the allocation of the calculation items from the heavy data processing unit to the light data processing unit is adopted.

Here, what is important is the data holding method. In the case of (1), there is a single data holding unit for managing the data of each item, and there is a problem that if the number of data processing units increases, data access competition occurs and the processing speed decreases. In the case of (2), since the data processing unit that accesses the data of each item is predetermined, a data holding unit that is paired with the data processing unit is prepared, and the data is stored therein to prevent data access competition. It is possible to reduce. However, in the MRP calculation, there is a possibility that the calculation load amount of each item dynamically changes due to inventory allocation or the like. Therefore, even if the items to be calculated are uniformly allocated to the respective data holding units in advance, the load balance between the respective data processing units is lost during the calculation process,
There is a high possibility that the overall processing speed will decrease. Therefore, (3) above is a promising method. However, there is also a problem in (3). In the case of (3), in order to avoid the problem of the data access conflict of (1), it is fundamental to have a data holding unit for each data processing unit as in (2).
However, in the course of the processing, there is a case where the calculation of a certain item is performed by a data processing unit assigned in advance and another data processing unit. At this time, a set of data on the item
The data must be transferred to the data holding unit of the data processing unit that is to be newly processed. Conventionally, data relating to each item is managed in a separate table for each of the parts table, stock / unscheduled information, and required quantity information. Therefore, in order to transfer data relating to a certain item from one data holding unit to another data holding unit, it is searched in each table and transferred. Due to the processing time of the data search and the data transfer, the overhead of the processing for maintaining the load balance is large, and the processing speed is eventually reduced. On the other hand, it is conceivable to make all data processing units hold data of all items, but there is a problem that a huge data area is required for each data processing unit.

The present invention solves the problem of the above method (3), shortens the processing time of the MRP calculation by parallel processing of the part expansion calculation, and quickly drafts a production plan in response to various fluctuations. That is its purpose. In particular, a method and system for calculating a required material amount, which enables the processing speed to be improved in accordance with the number of data processing units and does not reduce the processing speed even if the calculation load fluctuates in the course of the calculation.

[0035]

According to the present invention, in order to achieve the above object, it is necessary to read a separately determined production plan of a product and execute production of the product based on the production plan. In a material requirement calculation system including a plurality of MRP calculation means for calculating a type of a part or a raw material (hereinafter, referred to as an item), an amount of the item, and a delivery date of the item, the plurality of MRP calculation means include: Each M
With respect to an item that is determined in advance as an item to be calculated by the RP calculation means, at least a period required when manufacturing or purchasing the item, a parent-child relationship between the item and another item, and Holding means for holding the current inventory and the number of works in progress or the delivery date and the expected number of products already ordered and scheduled to be delivered in the future for each of the items, and holding them as item cell data corresponding to the items; MRP calculation means, based on the production plan and the item cell data corresponding to the item, respectively, the item corresponding to the item cell data required to execute the production of the product based on the production plan Of the child item, the quantity of the child item, and the delivery date of the child item can be calculated.

In addition, the plurality of MRP calculation means each execute a type of child item necessary for executing production of a product based on the production plan based on the production plan and the item cell data. Calculating means for calculating the quantity of the child item and the delivery date of the child item; and holding the item cell data corresponding to the necessary child item in its own holding means or in other MRP calculating means. Determination means to determine whether the means has, when the own holding means has,
The amount of the child item and the delivery date of the child item are added to the item cell data corresponding to the required child item, and the item cell data corresponding to the required child item is added to another MRP calculation means. In the case of having the holding means, the other MRP calculation means, the holding means of the other MRP calculation means has, in the item cell data corresponding to the required child item, the amount of the child item, It can be instructed to add the delivery date of the child item.

Further, the apparatus further comprises control means for controlling the operation of the plurality of MRP calculation means, wherein each of the plurality of MRP calculation means is one of the items which are predetermined to be calculated by the MRP calculation means. , The number of items for which the quantity of required items and the delivery date of the items have already been calculated,
Notifying means for notifying the control means as the number of calculable items, wherein the control means has a predetermined number of the calculable items in each of the plurality of MRP calculating means notified by the notifying means. Detecting means for detecting whether there is an MRP calculating means which has notified the number of items which can be calculated, and detecting that there is an MRP calculating means which has notified the number of items which can be calculated less than a predetermined number. Selecting means for selecting, among the plurality of MRP calculating means, an MRP calculating means having a large number of items that can be calculated,
For the MRP calculation means selected by the selection means,
An instruction to transmit the item cell data corresponding to a predetermined number of items out of the items for which the required amount of the item and the delivery date of the item have already been calculated to the MRP calculating means detected by the detecting means. Instruction means for performing the production of a product based on the transmitted item cell data in place of the MRP calculation means selected by the selection means. It is also possible to calculate the type of child item required for this, the amount of the child item, and the delivery date of the child item.

