JP2002259147A - Information processor and real time distributed processing system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an information processor for allocating tasks for realizing the function of a protection relay to computing elements while suppressing the probability of stopping the function of the protection relay. SOLUTION: The information processor allowing computing elements connected through a network to execute tasks for realizing the function of the protection relay is provided with a means for storing the combinations of tasks allowed to be executed by a specific computing element out of a plurality of tasks and a means for selecting any one of the combinations of tasks allowed to be executed by the specific computing element and allowing the specific computing element to execute the tasks of the selected combination.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a configuration of a distributed processing system in which a plurality of computers are connected on a network and executes a plurality of programs, and in particular, a real-time control represented by a power equipment protection control device and the like. Related to the processing system required.

[0002]

2. Description of the Related Art In a conventional general distributed processing system, various task allocation methods have been devised and used for the purpose of improving processing performance and reliability. For example,
JP-A-5-257908 and JP-A-7-282013 propose a method of allocating tasks according to the characteristics and status of each computer. Tasks are allocated and priorities are determined by referring to the performance such as the processing performance and reliability of each computer or the state such as the calculation load. A highly reliable distributed processing is performed as a whole by causing a computer having appropriate reliability to process tasks having different degrees of importance and urgency.

[0003] In addition to the above configuration, there is a configuration in which a cooperative processing mechanism is added as a relative relationship between tasks. For example, in Japanese Patent Application Laid-Open No. 9-231183, tasks are started simultaneously.
A management mechanism is provided to implement time constraints on task coordination. However, it does not attempt to solve the relative relationship between tasks from the viewpoint of reliability.

On the other hand, in a real-time processing apparatus requiring reliability, countermeasures by multiplexing have been conventionally taken. For example, in many control devices, such as power control such as power generation, transmission, and distribution, and traffic control such as air and railway, real-time performance is required, and prevention of malfunction is severely restricted. Therefore, in such real-time processing, high reliability is achieved by multiplexing control objects, multiplexing calculation processing, multiplexing algorithms, and the like.

[0005] Therefore, when the above object is implemented as highly reliable real-time distributed processing, there is a relative relationship from the viewpoint of reliability between the tasks to be distributed.
For example, when reliability is increased by implementing a plurality of multiplexed tasks for the same purpose, it is desirable from the viewpoint of reliability that those tasks are not processed by the same computing device. Therefore, in the highly reliable real-time distributed processing, it is necessary to consider not only the relative relationship between tasks in terms of the amount of calculation but also the relative relationship between tasks in terms of reliability.

Conventionally, when a real-time processing device requiring reliability is to be implemented in distributed processing, such a relative constraint has only been satisfied by prior design.
For example, Japanese Patent Laying-Open No. 57-3516 proposes a method in which a protection control device for electric power equipment is implemented in a distributed processing system, but adjustment of the amount of calculation and adjustment of reliability are manually performed at the time of design. . There is a need to design an optimal arrangement for each device configuration. Further, since only the predetermined task arrangement is used, the effect is lost when the apparatus configuration changes.

[0007]

As described in "Prior Art", a conventional general distributed processing system cannot be applied to a control device requiring reliability.

In the conventional distributed processing technology, it is only possible to define the reliability and urgency of each computer or task, and to improve the performance and reliability.

There is no means for defining the relative relationship between tasks in terms of reliability.

Therefore, in order to realize a highly reliable distributed processing system, in addition to the conventional relative relationship between tasks in terms of the amount of calculation, the relative relationship in terms of reliability such as the concept of multiplexing is also required. We need a mechanism that can be reflected.

An object of the present invention is to specify the reliability and urgency of each computer or task, and to improve the performance and reliability.

Another object of the present invention is to assign a task so that the function of the protection relay can be realized with a small number of arithmetic units while suppressing the possibility of stopping the function of the protection relay.

