JPH0991011A - Load distributing device and its method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To automatically controlling the distribution of load of each programmable controller(PC) in accordance with the load state of each PC. SOLUTION: The PC system is constituted of connecting plural PCs 100-1 to 100-n through a network 200. Each of the PCs 100-1 to 100-n calculates its own load and requests another PC to execute at least a part of its own processing.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a load balancing apparatus and method in a programmable controller system in which a plurality of programmable controllers are connected via a high speed network, and in particular, each programmable controller itself automatically manages its load. The present invention relates to a load distribution device and method for distributing a load among programmable controllers to improve processing efficiency.

[0002]

2. Description of the Related Art Recently, there has been proposed a programmable controller system configured by connecting a plurality of programmable controllers via a high speed network.

That is, in order to cope with the recent diversification of FA (factory automation) systems, one programmable controller has a limit in its control mode. Therefore, a plurality of programmable controllers are connected via a high speed network. By constructing a programmable controller system, the programmable controller system can be used to achieve an efficient F
Attempts have been made to build an A (Factory Automation) system.

[0004]

However, generally, in the conventional programmable controller, the relationship between the programmable controller main body and physical resources such as sensors and actuators related to the programmable controller main body is uniquely determined. Therefore, even when a programmable controller system is constructed by connecting a plurality of programmable controllers via a high-speed network as described above, each programmable controller main body and the sensors related to this programmable controller main body are connected. , The relationship with physical resources such as actuators is uniquely determined, and even if multiple programmable controllers have a sufficient load, the load distribution can be distributed among the programmable controllers. Cormorants system did not exist.

Therefore, it is an object of the present invention to provide a load balancing apparatus and method in a programmable controller system that can automatically control the load of each programmable controller according to the load state of each programmable controller. To do.

[0006]

In order to achieve the above object, the load balancing apparatus of the present invention is a load balancing apparatus in a programmable controller system configured by connecting a plurality of programmable controllers via a network. The programmable controller comprises: load calculating means for calculating its own load; and processing requesting means for requesting at least a part of its own processing to another programmable controller corresponding to the load calculated by the load calculating means. It is characterized by doing.

The load balancing method of the present invention is a load balancing method in a programmable controller system configured by connecting a plurality of programmable controllers via a network, wherein each programmable controller calculates its own load, The other programmable controller is requested to perform at least a part of its own processing corresponding to the calculated load.

Further, according to the present invention, the programmable controller executes a divisible ladder program, and the process requesting means requests another programmable controller to perform a part of the process of the divided ladder program. Characterize.

Further, the processing requesting means requests the processing after the ladder program is created and when the system is not operating.

The processing requesting means may request the processing after the ladder program is created and when the system is operating.

Further, there is further provided a shared storage means which is shared by the plurality of programmable controllers and stores the instruction execution time of each programmable controller, the average instruction execution time of all programmable controllers, and the programmable controller with the minimum load. The processing request means selects the programmable controller with the minimum load stored in the shared storage means as the processing request destination programmable controller when requesting the processing.

When requesting the processing, the processing requesting means checks the instruction execution time of each programmable controller stored in the shared storage means in a predetermined order, and based on the first found average instruction execution time. It is characterized in that a programmable controller having a short instruction execution time is selected as a process request destination programmable controller.

Further, when the processing request is made, the processing requesting means issues a message inquiring to all other programmable controllers whether or not the processing request can be made, and a response indicating that the processing request can be made is returned. It is characterized in that the programmable controller is selected as the programmable controller to which the process is requested.

Further, the processing requesting means selects the processing which minimizes the overhead generated when requesting at least a part of its own processing to another programmable controller and requests it to the other programmable controller. To do.

Further, the processing requesting means selects a processing which requires a minimum amount of data to be newly shared by the programmable controller when the processing is requested.

Further, the processing requesting means comprises a program code rewriting means for rewriting a program code describing an input and an output relating to the request of the processing when requesting at least a part of its own processing to another programmable controller. It is characterized by doing.

Further, each of the programmable controllers includes a program status table for managing a ladder program in the divided processing units, and the plurality of programmable controllers store data shared by the plurality of programmable controllers. It is characterized in that it has a shared data table to be managed, and actual processing in each programmable controller is executed by referring to the program status table, and the request for the processing is executed by referring to the shared data table.

Further, the program state table is such that identification information for identifying each process in each process unit, for each process unit, the process unit is being executed, or the process unit is another programmable controller. Status information indicating whether or not the processing unit is requesting the processing unit from another programmable controller, input data information indicating input data required by each processing unit, and each processing unit requires It is characterized by storing output data information indicating output data and code information describing the contents of actual processing in each processing unit.