[0038]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is an overall block diagram of a system having a data holding method according to the present invention.

In this system, a plurality of MRP calculation units are connected to one work instruction unit 8 for controlling the entire system. The configuration of the MRP calculator 7 will be described with one of the MRP calculators as the MRP calculator 7.

The MRP calculator 7 has an initial processor 1 for performing an initial process of the MRP calculator 7 and a data manager 2 for managing data necessary for the MRP calculation. The data management unit 2 includes a data holding unit 32 that stores data required for MRP calculation for each item, and a computable item extraction unit 33 that extracts items for which MRP calculation can be performed. Further, the MRP calculation unit 7 includes a computable item management unit 3 that holds the number of items that can perform the MRP calculation and an address in the data holding unit 32 that holds data related to the item; A part development calculator 4 for performing
The result calculated by the component expansion calculation unit 4 is transmitted to another MRP calculation unit, the result calculated by the component expansion calculation unit of the other MRP calculation unit is received, and the data stored in the data holding unit 32 Is sent to another MRP calculation unit, or another MR
It has a request amount transmitting / receiving unit 5 for performing processing for receiving data stored in the data holding unit of the P calculating unit, and a message managing unit 6 for managing message transmission / reception.

As will be described in detail later, the data holding unit of each MRP calculation unit differs from the conventional data holding unit in that data required for MRP calculation is stored for each item.

Each MRP calculation section mainly calculates the items managed by itself, and if there is no item that can be calculated among the items managed by itself, the other MRP calculation section performs the calculation. It is characterized in that part calculation is advanced in parallel by having the calculation unit transmit data stored for each of the items that can be calculated.

Next, the operation of the system shown in FIG. 1 will be described.

When the message management unit 6 receives the MRP calculation start instruction from the work instruction unit 8, the MRP calculation unit 7 sends a processing start instruction to the initial processing unit 1. The initial processing unit reads the contents of the MPS data, the bill of materials, and the stock / unscheduled data, which are the required quantities of various products, to the data management unit 2 and stores the contents for each item in the data holding unit 32 in the data management unit 2. Is instructed to write in the data area of the "item cell" which is a table prepared in the above. Then, it instructs the computable item extraction unit 33 in the data management unit 2 to send the data addresses of all the item cells of level 0 to the computable item management unit 3.

The computable item management unit 3 includes the data management unit 2
When the data address is received from the computable item extraction unit 33 in the, it is registered in the level 0 queue as a computable item. After that, the data address is taken out of the queue and sent to the component development calculation unit 4. The part development calculation unit 4
When receiving the data address of the computable item from the computable item management unit 3, the component development calculation of the item is performed. The required quantity of the child item calculated by the part expansion calculation is the data management unit 2 if the child item is the target of calculation by the same MRP calculation unit 7.
If it is a calculation target of another MRP calculation unit, it is sent to the required amount transmission / reception unit 5.

The data management unit 2 sends the required amount of each item to the data holding unit 32 and writes the required amount in the data area of the required amount of the cell of the item. The required amount transmitting / receiving unit 5 writes the required amount of each item in the transmission buffer. Each time the transmission buffer becomes full, another MRP is passed through the message management unit 6.
The contents stored in the transmission buffer are sent to the calculation unit. In the MRP calculation unit at the transmission destination, when the requested amount transmission / reception unit receives the contents of the transmission buffer via the message management unit, it sends the contents to its own data management unit as the requested amount of each item.

The data management unit 2 sends the required amount of each item to the data holding unit 32 and writes the required amount in the data area of the required amount of the item. The data management unit 2 includes a computable item extraction unit 3
Manages the transfer of the requested amount between parent and child items, and if there is an item for which the requested amount from the parent item has been written in the required data area, the item is set as a computable item and the data address of the item is set to low level. It is sent to the computable item management unit 3 together with the code. However, this is performed only when the data is first written. Upon receiving the data address and the low level code of the item, the computable item management unit 3 registers it as a computable item in the queue of the low level code. The component expansion calculation unit 4 that has completed the component expansion calculation of the item of the level 0 data address sent from the computable item management unit 3
The entire contents of the transmission buffer of the request amount transmitting / receiving unit 5 are transmitted,
After that, a level 0 calculation completion report is sent to the work instruction unit 8 via the message management unit 6. When receiving the report, the work instructing unit 8 receives the level 0 completion report from all the MRP calculation units and sends a restart instruction to all the MRP calculation units. When the resumable item management unit 3 receives the restart instruction via the message management unit 6, it extracts the data address from the next level queue and sends it to the component development calculation unit 4. Thereafter, the same operation is performed for each level. The work instructing unit 8 which has received the final level calculation completion report from all the MRP calculation units,
An end instruction is sent to all MRP calculation units. This gives M
The RP calculation ends.