[0013]

In a real-time processing device that requires reliability, such as power control or traffic control for air or railway, multiplexing is required to maintain normal operation with respect to an abnormality in an arithmetic unit. It is necessary to realize fail-safe backup. If this processing purpose is to be realized by a distributed processing system, malfunctioning cannot be prevented only by making the system itself fault-tolerant. Implementing multiple tasks based on the concept of computational multiplexing and algorithm multiplexing provides fail-safe and
Backup needs to be implemented.

In this case, for these tasks, it is necessary to consider the relationship between the tasks. As a correlation between tasks, there are some which are restricted from being calculated by the same arithmetic unit, or some which are not appropriate. Therefore, the high-reliability real-time distributed processing system of the present invention is provided with an execution operation unit management mechanism for managing the correlation.

The execution operation unit management mechanism determines the operation unit on which each task is to be executed, and determines the urgency and importance of the execution time of each task, and the degree of exclusion of reliability, which is the interrelationship between tasks. Is managed in the database.

The execution operation unit management mechanism can determine the optimal task arrangement for the situation of the distributed processing system by referring to the rules in the database.

In consideration of the urgency and importance of the execution time of each task, the priority of the important task is maximized by maximizing the CPU usage efficiency, and the degree of exclusiveness in reliability, which is the interrelationship between tasks, is not impaired. In addition, it is possible to carry out a process of selecting a task to be performed and a process of selecting a computing unit to which the task is assigned.

Further, if task scheduling is performed on the above-mentioned distributed processing system, high reliability can always be realized systematically. Even in the event of a failure or maintenance of a computing unit, it is possible to obtain a function maintenance that achieves both a balance between CPU usage efficiency and reliability.

According to another aspect of the present invention, there is provided an information processing apparatus for processing a plurality of tasks for realizing a function of a protection relay by a plurality of arithmetic units connected via a network,
Means for storing a combination of tasks that can be processed using a specific arithmetic unit among the plurality of tasks, and one of a combination of tasks that can be processed using the specific arithmetic unit. Some have means for selecting and causing the specific arithmetic unit to execute the task of the selected combination.

According to an aspect of the present invention, there is provided an information processing apparatus for causing a plurality of arithmetic units connected via a network to execute a plurality of tasks for realizing a function of a protection relay,
Of the plurality of tasks, means for storing a combination of tasks that are not to be processed using a specific arithmetic unit, and a task of a combination excluding a combination of tasks that are not to be processed by using the specific arithmetic unit to the specific arithmetic unit. Some have a means to execute.

[0021]

Embodiments of the present invention will be described below in detail with reference to the drawings. FIG. 1 shows a first embodiment.

A plurality of arithmetic units, in this example, three arithmetic units 102 to 104 are connected to a network via a communication network 101. The arithmetic unit referred to here is a computer that executes digital arithmetic. One computing unit is powered by C
Since it has a basic configuration such as a PU and a memory independently, it is a device that does not affect the operation performance even if other operation units stop or fail. Also, since each computing unit has the same program execution environment, independent program modules can be executed by any computer. In the figure, box-shaped symbols 105 to 109 described in arithmetic units 102 to 104 abstractly represent program modules executed in each arithmetic unit. Each of the arithmetic units can execute a plurality of independent program modules at the same time.

It is required to perform a plurality of tasks by using the above-mentioned plurality of arithmetic units. Each of these tasks has an execution time constraint. In the example of FIG. 1, five tasks 105 to 109 of task 1 to task 5
Is required to be implemented.

Further, among the above-mentioned plurality of tasks, as the correlation between the tasks, there are those which are restricted from being calculated by the same arithmetic unit, or those which are not appropriate. For example, in the case where one function is implemented by taking the AND of the calculation results of multiplexed tasks in order to prevent malfunction, calculating these tasks with the same arithmetic unit is not convenient for reliability. , Undesirable. Hereinafter, this is referred to as an inter-task constraint condition in terms of reliability.

The execution operation unit management mechanisms 110 to 112 mounted on one or a plurality of devices perform processing for allocating tasks in consideration of the rules of the inter-task constraint condition in terms of reliability. . As a means of realizing this,
There are several ways.