Further, the shared data table is data shared by the plurality of programmable controllers,
It is characterized in that an address of the shared data and identification information for identifying a process using the shared data are stored.

With this configuration, the programs can be distributed in each programmable controller (PLC) without the user consciously arranging the programs, and at the time of operation, each programmable controller (PLC) can be distributed. Since it is possible to reflect the decentralization while considering the load of, it becomes possible to build an efficient FA (factory automation) system.

Further, even if one new programmable controller (PLC) is added and programming is not performed for it, the programmable controller (P
The LC) can automatically reduce the load on the programmable controller (PLC) in the surroundings, and can easily obtain the performance improvement of the system commensurate with the cost of the equipment required.

[0022]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of a load balancing apparatus and method according to the present invention will be described below in detail with reference to the drawings.

FIG. 1 is a block diagram showing a schematic configuration of an embodiment of a programmable controller system configured by applying the load balancing apparatus and method according to the present invention.

In FIG. 1, this programmable controller system is constructed by connecting a plurality of programmable controllers (PLC) 100-1 to 100-n to a high speed network 200. In addition, the programmable controller system includes an in-network global memory 30 shared by a plurality of programmable controllers (PLC) 100-1 to 100-n.
0 is provided, and the global memory in this network 3
00 is also connected to the high speed network 200.

Programmable controller (PLC) 1
00-1 to 100-n are read-only memories (ROMs) 101 for storing system programs,
A program status table 102-1, a shared data table 102-2, a random access memory (RAM) 102 that stores a process request enable flag 102-3, an input / output memory (I / O memory) 103, a central processing unit (C).
PU) 104, network interface (I /
F) 105, which are connected to the internal bus 106 in the programmable controllers (PLC) 100-1 to 100-n.

Here, a read only memory (ROM)
The system program stored in 101 is a program for controlling the operation of the programmable controllers (PLC) 100-1 to 100-n main body, and describes the "processing request method" and "processing after request" according to the present invention. Contains. The "processing request method" and the "processing after request" will be described in detail later.

Random access memory (RAM)
Program state table 102-1 stored in 102
Describes in what state the divided program units are and what kind of input and output they have. Also,
It has data indicating the processing content regardless of whether the pointer is a processing code or the processing code itself.

Random access memory (RAM)
The shared data table 102-2 stored in 102 is
It describes the inputs and outputs related to the processing requested by another programmable controller (PLC) by its own programmable controller (PLC) via the network 200.

Random access memory (RAM)
The process requestable flag 102-3 stored in 102 is
A flag indicating whether or not one's programmable controller (PLC) can request processing to another programmable controller (PLC). This processing requestable flag is
It is turned on at the start of the operation of the programmable controller (PLC). In addition, the programmable controller (P
When it is found that there is no internally executed processing unit during the operation of (LC), this processing requestable flag is set to O.
Set to FF.

The central processing unit (CPU) 104
Performs user program execution or peripheral processing.

In-network global memory 300
1) instruction execution load of each PLC 2) average instruction execution load of PLC in the network 3) PLC with the least load in the network 4) input / output information shared between PLCs is stored.

Here, the "instruction execution load of each PLC" is
The instruction execution load of each of the plurality of programmable controllers (PLC) 100-1 to 100-n in the network 200 that configures this programmable controller system. As the instruction execution load, each programmable controller (PLC) 100-1 to 100-1. For example, a simple instruction execution time in one cycle, a ratio of the instruction execution time in one cycle, and the like of 100-n "execution instructions" can be used. In short, each programmable controller (P
Any value can be used as long as it is an index capable of relatively judging the load of the “execution instruction” of LC) 100-1 to 100-n.

The "average instruction execution load of PLCs in the network" is the average instruction execution load of each programmable controller (PLC) 100-1 to 100-n in the network 200. Here, if the dimension of the instruction execution load of each programmable controller (PLC) 100-1 to 100-n is the execution time, the arithmetic mean is calculated,
If it is a ratio, it is suitable to take a geometric mean.

The "PLC with the least load in the network" is a programmable controller (PLC) with the lightest load of executing the first instruction among the programmable controllers (PLC) 100-1 to 100-n in the network 200. Is a variable that indicates. This variable is used for each programmable controller (PLC) 100-1 to 100
The node address indicating -n may be an ID value unique to each programmable controller (PLC) 100-1 to 100-n.

The "input / output information shared between PLCs" is
By requesting processing to another programmable controller (PLC), the programmable controller (PL)
This is a reference area for input / output information that needs to be shared between C).

Before describing the detailed operation of the programmable controller system of this embodiment, the basic concept of the load balancing apparatus and method according to the present invention will be described.

First, as shown in FIG. 2, two programmable controllers (PLC1 and PLC2) 100 are provided.
-1, 100-2 to high-speed network (LAN) 200
Consider the configuration connected to.