Here, at each level, each computable item management unit manages the number of items in the queue. According to the number of items, the state of each MRP calculation unit is divided into three types: blue, yellow, and red. Blue indicates that there are enough items to calculate and can be supplied to other MRP calculation units.
Yellow is in a state where it is faster to calculate it as it is than to provide it to other MRP calculators. Red is a queue with few items left and requests supply from other MRP calculators. It is in the state that it is. Each of the computable item management units updates their own state at the beginning of the level and when the state changes, via the message management unit to the work instruction unit 8.
Report to

The work instructing section 8 receives these reports and keeps the latest state of each MRP calculating section. In particular, some M
Upon receiving a report that the RP calculation unit has entered the red state, one of the MRP calculation units in the blue state is selected, and the corresponding MR
Instruct the P calculator to send the load to the MRP calculator in the red state. Upon receiving the instruction, the MRP calculation unit selects a plurality of items from the queue and removes them from the queue, and at the same time, replaces the cells of the items with the MRP in the red state.
Send to calculation unit.

The MRP calculation unit that has received the cell of the item stores the cell as it is in its own data holding unit,
The item name is put in a queue and processed.

As a result, first, since the data accessed by each data processing unit can be basically obtained from each pair of data holding units, no data access contention occurs and the number of data processing units is reduced. By increasing the number, the processing speed can be improved. Further, when the load balance is lost in the course of processing, it is possible to cause another data processing unit to perform an item that was to be calculated by each data processing unit. At that time, when transferring data related to the item in each data holding unit, since the information about the item is already collected and managed in one place, the search for the item only needs to be performed once. Since the information on the item is collected and can be easily sent together in the form of several items, the overhead of data transfer can be reduced. As a result, it is possible to keep the processing speed from lowering even if the load balance between the MRP calculation units is lost during the calculation process.

As shown in FIG. 2, the distributed memory type parallel computer has a PE (Proces) having one CPU and one memory.
sor Element), which are connected by a network. In such a distributed memory type parallel computer, component development calculation of a plurality of items is performed simultaneously in the MRP calculation process generated in each PE. Here, the assignment processing assumed in the present embodiment will be briefly described.

In the assignment process, an item is read from the parts table, assigned to each PE, and the result is stored in a table indicating the assigned PE number using the item number as a key. There are various methods for determining the assignment PE. For example, the simplest assignment determination method is to assign items in a bill of materials to each PE in turn. After determining the allocation PE for all the items using such an allocation determination method, the MPS, the bill of materials, the stock / unscheduled data on the items are distributed and arranged according to the allocation PE number of each item,
finish.

The configuration of the present invention in the distributed memory type parallel computer is similar to the MRP calculation unit 7 for each PE as shown in FIG.
A control process 10 including an RP calculation process 9, managing the processing states of the plurality of MRP calculation processes, and including a work instruction unit 8 for instructing start / end of the MRP calculation.
Prepare one. In the MRP calculation process, the data management unit manages input / output of data on the memory. The memory has an item cell 11 in which parts table data, stock / unscheduled data, and the like are rearranged for each item. The computable item management unit includes a job queue 15 that stores the item cell head address of a computable item for each low level, and a job counter 14 that indicates the number of jobs in the job queue for each low level. It also has a level counter 16 that indicates the low level being calculated. The request amount transmitting / receiving unit includes a request amount carrier 17 which is a transmission buffer for transmitting a request amount to an item allocated to another MRP calculation process, and an item cell carrier for transmitting item cell data for moving a load. 18 are provided for each of the other MRP calculation processes. The control process 10 prepares a state table 40 for holding the state of each MRP process.

Next, a specific processing flow of the present invention in the distributed memory type parallel computer will be described with reference to the drawings.

FIG. 4 is a flowchart showing the processing of the control process. The control process starts with M
An RP calculation process is generated (step 402), and the PRP of the MRP calculation process is calculated from the generated MRP calculation process.
The E number and the process number are received (step 404).
It is determined whether or not it has been received from all MRP calculation processes (step 406).
Return to step 404. When the PE number and the process number are received from all the MRP calculation processes, the received information indicates the name and M of the PE in which the MRP calculation process exists.
A list of RP calculation process numbers is created and distributed to all MRP calculation processes (step 408). The MRP calculation process creates a required amount carrier and an item cell carrier based on the distributed list. After distributing the list, the control process instructs all MRP calculation processes to read data such as a parts list necessary for MRP calculation at a time (step 410), and sends a batch read completion report from the MRP calculation process. Wait (step 412). When the batch read completion report is received from all the MRP calculation processes (step 414), next, an MRP calculation instruction is issued to all the MRP calculation processes (step 41).
6) Wait for a status report or level end report from the MRP calculation process (step 418).