First, as shown in FIG. 1, the execution operation unit management mechanisms 110 to 112 can be realized by a configuration holding a database. The execution operation unit management mechanism acquires in advance the inter-task constraint condition in terms of reliability for each task, and stores this as the database contents 113.

The contents of the database describe the execution time constraints and the inter-task constraints on reliability for each task. In FIG. 1, the description 114 regarding the task 1 indicates an execution time evaluation index 115 regarding an execution time constraint that determines real-time performance, an execution computing unit evaluation index 116 regarding a relationship between the task 1 and the task 4, and the like.

According to this, the urgency and importance of the execution time of the task 1 are defined by the execution time evaluation index 115, and the degree of reliability exclusiveness with the task 4 is determined by the execution arithmetic unit evaluation index 116. Stipulated by Similarly, other tasks and descriptions of the tasks are described. That is, from the database contents 113,
The configuration is such that information that can determine the degree of urgency, importance, and reliability in the processing content composed of a plurality of tasks is obtained.

The execution operation unit management mechanisms 110 to 112 obtain information of the contents 113 of the database and determine an operation unit to execute each task. In consideration of the execution time constraints of each task, the CPU use efficiency is maximized, and the important tasks are prioritized, and the tasks to be performed are selected so that the inter-task constraints are not impaired in reliability (task selection processing). Then, a process of selecting a computing unit to which the task is assigned (computer selecting process) is performed.

In the example shown in FIG. 1, strict inter-task constraint conditions are imposed on task 1 and task 4, task 1 and task 2, task 2 and task 3, task 3 and task 5 in terms of strict reliability. Therefore, a task mounting pattern is created so that these combinations are not mounted on the same computing unit. One of the solutions is the task assignment pattern of FIG. Task 1 and task 3 are assigned to operator X, task 2 and task 4 are assigned to operator Y, and task 5 is assigned to operator Z. All of the above-mentioned constraints between tasks are satisfied in terms of reliability.

Another means for the execution arithmetic unit management mechanism to consider the rules of the inter-task constraint condition in terms of reliability will be described. It is possible for the execution operation unit management mechanism to obtain the above effects without having a database.

In the example shown in FIG. 2, each task is implemented such that the execution time evaluation index for the task and the execution operation unit evaluation index for each of the correlations between the other tasks are incorporated in the program module. The communication network 101 and the plurality of arithmetic units 102 to 104 have the same configuration as that illustrated in FIG. The difference from the embodiment of FIG. 1 is that
Tasks 201 to 205 have the index. The execution arithmetic unit management mechanisms 206 to 208 in FIG.
Is provided with a mechanism capable of acquiring, via communication, an index contained in the tasks 201 to 205. Therefore, the execution operation unit management mechanism can execute the task selection processing and the operation unit selection processing by communicating with the tasks and referring to the transmitted index as necessary.

Another method will be described as a means for the execution arithmetic unit management mechanism to consider the rules of the inter-task constraint condition in terms of reliability. The above-described effect can be obtained even if indexes for all other tasks are not built in the program module in advance. The example of FIG. 3 shows a method in which each task has a task name based on a predetermined grammar. A communication network 101 and a plurality of arithmetic units 102 to 1
04 is the same as the configuration described in FIG. The difference from the embodiment of FIG. 1 is that tasks 301 to 305 have preset task names. The execution operation unit management mechanisms 306 to 308 in FIG. 3 calculate the execution operation unit evaluation index with reference to the task names of the tasks 301 to 305. To do this, the execution operation unit management mechanism 30
6 to 308 hold predetermined grammar rules 309. The grammars 310 to 312 describe an exclusive constraint 313 in the same arithmetic unit as a relationship between two task name patterns.

The execution arithmetic unit management mechanisms 306 to 308
The task name is recognized when the program module is installed. Referring to grammar rule 309, grammar 3
From 10 to 312, a grammar with a matching task name can be extracted to obtain an inter-task constraint condition in terms of reliability. As described above, the task selection processing and the arithmetic unit selection processing can be performed in the same manner as in the embodiment of FIGS.