Here, the programmable controller (P
It is assumed that the processing in LC1) 100-1 is much larger than the processing in programmable controller (PLC2) 100-2. In this case, in the load distribution apparatus and method according to the present invention, a part of the processing in the programmable controller (PLC1) 100-1 is requested to the programmable controller (PLC2) 100-2 to distribute the load of the entire system.

Here, the programmable controller (P
Assuming that the programs of LC1) 100-1 and programmable controller (PLC2) 100-2 are composed of ladder programs, this ladder program can be divided into small circuits as shown in FIG.

In FIG. 3, the load of the execution process of the ladder program of the programmable controller (PLC1) 100-1 (programmable controller 1) is very large, and the programmable controller (PLC2) 10
0-2 (programmable controller 2) shows a case where the load of the execution process of the ladder program is very small.

In this case, the programmable controller 1
The ladder program is divided into processes A to D, and a part of the divided processes A to D is requested to the programmable controller 2. Here, even if the order of the divided processes A to D is changed, there is no influence related to the execution of the ladder program.

Now, the time when the programmable controller 1 executes the ladder program is T1, the time when the programmable controller 2 executes the ladder program is T2, and the time T1 is sufficiently larger than the time T2. Consider a case where the controller 2 is requested to execute a program.

Here, if the relationship of (T1−T2) <(TA + ΔTA) is established for the execution time TA of the process A, the programmable controller 1 requests the programmable controller 2 to perform the process A, so that the entire system is executed. Program execution time is reduced. Here, ΔTA is an overhead required for the programmable controller 1 to request the programmable controller 2 to perform the process A. Specifically, this overhead is the time required to make the input information related to the process A available to the programmable controller 2 and to reflect the output information of the process A by the programmable controller 2 in the programmable controller 1. is there.

When the above conditions are satisfied, it is possible to request the process A from the programmable controller 1 to the programmable controller 2.

FIG. 4 shows a processing program by the programmable controller 1 and the programmable controller 2 when the programmable controller 1 requests the programmable controller 2 for the processing A. That is, since the process A of the programmable controller 1 is requested to the programmable controller 2,
The processing by the programmable controller 1 is processing B to
The processing becomes C, and the processing by the programmable controller 2 becomes processing A, so that the load distribution of the entire system can be achieved.

In the above description, the case where the process A of the programmable controller 1 is requested to the programmable controller 2 is shown, but the process other than the process A of the programmable controller 1, that is, the processes B to C are requested to the programmable controller 2. You may do it. Here, which of the processes that can be requested is requested,
The overhead caused by the request for this processing is determined to be as small as possible. That is, when requesting processing, the one with the least overhead, that is,
It is an effective method from the viewpoint of effective use of system resources to newly select the input / output information that must be shared between the programmable controller 1 and the programmable controller 2 when requesting processing.

After requesting the process A from the programmable controller 1 to the programmable controller 2,
When the load of the programmable controller 2 is still smaller than that of the programmable controller 1, the programmable controller 1 can request the programmable controller 2 to perform other processing.

Also in this case, a process having a small amount of input / output information that needs to be shared between the programmable controller 1 and the programmable controller 2 is selected, and the programmable controller 1 requests this process from the programmable controller 2. That is, programmable controller 1 to programmable controller 2
Among the processes that can be requested to, a process that requires less new information exchange will be requested. For example, if the input / output information used in the process A is also used in the process C, the amount of information that requires new information exchange is reduced by requesting the process C.

By repeating the above processing, programmable controller 1 and programmable controller 2
The load can be made uniform between the two, and the efficiency of the entire system can be improved.

In the above description, two programmable controllers (PLC1 and PLC2) are used.
High speed network (LAN) 100-1 and 100-2
Although a configuration in which the programmable controller system is connected to the high-speed network (LAN) 200 is considered in general when configuring a programmable controller system, the number of programmable controllers (PLC) connected to the high-speed network (LAN) 200 is not two as shown in FIG. Not exclusively.

Therefore, next, as shown in FIG.
Consider a case where C1 to PLC4) 100-1 to 100-4 are connected.

In this case, there are two possible timings for distributing the program load. 1) Each programmable controller (PLC1 to PLC
4) After creating the programs 100-1 to 100-4, load distribution when the system is not operating (hereinafter, this is referred to as static load distribution). 2) Each programmable controller (PLC1 to PLC
4) After creating the programs 100-1 to 100-4, load distribution during system operation (hereinafter referred to as dynamic load distribution).

Here, for the static load distribution of the above 1), the program is appropriately changed by a programming tool or the like to each programmable controller (PLC1 to PL).
C4) It can be dispersed in 100-1 to 100-4.