Here, the definition of the state of each MRP process will be described. The state of each MRP process is determined by the number of items in the low-level queue currently being calculated. The status is divided into three types, blue, yellow, and red, in descending order of the number of items. For the number of items at the boundary between blue and yellow, for example, a limit value that is calculated earlier by itself rather than passed to another MRP process is used. This takes time to move the item cell data. If the difference in the number of items in the queue in each MRP process is smaller than a certain value, the item cell data is moved to equalize the number of items in the queue. This is because it may be faster to continue the calculation in each MRP calculation process while there is a difference in the number of items. The number of items at the border between yellow and red is
For example, the number of items that can be processed in the time from when a request for data of an item cell is requested from another MRP calculation process to when the data actually arrives is used. This allows
Each MRP calculation process completes all the items to be processed, and it is possible to reduce the state of playing without doing anything. Since the number of items at the boundary of each state depends on the performance of the parallel computer, it is determined based on experiments and the like. In summary, in the blue state, another load can be provided, in the red state, a load is requested from the other, and in the yellow state, no other load can be provided and no other request is made.

In step 418 of FIG. 4, if the message received from the MRP calculation process is a status report, the flow advances to step 426 to perform a status report corresponding process.

FIG. 5 shows a flowchart of the status report handling process in the control process. MR with control process
When the status report is received from the P calculation process, the corresponding column of the status table 40 shown in FIG. 6 is updated (step 50).
2). As shown in FIG. 6, the status table 40 includes a PE number and the MRP executed in the PE having the PE number.
The calculation process number, the number of jobs executed in the MRP calculation process having the MRP calculation process number, and the state of the MRP calculation process are stored as a pair for each PE.

It is determined whether the status report has reached the red status (step 504). If the status report has reached the red status, the MRP calculation process in the blue status and having the largest number of items in the queue is performed. Is retrieved (step 506). As a result of the search, it is determined whether or not the corresponding MRP calculation process exists (step 508), and a status report is sent to the corresponding MRP calculation process.
Send Export_MSG to the RP calculation process (step 510). This means that the corresponding MRP process in the search result is instructed to shift the load to the MRP process that has become red. Step 508
If there is no MRP calculation process in the blue state at
The "rived_MSG" is sent to the MRP calculation process that has reported the status (step 512). If it is determined in step 504 that a status report other than the status report of the red status has been received, Arrived_MSG is sent to the MRP calculation process (step 512).

Returning to the flowchart of FIG. Step 4
If it is determined at 18 that a level end report has been received, and at step 420 it is determined that a level end report has been received for a calculation item at that level from all MRP calculation processes, then for all MRP calculation processes, An MRP calculation instruction is issued to perform the next level calculation (step 422). It is determined whether or not all levels have been completed (step 424). If all levels have been completed, an instruction is issued to collectively write the required amount data of each item, which is the MRP calculation result (step 428), and the MRP calculation process starts. Wait for a batch write completion report (step 430). It is determined whether or not a batch write completion report has been received from all the MRP calculation processes (step 432). When a batch write completion report has been received from all the MRP calculation processes, the MRP calculation ends for all the MRP calculation processes. An instruction is issued (step 434), and the control process ends.

FIG. 7 is a flowchart showing the processing of the MRP calculation process. After being generated by the control process, the MRP calculation process sends the PE number of the PE in which the MRP calculation process exists and the process number of the MRP calculation process to the control process (step 702). A number list is received (step 704). From the received list, select another MR
A request amount carrier for transmitting the request amount to the child item allocated to the P calculation process and an item cell carrier for shifting the load are created for each MRP calculation process other than the own (step 706). Then, the system enters a state of waiting for a work instruction message from the control process (step 708), determines the type of the work instruction message received in step 710, and performs the following four processes according to the work instruction of the control process.

(1) Batch read processing (step 71)
2) The batch reading process is a process performed in response to a batch reading instruction. From the hard disk, data necessary for MRP calculation, such as MPS and a bill of materials, for the assigned item is collectively developed in the memory. At this time, an item cell 11 is created for each item for data such as a parts table, inventory / unscheduled data, and the like, and stored in each data area of the item cell as shown in FIG. Here, each data area of the item cell will be described. As shown in FIG. 8, the item cell includes an item number, a parent item counter 13, a low-level code, a required amount data 20, an item data, and a bill of materials data 2.
3. A data area is provided for storing inventory data, backlog data, lot data, and an address for storing the next item cell. Each item cell uses the item number of the item stored in the item cell as a key, and
Obtained by key conversion using the well-known Hash function
Hash tables [0] to [n] are stored in the corresponding table.