As shown in the embodiments of FIGS. 1 to 3, the execution operation unit management mechanism does not need to be executed by one operation unit. As a function of the OS, it is also possible to implement it in all the arithmetic units. It may be composed of a master / slave. Furthermore, it may be mounted on a computer that is intermittently connected via an external communication path without using the communication network 101, for example, a remote control console.

By introducing the execution arithmetic unit management mechanism described above, it is possible to achieve both real-time performance and reliability constraints in accordance with the configuration and capabilities of the arithmetic unit.

Next, a description will be given of a configuration for realizing a task killing function while maintaining high reliability by utilizing the present invention. It is assumed that a distributed processing system is provided with a task reproduction / operation unit movement function in addition to the task selection processing and operation unit selection processing functions of the present invention.

This will be described with reference to FIG. The example of FIG.
Based on the example, it is assumed that a failure has occurred in the arithmetic unit X. Except for a failure in the arithmetic unit X, the communication path 4
01, a plurality of computing units 402 to 404, and task 4
05 to 409, the execution arithmetic unit management mechanisms 410 to 412
It is the same as FIG.

In the example of this figure, tasks 1 to 5 are all tasks that require the same amount of computation.
In terms of importance, it is assumed that task 5 has a high priority. Further, it is assumed that only two of the tasks 1 to 5 can be executed by the two operators Y and Z.

In this state, the execution unit management mechanism 410 installed in the execution unit X or the execution execution unit management mechanisms 411 and 412 installed in the execution units Y and Z are constantly monitored, and the execution unit When a device failure of X is detected,
The execution operation unit management mechanisms 411 and 412 again execute the task selection processing and the operation unit selection processing, reflecting the device failure of the operation unit X.

FIG. 5 shows the state of the distributed processing system after re-execution. On the communication path 501, the failed arithmetic unit 502
And sound arithmetic units 503 and 504 are connected.
The execution arithmetic unit management mechanisms 505 and 506 perform task selection processing and arithmetic unit selection processing. The assignment of the three tasks 507 to 509 (task 2, task 4, task 5) does not change before the device failure. The difference from before the failure is that a task 510 is newly assigned to the device 504 as an alternative to the task 1 calculated by the device 502.

This assignment is the result of a combination that balances the individual properties of the task with the properties of the correlation between the tasks. The task selection process maximizes the importance and urgency of the entire task, and the arithmetic unit selection process maximizes the reliability of the task combination implemented in the same arithmetic unit. In the case of the example of FIG. 5, since the computation amounts of the tasks are all the same, a plurality of mounting patterns for executing the four tasks with the two computing units can be considered. However, among these mounting patterns, the mounting patterns that most satisfy the execution time evaluation and the execution arithmetic unit evaluation described in FIG. 1 are the arithmetic unit Y (task 2, task 4) and the arithmetic unit Z (task 1). , Task 5). The execution arithmetic unit management mechanism processes such a task selection and an arithmetic unit selection, and as shown in FIG.
The function of automatically assigning the alternative task 510 of the task 1 calculated in 02 to the device 504 can be realized.

Next, a configuration for realizing a function of performing task scheduling while maintaining high reliability by using the present invention will be described. The execution arithmetic unit management mechanism that performs task scheduling is to grasp the presence or absence of a task that will occur in the future from the current calculation section in addition to the functions described above. Then, the task selection processing and the computing element selection processing are performed for each of a plurality of time slices in which the processing target task occurs, and an optimal solution is obtained through the plurality of time slices, thereby implementing task scheduling.

An embodiment will be described with reference to FIGS.
FIG. 6 conceptually shows the occurrence schedules of tasks recognized by the execution arithmetic unit management mechanism, and the results of task scheduling assigned to the arithmetic units of these tasks. In the figure, a horizontal axis 601 indicates a time axis representing occurrence and processing of a task relating to task 1. An arrow 602 below the horizontal axis 601 indicates generation of a task 1 execution request. A rectangle 603 above the horizontal axis 601 indicates execution of task 1. The computing unit name inside the rectangle 603 indicates the computing unit in charge of execution.
In this case, the task 1 execution request 602 means that it has been assigned to execution 603 by the arithmetic unit X. Further, when an execution corresponding to a new execution request 604 is allocated during the execution 603, it is indicated as a rectangle 605.