Further, in the case of the dynamic load distribution of 2) above, each programmable controller (PLC1 to PLC)
4) 100-1 to 100-4 determine how much load their programmable controller has relative to the entire system when the system is operating, and if the load of its programmable controller is from the entire system If the number is relatively large as viewed, processing is requested to another relatively small programmable controller.

FIG. 6 shows a configuration for enabling the dynamic load distribution. In the configuration shown in FIG. 6, the four programmable controllers (PLC1 to P
LC4) A memory 300 shared by all of 100-1 to 100-4 is provided, and 1) instruction execution time in PLC1 is provided in a memory area of this memory 300 2) instruction execution time in PLC2 3) instruction execution time in PLC3 4) Instruction execution time in PLC 4 5) Average instruction time of all PLCs in the network 6) PLC number with the least load in the network is stored, and each programmable controller (PLC1 to PLC1
PLC4) 100-1 to 100-4 are connected to this memory 30.
By referring to the above information stored in the memory area of 0, the relative load of the programmable controller as viewed from the entire system is determined.

The instruction execution time in each of the PLCs 1) to 4) stored in the memory area of the memory 300 corresponds to the programmable controller (PL).
C1 to PLC4) 100-1 to 100-4 are the memory 30
Dynamically write to memory area 0.

That is, the instruction execution time in PLC1 is
The programmable controller (PLC1) 100-1 writes, and similarly, the instruction execution time in the PLC2 is
The instruction execution time in LC3 is the programmable controller (PLC3) 100-3, and the instruction execution time in PLC4 is the programmable controller (PLC4) 100-3.
-4 respectively writes in the corresponding memory areas of the memory 300.

The 5) "average command time of all PLCs in the network" and the 6) "PLC number with the least load in the network" are the programmable controllers determined to have the least load in the network. (P
LC3) 100-3 calculates and writes to the corresponding memory area of memory 300.

The calculation of 5) "average command time of all PLCs in the network" and 6) "PLC number with the least load in the network" and writing to the corresponding memory area of the memory 300 are performed by the network. The programmable controller (PLC3) 100-3 that is determined to have the least load among them is configured to perform this processing, but this processing is performed by another programmable controller (PL).
It may be configured to be performed by C). That is, in this embodiment, the calculation of 5) "average command time of all PLCs in the network" and 6) "PLC number with the least load in the network" and writing to the corresponding memory area of the memory 300 are also performed. It is being processed for load balancing.

If it is determined that the relative load of the programmable controller of the own system as a whole is large by referring to the above information stored in the memory area of the memory 300, this programmable controller becomes the processing request destination. Select another programmable controller.

The selection of another programmable controller to which this processing is requested is performed by the method described below. 1) From the PLC number with the least load in the network stored in the memory area of the memory 300, the programmable controller corresponding to that PLC number is selected as another programmable controller to be the processing request destination. 2) Each PLC stored in the memory area of the memory 300
By comparing the instruction execution time at the average instruction time of all PLCs in the network and the memory area storing the average instruction time of the programmable controller of one's own in order, the average found first A programmable controller (PLC) having an instruction execution time shorter than that of the instruction processing is selected as another programmable controller to which the processing is requested. 3) A response from the programmable controller (PLC) that issues a message to all other programmable controllers (PLCs) other than itself and asks whether or not the processing can be requested. Based on the above, another programmable controller to be the request destination of this processing is selected.

Which of the above methods 1) to 3) is adopted is arbitrary in terms of system configuration. However, the method of 1) can be easily obtained by “the PLC number with the least load in the network” of 6) stored in the memory area of the memory 300 calculated by a specific programmable controller (PLC). However, when the number of programmable controllers (PLCs) in the entire system increases, a plurality of programmable controllers (PLCs) may simultaneously select the same programmable controller (PLC) as a request destination of processing. In the methods 2) and 3), the possibility that the same programmable controller (PLC) is selected as the processing request destination is low, but the programmable controller (PLC) that is the processing request destination is less likely to be selected. There is a problem that the selection process of) becomes complicated.

Next, the programmable controller (PLC) as the processing request source determines the processing to be requested. In determining the processing to be requested, a processing in which the overhead caused by this processing request is minimized is selected.

FIG. 7 is a flow chart showing the processing at the time of processing request in each programmable controller (PLC).

First, the relative load of the own PLC is calculated (step 111). This process, as described above,
It is performed based on the "average command time of all PLCs in the network" stored in the memory area of the memory 300.