The level counter and other various counters are set to 0.
Initialize to After that, the MPS data is read one by one and written to the required amount data of the corresponding item. At this time, for the item for which the required amount data has been written, the item cell address is registered in the job queue of the low-level code of the item in the computable item management unit, and how many items are registered in the queue for each low level, respectively. , The job counter is incremented. After all the data has been loaded into the memory, a batch reading completion report is sent to the control process (step 714), and the process returns to step 708 to wait for a work instruction message.

(2) MRP Calculation Process (Step 716) The MRP calculation process is a process performed in response to an MRP calculation instruction. FIG. 9 shows a flowchart of this processing. The description will be given using the item-related data structure in FIG. 8 and the data structure of the job counter and the job queue in FIG. As shown in FIG. 10, the job counter 14 is prepared for each level,
Stores the number of computable items at that level. Further, the job queue 15 stores an address of an item which can be calculated at the level, which stores an item cell in the data management unit. Specifically, the job queue 15 has a List structure as shown in FIG. 10, and the first box of the job queue 15 has the address of the next box as the start address of a job that can be calculated at that level. Is stored in the next box address, the item cell address of the item that can be calculated at the level and the next box address of the box. Hereinafter, in each box, the item cell address of the item that can be calculated at the level and the address of the box next to the box are sequentially stored.

Referring to the flowchart of FIG. 9, first, a transmission message counter is initialized to 0 (step 902). After referring to the value of the job counter at the level indicated by the level counter and determining the status of blue, yellow, and red, a status report is sent to the control process together with the counter value. Then, the transmission message counter is incremented (step 904).

Then, a message reception check is performed (step 906). In the message reception check, if Carry_MSG prompting reading from the requested amount carrier has been received, “Carry_MSG corresponding processing” is performed (step 9).
08). Cell_M to prompt reading from item cell carrier
If the SG has been received, “Cell_MSG corresponding processing” is performed (step 910). Export_MSG indicating load transfer instruction
If it has received “Export_MSG”, it performs “Export_MSG corresponding processing” (step 912). If Arrived_MSG indicating that the transmitted message has arrived successfully has been received, the transmitted message counter is decremented (step 91).
4). After the processing corresponding to these messages is completed, the statuses of the blue, yellow, and red are determined by referring to the values of the job counter (step 916). If the status has changed, the status report is sent together with the job counter value. Is performed on the control process, and the transmission message counter is incremented (step 918). If the state has not changed, go to step 920.

Next, it is determined whether or not the value of the job counter is 0 (step 920). If the value of the job counter is 0, the destination MRP except for the empty requested amount carrier
Send all by Carry_MSG with process name,
The transmission message counter is incremented by the number of transmitted carriers (step 922). The transmitted request amount carrier is initialized (step 924). If all requested quantity carriers are empty, do nothing. Thereafter, it is further determined whether or not the transmission message counter is 0 (step 926). If the transmission message counter is 0, a level end report is made to the control process, and the level counter is incremented (step 928). Step 90
The process returns to the work instruction message reception waiting state of No. 6. If the transmission message counter is 1 or more in step 926, the process returns to step 906 for checking message reception.

If the job counter is not 0 in step 920, the item cell address of the computable item is extracted from the job queue (step 930), and "parts development calculation processing" is performed (step 932). Then, the job counter is decremented (step 934), and the process returns to the message reception check of step 906.

FIG. 11 is a flowchart of the “parts development calculation process”. This will be described with reference to the item-related data structure shown in FIG. The part development calculation process is described in
As described in the example of "500 RP terms", each process is performed in the order of total requirement calculation, net requirement calculation, lot combination, lead time calculation, and requirement development.

What is different from the conventional MRP is that “multiple PEs
In this case, related processing is added to the required amount development. Determining whether or not the request quantity transfer spans multiple PEs, processing if multiple PEs are processed, and, after calculating the required quantity, determining whether there is a registered item in the job queue and request quantity when there is no registered item Carrier transmission processing is added. More specifically, as shown in FIG.
If the number 23 is the same PE as the parent item, the item cell of the child item is searched, the delivery date and the required amount for all lots are calculated, and the “item cell writing process” is performed. If the child item has an assigned PE number different from that of the parent item, “request amount carrier write data area determination processing” is performed. And the same PE
In the same manner as described above, the delivery date and the required quantity for all the lots are calculated and written in the column of the item of the required quantity carrier.

The processing shown in FIG. 11 will be described one by one in order. First, in step 1102, a total requirement calculation, a net requirement calculation, a lot combination, and a lead time calculation are performed, and the requirement development is performed in step 1104 and thereafter. In step 1104, it is determined whether the required quantity has been developed for all the child items. When the required quantity has been developed for all the child items, the processing is terminated. If the required amount has not been developed for all child items, step 110
Proceeding to 6, it is determined whether the child item is an item assigned to the PE.