The first time section 606 can be realized by the task assignment by the execution operation unit management mechanism described above. The task request generated in the time section 604 has the same contents as those in the first embodiment. That is, the configuration of the task assignment shown in FIG. 1 is obtained. That is, FIG.
The assignments of the computing units for tasks 1 to 5 are the same as those in FIG.

In the second time section 607, a request 604 for task 1 is newly generated. In response to this request,
The task scheduling by the execution arithmetic unit management mechanism plans a new task allocation in the second time slice while keeping the inter-task constraint condition on the reliability in FIG. In addition to the state of task assignment in FIG. 1, task 1 is newly assigned to the computing unit Z as in execution 605. The result is as shown in FIG. In FIG. 8, inside the computing unit Z801,
There is no inter-task constraint condition between the newly assigned task 802 (the task in the bold line in the figure) and the running task 803 in terms of reliability. As a whole, the inter-stuck constraint condition is satisfied in terms of reliability in FIG.

At the time point of the third time section 608, the task 1 execution processing 603 executed by the arithmetic unit X and the task 2 execution processing 610 executed by the arithmetic unit Y end the processing. The execution request 611 of the task 5 has been received.
Task scheduling by the execution arithmetic unit management mechanism
The task assignment of FIG. 9 is performed. Due to the problem of the processing capability of the computing unit, the task 5 can be executed only by the computing unit X901.
Alternatively, it becomes the arithmetic unit Y902. However, when the task 5 is assigned to the arithmetic unit X, the inter-task constraint condition is violated in terms of reliability in relation to the running task 903 (task 3). As a result of the assignment that gives the highest evaluation, it is selected that the arithmetic unit Y executes the new task 904 (task 5). This is because there is no problem in reliability with the running task 905 (task 4).

At the time point of the fourth time slice 609, all the execution processes of the tasks other than the execution process 612 of the task 5 executed by the arithmetic unit Y have been completed. Here, task 1
5 are received. By the same process as described above, the task scheduling by the execution arithmetic unit management mechanism performs the task assignment of FIG. In addition to the executing task 1002 (task 5) of the computing unit Y1001, five new tasks 1003 to 1007 are assigned. There is no violation of inter-task constraints on reliability.

As described above, the task scheduling can be realized by processing the task selection and the operation unit selection together with the currently executing task in each time section in which the execution request is generated.

There are a plurality of methods for performing task scheduling by allocating tasks in a plurality of time sections, depending on the mounting method of the optimization method. In the above example, task scheduling is performed sequentially along the flow of the time section. Apart from this, the task scheduling can also be executed by optimizing the task assignment in all time slices at the same time and solving a combination problem that minimizes the difference in the task assignment state between adjacent time slices.

With this task scheduling function, the distributed processing system can always achieve high reliability systematically. For example, when a request to start a new task other than those recognized by the execution operation unit management mechanism is issued from within the system, or when a new task start command is issued from outside, these changes are reflected. By re-executing task scheduling, high reliability after the change can be maintained. In addition, when a command is issued for the operation state of the operation unit via an external computing device due to a failure or maintenance of the operation unit, task scheduling is performed based on the assumption of the operation state of the operation unit according to the instruction content. By re-executing
Function maintenance that achieves both CPU usage efficiency and reliability can be obtained.

Next, as a second embodiment, a development tool to which the present invention is applied is shown in FIG.

The processing system to be developed is the same as that of the embodiment shown in FIG. A plurality of arithmetic units, in this example, three arithmetic units 1102 to 1104 are network-connected via a communication path 1101 used for control. These arithmetic units are required to execute all of the plurality of tasks 1105 to 1109 to which the execution time is restricted.

An external development computer is used to design this processing system. A development computer 1111 is connected via an external communication network 1110 that does not need to be used for control. On this development computer, the basic design of the task and the setting of various setting values are performed.