If it is determined that the relative load of the own PLC is large, then the processing request destination PLC is obtained (step 11).
2). This processing request destination PLC is the above-mentioned 1) Another programmable controller which requests the programmable controller corresponding to the PLC number from the PLC number with the least load in the network stored in the memory area of the memory 300 as the processing request destination. To choose as. 2) Each PLC stored in the memory area of the memory 300
By comparing the instruction execution time at the average instruction time of all PLCs in the network and the memory area storing the average instruction time of the programmable controller of one's own in order, the average found first A programmable controller (PLC) having an instruction execution time shorter than that of the instruction processing is selected as another programmable controller to which the processing is requested. 3) A response from the programmable controller (PLC) that issues a message to all other programmable controllers (PLCs) other than itself and asks whether or not the processing can be requested. Based on the above, another programmable controller to be the request destination of this processing is selected. It is carried out by any of the following methods.

Next, the process to be requested is selected (step 114). As described above, the selection of the processing to be requested is selected in consideration of the overhead caused by the processing request.

When the processing request destination PLC and the processing to be requested are determined, a processing requesting processing for the processing is performed (step 114), and this processing ends.

As described above, the ladder program can be divided for each circuit. Here, the divided processing unit possesses an input 121, a processing 122, and an output 123, as shown in FIG.

What is important here is that necessary input information and output information can be defined for the divided processing units, and thus the overhead due to sharing of information can be calculated. In the ladder program, since the input contact and the output contact can be assigned as necessary input information and output information, the definition of the input information and the output information and the overhead due to the sharing of the information can be easily calculated.

Next, a program dividing method for a general program will be described.

1) ST Language The ST language is a language similar to FORTRAN and includes “sequential execution”, “judgment sentence”, “repeated sentence” and the like. The following shows how they can be divided.

A) "Sequential execution""Judgment" and "sequential execution" that do not include "repeat" can be arbitrarily divided as shown in FIG. That is,
"Sequential execution" that does not include "judgment" and "repetition"
It does not matter how the division is performed or which programmable controller (PLC) performs the divided processing. That is, if each process does not have a dependency relationship, the process may be divided in any way, and may be distributed in any manner.

At this time, each processing includes:
Almost all assignment processing. Therefore, this process is shown in FIG.
As shown in FIG. 4, the input 141 of the assignment statement, the process 142, and the output 143 of the assignment statement can be defined. By this, the overhead when distributing the processes can be calculated.

B) "Judgment sentence" The "judgment sentence" is, as shown in FIG.
1 and processing 152. Here, this "judgment sentence" can be considered as one processing unit. At this time, as the input / output information of the process, as shown in FIG.
It can be defined by the input to the assignment statement + the condition 161, the process 162, and the output 163 in the assignment statement, whereby the overhead when distributing the processing can be calculated.

C) "Repeat sentence""Repeatsentence" has no input / output information required for its processing, and therefore, when this is regarded as one process, no information is added.

2) SFC In SFC, "STEP" and "TRANSI" are used.
There is a framework called "TION". Then, as processing, "ACTION" and "TRANS" in "STEP"
Each of the "ITION" processes can be divided into individual processes.

The process of "TRANSITION" is "S
With the active state of "TEP" as input, "S"
Since the active state of "TEP" is output, as shown in FIG. 13, the active state 171 of STEP, TRANSITION
This can be defined by the processing 172 and the active state 173 of STEP, and thus the overhead when distributing the processing can be calculated.

In addition, "ACTION" in "STEP"
Since the execution / non-execution of the process of is obtained from the active state of its own STEP, it can be defined by the active state 181 of STEP and the ACTION process 182 in STEP as shown in FIG. The overhead can be calculated.

As described above, ST language or SF
Even in programs such as C where ordering is important,
This program can be divided into small processing units, and the division of this program into small processing units enables load distribution.

Further, the processing of the program divided as described above can be further subdivided as shown in FIG. That is, input A, processing 190, output B
The processing defined by is input A, processing 190-1 (processing 1), output (A + B + α) and input (A + B + α),
Process 190-2 (process 2) can be divided into output B. In this case, generally, only “input + output + temporary use variable” is necessary for exchanging information between the subdivided processes.

FIG. 16 shows a program status table 1 in the programmable controller system shown in FIG.
02-1 shows the details of No. 02-1. The program status table 102-1 is a table that manages programs in units of divisions.
The components of -1 consist of 1) tag number 2) status 3) input data 4) output data 5) code system to be processed.

Here, "tag number" means "each processing unit"
Is a unique number.

The "status" is "running",
It consists of three types of statuses, "Requesting" and "Requested", where "Running" indicates that this "processing unit" is being executed by this programmable controller (PLC). "Medium" indicates that this "processing unit" is being requested to another programmable controller (requested programmable controller), and "Requested" is
This "processing unit" is requested by another programmable controller (request source programmable controller).