If the child item is an item assigned to the PE, the flow advances to step 1108 to search for an item cell of the child item. It is determined whether the delivery date and the required quantity have been calculated for all the lots of the child item (step 111).
0), when the delivery date and the required amount have been calculated for all lots, the process proceeds to step 1118, and the process returns to step 1104 with the next child item as the child item. Step 1110
If the delivery date and the required quantity have not been calculated for all the lots, the delivery date and the required quantity are calculated for the lot (step 1112), and a child item cell writing process is performed (step 1114). Is set as the lot (step 1116), and the process returns to step 1110.

At step 1106, the child item is
If not, the process proceeds to step 1120 to perform a required-amount carrier writing data area determination process. It is determined whether the delivery date and the required amount have been calculated for all the lots of the child item (step 1122). If the delivery date and the required amount have been calculated for all the lots, the process proceeds to step 1118, and the next child item is executed. , And returns to step 1104. If the delivery date and the required amount have not been calculated for all lots in step 1122, the delivery date and the required amount are calculated for the lot (step 11).
24), the delivery date and the requested amount are added to the requested amount carrier (step 1126), the next lot is set as the lot (step 1128), and the process returns to step 1122.

FIG. 12 is a flowchart of the "request amount carrier write data area determination process". An explanation will be given in conjunction with the item-related data structure in FIG. 8 and the required amount carrier data structure in FIG. As shown in FIG. 13, the requested amount carrier can store the destination PE number 26, the destination process number 27, the item number 28, and the delivery date and requested amount of the item having the item number.

In the request amount carrier write data area determination processing, first, the destination PE number 26 of the request amount carrier is
The requested amount carrier having the same PE number 23 as the child item data is searched (step 1202). Then, it is checked whether the requested amount of the same child item has already been written in the requested amount carrier (step 1204). If it has been written, the data area is used as a writing area for the child item (step 1206). If not, the new data area is used as the writing area for the child item. At this time, it is determined whether or not the requested data carrier is full and a new data area cannot be obtained (step 1208).
The content stored in the requested amount carrier is transmitted via the message management unit (step 1210). At this time, the transmission message counter is incremented. After the transmission, the required amount carrier is initialized (step 1212), and the new data area is set as a writing area for the child item (step 1214). In step 1208, if there is a vacancy in the requested amount carrier and a new data area is obtained, the process proceeds to step 1214, where the new data area is set as a writing area of the child item.

FIG. 14 is a flowchart of the "child item cell writing process". This will be described with reference to the item-related data structure shown in FIG. The child item cell writing process adds the delivery date and required amount of the child item to the required amount data of the item cell of the child item (step 1402). Then, it is determined whether or not the requested amount is for the final lot (step 140).
4) If the requested quantity is not for the final lot, the process is terminated. If the requested quantity is for the last lot, the parent item counter 13 of the child item is incremented (step 1406), and it is determined whether the value of the parent item counter 13 is 1 as a result of the increment. (Step 14
08). If the value of the parent item counter 13 is 1, the address of the item cell is registered in the job queue of the level code of the item (step 1410), and the job counter of the level code is incremented (step 141).
2) End. At step 1408, parent item counter 1
If it is determined that the value of 3 is not 1, the process ends.

FIG. 15 is a flowchart of the "Carry_MSG corresponding process". This will be described with reference to the item-related data structure of FIG. 8 and the required amount carrier data structure of FIG. Carry_MS
In the G corresponding processing, the delivery date and the requested amount of all the stored child items are read from the received requested amount carrier. And
The delivery date and the required quantity are written in the item cell of the child item. Also,
The parent item counter 13 is incremented, and the process ends. As a result of the increment, if the value of the counter is 1, the address of the item cell is registered in the computable job queue of the level code of the item, and the job counter is incremented. This is performed for all items in the requested amount carrier.
After that, Arrived_MSG is sent to the MRP process of the request amount carrier transmission source.

The processing shown in FIG. 15 will be described one by one.
First, in step 1502, it is determined whether or not the delivery date and the requested amount of all the stored child items have been read from the received requested amount carrier. If it is determined that the delivery date and the requested amount of all the child items have been read, the Carry_MSG transmission source M
Arrived_MSG is transmitted to the RP process (step 1514), and the process ends. If it is determined that the delivery date and the requested amount of all the child items have not been read, the delivery date and the requested amount of one item are read from the requested amount carrier and written in the item cell (step 1504). Further, the parent item counter 13 of the item cell is incremented (step 1506). Next, the value of the parent item counter 13 is 1
Is determined (step 1508). If the value of the parent item counter 13 is not 1, step 1
Return to 502. If the value of the parent item counter 13 is 1, the address of the item cell is registered in the computable job queue of the level code of the item (step 151).
0), the job counter is incremented (step 1512). Then, the process returns to step 1502.