The development computer 111 includes an execution arithmetic unit management mechanism 1112 having the same functions as those described in the embodiment of FIG. Therefore, by instructing the device configuration of the processing systems 1101 to 1104 to be developed, it is possible to perform the initial assignment of a task that achieves both a balance between CPU usage efficiency and reliability.

By using such a development tool, it is possible to perform an initial design capable of executing a process of selecting a task to be executed and a process of selecting an operation unit for calculating the selected task, and realize a highly reliable real-time. Design a distributed processing system. Also, by adding / deleting / modifying tasks via an external communication network, the functions of the highly reliable real-time distributed processing system can be achieved without adding an execution operation unit management mechanism to the operation units that constitute the processing system. Many can be realized.

As a third embodiment, utilizing the present invention,
A distributed processing system having a function of monitoring a communication path will be described.

To configure the processing system, the communication path 1
A plurality of operation units, in this example, three operation units 1202 to 1204 are connected to each other via a network 201.

Here, a communication path monitoring task for verifying the communication path is added to make the processing system more reliable. As a communication path monitoring method, a plurality of tasks for performing communication via a communication path to be monitored are set, and the health of the communication path is monitored by checking the validity of each other's communication contents.

At this time, task assignment and task scheduling of these monitoring tasks are complicated.
First, in order to realize the monitoring function, it is preferable to perform not only the minimum necessary monitoring task but also the monitoring task that should be performed as much as possible. However, the execution of the task, which is the original processing purpose, must not be prevented. Second, the individual tasks constituting a set need to be executed by a different arithmetic unit from the same set of task groups. The assignment and scheduling of a monitoring task having such a plurality of restrictions can be realized using the present invention.

As in the example described in the embodiment of FIG. 1, a plurality of arithmetic units in the processing system are provided with an execution arithmetic unit management mechanism 1.
205 to 1207 are provided. These execution arithmetic unit management mechanisms manage the meaning of tasks to be executed as database contents 1208. The contents of the database describe the execution time constraint and the inter-task constraint on reliability for each task. In FIG. 12, a description 1209 relating to task 1 includes an execution time evaluation index 1210 relating to an execution time constraint that determines real-time performance, and an execution computing element evaluation index 121 relating to the relationship between task 1 and task 4.
1 and so on.

In this example, it is assumed that there are two points different from the embodiment of FIG. First, the execution time evaluation index 12 relating to the communication path monitoring task A1212
13, it is assumed that an execution time evaluation index 1215 related to the communication path monitoring task B 1214 and an execution computing unit evaluation index 1216 between tasks are added.

Second, the execution time evaluation index 1217 for the task 5 is set to have a lower execution priority than the execution time evaluation indexes of the tasks 1 to 4 and the execution time evaluation index of the monitoring task. Suppose you have That is,
Task 5 is set not to be an essential task.

Here, a function of monitoring a communication path is added as a set of two tasks. Since this monitoring function is required at a minimum, an execution priority higher than that of the task 5 is set.

The above setting is executed.
05 to 1207, the task assignment to the three computing units 1202 to 1204 becomes 122
It is implemented as 0-1225. Execution of task 5 is omitted. In addition, the arrangement of each of the other six tasks is determined while satisfying the mutual exclusion.

As described above, by making the task of monitoring the communication path a part of the management target of the execution arithmetic unit management mechanism, a highly reliable real-time distributed processing system for grasping the normality of the communication path can be realized.

Further, monitoring can be performed periodically by using task scheduling using the execution arithmetic unit management mechanism. By periodically requesting the processing of a set of monitoring tasks having a low execution priority, it is possible to achieve the highest possible reliability within the range of the surplus computing capacity of the processing system.

In the above embodiment, the monitoring of the communication path has been described. However, the monitoring function of the arithmetic unit can be implemented in a similar manner. First, as an arithmetic unit monitoring task, a test for checking a predetermined arithmetic expression is performed, and a plurality of tasks for verifying the result through communication with other multiple arithmetic units connected via a network are prepared. Then, a constraint or condition is given to the execution operation unit management mechanism so that each of the operation unit monitoring tasks is executed by a different operation unit. As a result, the computing unit monitoring task automatically receives appropriate task assignment / scheduling, and can specify an abnormal computing unit or detect a network failure.