The request-destination programmable controller is designated by the request-destination node address, and the request-source programmable controller is designated by the request-source node address.

"Input data" indicates input data required by this "processing unit", and "output data"
Indicates output data required by this “processing unit”.

The "processing code system" describes the actual processing in this "processing unit".

That is, in the program state table 102-1 shown in FIG. 16, the "processing unit" of the tag number "1" is "in execution", and the "input data" of the "processing unit" is ""A" and "B", "output data"
Indicates that it is “C”.

The "processing unit" of the tag number "2" is
“Requesting”, the requesting node address of the “processing unit” is “node Z”, and the “input data” is “A”.
And “G”.

The "processing unit" of the tag number "3" is
It indicates that "request is in progress" and the request source node address of the "processing unit" is "node W".

The "processing unit" of the tag number "4" is
"Running", and the "input data" of the "processing unit"
Indicates that it is “A”.

The "processing unit" of the tag number "5" is
“Requesting”, the requesting node address of the “processing unit” is “Node L”, and the “output data” is “G”.
And “V”.

FIG. 17 shows a shared data table 102 in the programmable controller system shown in FIG.
2 shows details of -2.

The shared data table 102-2 is a table for managing data that needs to be shared between programmable controllers. The components of the shared data table 102-2 are composed of 1) data 2) shared memory address 3) related program tag number.

Here, data to be shared is described in "data". In the “shared memory address”, the address of the shared data in the shared memory is described. Also, in the "related program tag number", the tag number of the program using this data is described.

That is, the shared data table 102-2 shown in FIG. 17 is shown in relation to the program state table 102-1 shown in FIG. 16, and in the program state table 102-1 shown in FIG. Since the “processing unit” of the tag number “2” and the tag number “5” is “requesting”, the input / output data and the tag number related to the “processing unit” of the “requesting” are described. .

Specifically, since the input / output data related to the "processing unit" of "requesting" is the data "A", "G", "V" as shown in FIG. The "shared memory address" and the "related program tag number" are described corresponding to the data "A", "G", and "V".

That is, in the shared data table 102-2 shown in FIG. 17, the address "0121" in the shared memory and the associated program tag number "2" are described in correspondence with the data "A", and the data "G". "0200" in the shared memory and the related program tag numbers "2" and "5" are described corresponding to ".", And the address in the shared memory "020" corresponding to the data "V".
1 ”and the related program tag number“ 5 ”are described.

It should be noted that from one programmable controller (PLC) to another programmable controller (PL
When requesting the processing to C), it is rewritten in the "processing code system" of the program state table 102-1 shown in FIG.

For example, in the program state table 102-1 shown in FIG. 16, when requesting the processing of the tag number "1" to another programmable controller (PLC), input data "A" and "B" are output. The part indicating the data “C” is rewritten in the “code system to be processed”.

FIG. 18 is a flowchart showing the operation of the programmable controller (PLC) that is the request source of the processing in the programmable controller system shown in FIG.

In FIG. 18, first, the process request enable flag 102-3 stored in the random access memory (RAM) 102 shown in FIG. 1 is checked (step 40).
1). Here, when the process requestable flag 102-3 is on, a value obtained by subtracting the average instruction execution time in the network from the instruction execution time of the own PLC is obtained as the "load" (step 402).

Next, it is judged whether or not the "load" obtained in step 402 is large (step 403). Specifically, the determination as to whether or not this “load” is large is step 402.
This is done by checking whether the "load" obtained in step 1 is positive or negative.

That is, when the "load" obtained in step 402 is positive, it is judged that the "load" is large (step 4
If YES, the load is judged not to be large (NO at step 403).

If it is determined that the "load" is not large (no in step 403), this programmable controller (PLC) ends this processing without requesting the processing to another programmable controller (PLC). .

If the "load" is determined to be large in step 403, the "request processing" is set to "none" and the "overhead time" is set to "maximum" (step 40).
4). Next, the "status" of the program status table 102-1 is checked to see if this programmable controller (PLC) is executing the "processing unit" (step 405).

If the programmable controller (PLC) is not executing the "processing unit", the process proceeds to step 413 which will be described later.

If the programmable controller (PLC) is executing the "processing unit" in step 405, the time (time) is set to "0" (step 406), and the data is shared data table 102. −
Check whether it is registered in No. 2 (step 40
7).

In step 407, if the data is not registered in the shared data table 102-2 (no in step 407), the corresponding data is registered in the shared data table 102-2 (step 408), and time (time) is entered.
Is added to Δt (step 409), and then step 41
Go to 0. Here, Δt is an overhead required to share one data.

When the data is registered in the shared data table 102-2 in step 407 (yes in step 407), the process proceeds to step 410 without performing the processes of steps 408 and 409.