FIG. 16 is a flowchart of the “Cell_MSG corresponding process”. In the Cell_MSG corresponding process, the content of each item cell of the received item cell carrier is newly registered as an item cell, the address of the item cell is registered in a queue at the level indicated by the level counter, and the job counter is incremented. After performing for all items of the item cell carrier, the transmission message counter is decremented, and Arrived_MSG is sent to the MRP calculation process of the item cell carrier transmission source.

FIG. 17 shows the data structure of the item cell carrier. As shown in FIG. 17, the item cell carrier can store a destination PE number 29, a destination process number 30, and an item cell 31.

The processing shown in FIG. 16 will be described one by one.
First, in step 1602, it is determined whether or not all item data has been read from the item cell carrier. If it is determined that all the item data has been read from the item cell carrier, the transmission message counter is decremented (step 1609), and Arrived_MSG is transmitted to the MRP calculation process of the Cell_MSG transmission source (step 161).
0) End. If it is determined that all the item data has not been read from the item cell carrier, the item cell data of one item from the item cell carrier is transferred to the data holding unit (step 1604). Then, the address of the item cell is registered in the queue at the level indicated by the level counter (step 1606), and the job counter is incremented (step 1608). After this, step 16
Return to 02.

FIG. 18 is a flowchart of “Export_MSG corresponding processing”. Item related data structure of FIG. 8, FIG.
The description will be made using the item cell carrier data structure of FIG. Expo
In the rt_MSG corresponding process, first, the state of blue, yellow, and red is determined with reference to the value of the job counter (step 1802). This is performed because the state of the state table managed by the control process may be different from the current state. It is determined whether or not the status is blue (step 1804).
ved_MSG is sent (step 1814), and the process ends. This means that no job is supplied to the red MRP process.

When the MRP process receiving Export_MSG is in the blue state, the number of item cell addresses that can be mounted on the item cell carrier is extracted from the queue (step 18).
06), the contents of the item cell are transferred to the item cell carrier (step 1810) and sent to the destination MRP calculation process (step 1812). At this time, the job counter is decremented by the amount taken out (step 180).
8) Increment the transmission message counter.

(3) Batch write process In the batch write process, the required amount data of each item, which is the output result of the MRP calculation in the memory, is written to the hard disk at a time (step 718 in FIG. 7). After the writing, a batch writing completion report is made to the control process (step 720), and the system enters a work instruction message reception waiting state.

(4) Termination processing Termination processing of the process is performed.

The embodiment of the distributed memory type parallel computer has been described above. However, the present invention can be realized not only in this information processing environment but also in a conventional single CPU computer and a shared memory type parallel computer.

[0089]

As described above, a plurality of MRP calculation units or a part development calculation unit can simultaneously perform a part development calculation of a plurality of items. When an imbalance occurs in the number of computable items between the MRP calculation units, the imbalance is eliminated by transferring the item cell data compiled for each item of the computable items, so that a large overhead is required. The load can be moved without any change, whereby it is possible to obtain a calculation result equivalent to the conventional sequential MRP calculation in a short time.

For this reason, daily changes from sales, design, materials, and workplaces can be immediately reflected in the production plan, and problems caused by the changes can be quickly extracted and dealt with.

[Brief description of the drawings]

FIG. 1 is an overall block diagram.

FIG. 2 is an explanatory diagram of a distributed memory type parallel computer.

FIG. 3 is a configuration diagram of the present invention on a distributed memory type parallel computer.

FIG. 4 is an explanatory diagram of a flowchart of a control process.

FIG. 5 is an explanatory diagram of a state report correspondence processing flowchart.

FIG. 6 is a data structure of a state table.

FIG. 7 is an explanatory diagram of a flowchart of an MRP calculation process.

FIG. 8 is an explanatory diagram of a data structure for storing item-related data.

FIG. 9 is an explanatory diagram of a flowchart of an MRP calculation process.

FIG. 10 is an explanatory diagram of a data structure of a job count and a job queue.

FIG. 11 is an explanatory diagram of a flowchart of a component development calculation process.

FIG. 12 is an explanatory diagram of a flowchart of a required amount carrier write data area determination process.

FIG. 13 is an explanatory diagram of a data structure of a requested amount carrier.

FIG. 14 is an explanatory diagram of a flowchart of a child item cell writing process.

FIG. 15 is an explanatory diagram of a flowchart of Carry_MSG corresponding processing.

FIG. 16 is an explanatory diagram of a Cell_MSG corresponding processing flowchart.

FIG. 17 is an explanatory diagram of a data structure of an item cell carrier.

FIG. 18 is an explanatory diagram of an Export_MSG corresponding processing flowchart.