In step 410, it is checked whether or not to repeat for all input / output data (step 410), and if to repeat for all input / output data (step 410).
Yes in step 410, the process returns to step 407, and if not repeated for all input / output data (step 410, n
o), comparing the time with the overhead time,
It is checked whether time <overhead time is satisfied (step 411).

If it is determined in step 411 that time <overhead time is satisfied (step 411
Yes), the requested process is set to the corresponding process, and the overhead time is set to time (time) (step 412), and the process proceeds to step 413.

If it is determined in step 411 that time <overhead time is not satisfied (no in step 411), the process proceeds to step 413 without performing the process of step 412.

At step 413, it is checked whether the processing unit is repeated. If the processing unit is repeated (step 413).
Yes), the process returns to step 405, and if the processing unit is not repeated (no in step 413), then the requested process is checked for "none" (step 414). If the requested process is "none" (step 414)
Yes), and the processable flag is turned off (step 4
18) and this process ends.

If it is determined in step 414 that the requested process is not "none" (no in step 414), the requested PLC is set to the least loaded PLC in the network (step 415). The code corresponding to the input / output in the processing code body is transmitted to the requested PLC (step 416).

Then, the program status table 102-
The "status" of 1 is set to "requesting", and this processing ends.

FIG. 19 is a flow chart showing the operation of the programmable controller (PLC) to which the process is requested in the programmable controller system shown in FIG. That is, in the programmable controller (PLC) to which the process is requested, the requested process is simply registered in the program status table 102-1 (step 421), and this process is terminated.

[0118]

As described above, according to the present invention,
In a load distribution method in a programmable controller system configured by connecting a plurality of programmable controllers via a network, each programmable controller calculates its own load, and at least one of its own processes corresponding to the calculated load. Since a part is configured to request another programmable controller, a load distribution device and method in a programmable controller system that can automatically distribute and control the load of each programmable controller according to the load state of each programmable controller. There is an effect that it can be provided.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration of an embodiment of a programmable controller system configured by applying a load balancing apparatus and method according to the present invention.

FIG. 2 is a block diagram showing a programmable controller system in which two programmable controllers are connected to a high speed network.

3 is a diagram showing a case where the load of execution processing of a ladder program of the programmable controller 1 is very large and the load of execution processing of a ladder program of the programmable controller 2 is very small in the programmable controller system shown in FIG.

FIG. 4 is a view showing a processing program by the programmable controller 1 and the programmable controller 2 when the programmable controller 1 requests the programmable controller 2 for processing A in the state shown in FIG. 3;

FIG. 5 is a block diagram showing a programmable controller system in which a plurality of programmable controllers are connected to a high speed network.

FIG. 6 is a diagram showing a configuration for enabling dynamic load distribution in a programmable controller system in which a plurality of programmable controllers according to the present invention are connected to a high-speed network.

FIG. 7 is a flowchart showing a process at the time of requesting a process in each programmable controller (PLC) according to the present invention.

FIG. 8 is a diagram showing divided processing units in a ladder program.

FIG. 9 is a diagram for explaining division into processing units of “sequential execution” in the ST language.

FIG. 10 is a diagram showing a divided processing unit of “sequential execution” in the ST language.

FIG. 11 is a diagram illustrating processing of a “judgment sentence” in the ST language.

FIG. 12 is a diagram showing divided processing units of a “judgment sentence” in the ST language.

FIG. 13 is a diagram showing a divided processing unit of “TRANSITION” processing in SFC.

FIG. 14: “ACTI” in “STEP” in SFC
The figure which shows the divided process unit of the process of "ON".

FIG. 15 is a diagram showing a process of further dividing the process of the divided program.

16 is a diagram showing details of a program state table shown in FIG.

17 is a diagram showing details of the shared data table shown in FIG. 1. FIG.

FIG. 18 is a flowchart for explaining the operation of the programmable controller (PLC) that is the request source of the processing in the embodiment of the present invention shown in FIG.

FIG. 19 is a flowchart for explaining the operation of the programmable controller (PLC) to which the process is requested in the embodiment of the present invention shown in FIG.