FIG. 19 is an explanatory diagram of a component development view and a low-level code.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Initial processing part 2 ... Data management part 3 ... Computable management part 4 ... Parts development calculation part 5 ... Request amount transmission / reception part 6 ... Message management part 7 ... MRP calculation part 8 ... Work instruction part 9 ... Distributed memory type parallel computer MRP calculation process on the above 10 ... Control process on the distributed memory type parallel computer 11 ... Item cell 13 ... Parent item counter 14 ... Job counter 15 ... Job queue 16 ... Level counter 17 ... Requested amount carrier 18 ... Item cell carrier 20 … Item cell request amount data 22… Item cell low-level code 23… Item cell child item data allocation PE number 26… Request amount carrier destination PE number 27… Request amount carrier destination process number 28… Request amount carrier Item number 29… Item cell carrier destination PE number 30… Item cell Destination process number 31 ... item cell of item cell carrier Yaria 40 ... state table

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Tetsushi Tomizawa 890 Kashimada, Saiwai-ku, Kawasaki-shi, Kanagawa Prefecture Information Systems Division, Hitachi, Ltd.

Claims (4)

[Claims] 1. A method for reading a separately defined production plan of a product, the types of parts or raw materials (hereinafter referred to as "items") required to execute production of the product based on the production plan, and the type of the item. MRP to calculate quantity and delivery date of the item
In the material requirement calculation system including a plurality of calculation means, each of the plurality of MRP calculation means manufactures or purchases at least the above-mentioned item with respect to an item predetermined as an item to be calculated by the MRP calculation means. The required time period, the parent-child relationship between the item and other items, the current inventory and the number of works in progress for the item, or the delivery date and the expected number of items already ordered and scheduled to be delivered in the future, for each item And holding means for holding as item cell data corresponding to the item, wherein the plurality of MRP calculation means each include the production plan,
Based on the item cell data corresponding to the item, the type of the child item of the item corresponding to the item cell data required to execute the production of the product based on the production plan, and the amount of the child item And a delivery date of the child item. 2. The method according to claim 1, wherein the plurality of MRP calculation means are each based on the production plan and the item cell data, and execute the production of a product based on the production plan. Calculating means for calculating the quantity of the child item and the delivery date of the child item; and holding the item cell data corresponding to the required child item in its own holding means or in another MRP calculating means. A determination means for determining whether or not the means has the means, and if the holding means has its own means, the quantity of the child item, the delivery date of the child item, and If the holding means of the other MRP calculation means has the item cell data corresponding to the necessary child item, the holding means of the other MRP calculation means has the holding means of the other MRP calculation means. Corresponding to the necessary child items The item cell data, the child item quantities and, material requirements computing system of claim 1, wherein the instructing to append the delivery of the child item. 3. An apparatus according to claim 1, further comprising control means for controlling an operation of said plurality of MRP calculation means, wherein each of said plurality of MRP calculation means is one of items predetermined to be calculated by said MRP calculation means. A notification unit for notifying the control unit of the number of items for which the amount of the required item and the delivery date of the item have already been calculated, as the number of calculable items, wherein the control unit includes the notification unit Detecting means for detecting whether there is an MRP calculating means that has notified the number of calculable items smaller than a predetermined number, among the number of calculable items in each of the plurality of MRP calculating means, When the detecting means detects that there is an MRP calculating means that has notified the number of calculable items less than a predetermined number, the MRP meter having a large number of calculable items among the plurality of MRP calculating means. Selecting means for selecting the calculating means; and MRP calculating means selected by the selecting means.
An instruction to transmit the item cell data corresponding to a predetermined number of items out of the items for which the required amount of the item and the delivery date of the item have already been calculated to the MRP calculating means detected by the detecting means. And MRP calculating means detected by the detecting means, in place of the MRP calculating means selected by the selecting means,
A method for calculating a type of a child item, an amount of the child item, and a delivery date of the child item required to execute product production based on the transmitted item cell data. The material requirement calculation system according to 1 or 2. 4. A part or raw material (hereinafter referred to as "item") required for reading a separately defined production plan of a product and executing production of the product based on the production plan, and the type of the item. MRP to calculate quantity and delivery date of the item
A material requirement calculation method in a material requirement calculation system including a plurality of calculation means, wherein the plurality of MRP calculation means each include at least an item which is predetermined as an item to be calculated by the MRP calculation means. , The period required when manufacturing or purchasing the item, the parent-child relationship between the item and other items, the current inventory and the number of works for the item or the delivery date that has already been ordered and will be delivered in the future. The planned number is grouped for each of the items and stored as item cell data corresponding to the item.
Based on the item cell data corresponding to the item, the type of the child item of the item corresponding to the item cell data required to execute the production of the product based on the production plan, and the amount of the child item And a delivery date of the child item.