[Explanation of symbols]

100-1 to 100-n Programmable controller (PLC) 101 Read only memory (ROM) 102 Random access memory (RAM) 102-1 Program status table 102-2 Shared data table 102-3 Process requestable flag 103 Input / output memory ( I / O memory) 104 Central processing unit (CPU) 105 Network interface (I / F) 106 Internal bus 200 Network 300 Global memory in network

Claims (23)

[Claims] 1. A load distribution device in a programmable controller system configured by connecting a plurality of programmable controllers via a network, wherein each programmable controller calculates its own load, and the load calculation device. A load distribution apparatus comprising: a processing requesting unit that requests another programmable controller to perform at least a part of its own processing in accordance with the load calculated by the unit. 2. The programmable controller executes a divisible ladder program, and the processing requesting unit requests a part of the divided ladder program to be processed by another programmable controller. Item 1. A load balancer according to item 1. 3. The processing requesting means requests the processing after the ladder program is created and when the system is not operating.
The load balancer described. 4. The load balancer according to claim 2, wherein the processing requesting means requests the processing after the ladder program is created and when the system is in operation. 5. The shared storage means is further shared by the plurality of programmable controllers, and stores a command execution time of each programmable controller, an average command execution time of all programmable controllers, and a programmable controller with a minimum load. The load balancer according to claim 1, wherein the processing requesting means selects a programmable controller having a minimum load stored in the shared storage means as a processing request destination programmable controller when requesting the processing. 6. The processing requesting means, when requesting the processing, checks the instruction execution time of each programmable controller stored in the shared storage means in a predetermined order, and from the first found average instruction execution time. The load balancing apparatus according to claim 5, wherein a programmable controller having a short instruction execution time is selected as a process request destination programmable controller. 7. The processing request means, when requesting the processing, issues a message inquiring to all other programmable controllers whether or not the processing request can be made, and a response indicating that the processing request can be made is returned. The load balancer according to claim 5, wherein the programmable controller is selected as a processing request destination programmable controller. 8. The process requesting means selects a process that minimizes the overhead generated when requesting at least a part of its own process to another programmable controller and requests the other programmable controller. The load balancer according to claim 1. 9. The load balancer according to claim 8, wherein the processing requesting means selects a processing which requires a minimum amount of data to be newly shared by the programmable controller when the processing request is issued. 10. The processing requesting means comprises a program code rewriting means for rewriting a program code describing an input and an output related to the request of the processing when requesting at least a part of its own processing to another programmable controller. The load balancer according to claim 1, wherein 11. Each of the programmable controllers includes a program status table that manages a ladder program in the divided processing units, and the plurality of programmable controllers stores data shared by the plurality of programmable controllers. A shared data table to be managed is provided, and actual processing in each programmable controller is executed by referring to the program status table, and the request for the processing is executed by referring to the shared data table. 2. The load balancer according to 2. 12. The program status table is identification information for identifying each process in each process unit, for each process unit, the process unit is being executed, or the process unit is another programmable controller. Status information indicating whether or not the processing unit is being requested by another programmable controller, input data information indicating input data required by each processing unit, each processing unit requires The load distribution apparatus according to claim 11, wherein output data information indicating output data, code information describing the content of actual processing in each processing unit, are stored. 13. The shared data table stores data shared by the plurality of programmable controllers, an address of the shared data, and identification information for identifying a process using the shared data. The load balancer according to claim 11, wherein: 14. A load distribution method in a programmable controller system configured by connecting a plurality of programmable controllers via a network, wherein each programmable controller calculates its own load and corresponds to the calculated load. A load balancing method, wherein at least a part of self processing is requested to another programmable controller. 15. The programmable controller executes a divisible ladder program, and the request for the processing is made by requesting a part of processing of the divided ladder program to another programmable controller. 15. The load balancing method according to claim 14. 16. The load balancing method according to claim 15, wherein the request for the processing is made after the ladder program is created and when the system is not operating. 17. The load balancing method according to claim 15, wherein the processing request is made after the ladder program is created and when the system is in operation. 18. The plurality of programmable controllers share and store the instruction execution time of each programmable controller, the average instruction execution time of all programmable controllers, and the programmable controller with the smallest load, and request the processing described above. Is performed on the programmable controller with the least shared and stored load.
4. The load balancing method described in 4. 19. The request for the processing is to check the instruction execution times of the respective programmable controllers stored in common in a predetermined order, and the programmable controller having an instruction execution time shorter than the average instruction execution time found first. 19. The load balancing method according to claim 18, wherein is selected as a processing request destination programmable controller. 20. The processing request issues a message inquiring to all the other programmable controllers whether or not the processing request can be made, and requests the programmable controller to which a response indicating that the processing request can be sent is returned. 2. The first programmable controller is selected as the first programmable controller.
8. The load balancing method according to item 8. 21. The request for the process is characterized in that a process that minimizes the overhead that occurs when requesting at least a part of its own process to another programmable controller is requested to the other programmable controller. The load balancing method according to claim 14. 22. The load balancing method according to claim 21, wherein the request for the process selects a process that requires a minimum amount of data to be newly shared by the programmable controller when the process is requested. 23. The request for the processing is characterized in that, when requesting at least a part of its own processing to another programmable controller, a program code describing an input and an output relating to the request for the processing is rewritten. The load balancing method according to item 14.