As a fourth embodiment, consider an example in which a processing system to which the present invention is applied is used to realize a digital protection relay device for protecting one or a plurality of power devices. As shown in FIG.

First, for each of the power devices 1301 to be protected and controlled, such as buses, transformers and transmission lines, as input target devices, the voltage and current used for an accident detection function required for device protection and an observation function required for various controls. Information 1302 is extracted to input devices 1303 connected to each. Similarly, as the output target device, a control signal 1 such as a shutoff command is output for a power device 1304 to be controlled such as a circuit breaker.
305 is an output device 1306 connected to each device.
Transmitted from.

The input device 1303 and the output device 130
Reference numeral 6 denotes a plurality of arithmetic units 1308 via a communication line 1307.
To 1311. The arithmetic unit here is
It is a computer that performs digital operations. One arithmetic unit independently has basic components such as a power supply, a CPU, and a memory, so that a stop or failure of another arithmetic unit does not affect the arithmetic performance. Also, since each computing unit has the same program execution environment, independent program modules can be executed by any computer. In the figure, a plurality of arithmetic units 1308 to 131
A box-shaped symbol 1311 described in 0 is an abstract representation of a program module executed in each arithmetic unit. Each of the arithmetic units can execute one or more independent program modules simultaneously.

The execution operation unit management mechanism 1312 manages to which operation unit the execution of each program module is assigned. This is the same function as the execution arithmetic unit management mechanism of the first embodiment shown in FIG. Therefore, in the database inside the execution arithmetic unit management mechanism, the execution time constraint and the inter-task constraint on reliability are described for each task.

In the case of FIG. 13, the execution arithmetic unit management mechanism manages the contents of the database shown in FIG. It describes the execution time constraints and the inter-task constraints on reliability for each task. In the description of FIG. 14, the description 1401 of each task includes an execution time target index 1402 related to an execution time constraint that determines real-time performance, and a task execution calculator evaluation index 1403 with the other.

For example, in a group of operations for protecting the device A, a program module “device A / main / constant system” which always performs main protection operation, and immediately when the program module becomes unexecutable, The “device A / main / standby system” waiting for the switched backup needs to be executed by a different arithmetic unit. Therefore, a constraint between tasks is conditioned on the relationship between the task “device A / main / always” and the task “device A / main / standby” in terms of reliability.

Similarly, in order to prevent a malfunction, “device A
The "main / always system" and the "device A / failsafe / always system" need to be executed by different arithmetic units. Therefore, the inter-task constraint is conditioned on the relationship between the two tasks in terms of reliability. As a result of the conditioning by such a constraint on the arithmetic unit, the protection function for the device A is provided as long as the two arithmetic units on which the program modules representing the two tasks are executed do not fall into the inoperable state in the same time zone. There is no possibility of malfunction.

Further, even if a single failure occurs in the apparatus, it is possible to limit the task allocation so as to maintain higher reliability. For example, even if one of the tasks of the regular system is switched to the standby system due to a device failure, if the design is made so that the main and fail-safe tasks are executed by different arithmetic units, "Equipment A, main, A task execution computing unit evaluation index can be set as a task-to-task condition on reliability in the relationship between tasks of “always system” and “device A / fail-safe / standby system”. In this case, the exclusion condition 140
4 may be set at a lower priority than the inter-task constraint 1403 in terms of reliability.

As described above, the design as shown in FIG. 14 is performed for all the task groups related to the device A. In addition, according to the same concept, the task group for the device B is designed. The execution unit on which each program module is executed is appropriately determined by the execution unit management mechanism having this database. The result is shown in FIG.
3 is task assignment. Satisfaction of all inter-task constraints and conditions in terms of reliability cannot be realized from the number of arithmetic units. The arrangement satisfies the constraint 1403 having a high priority and the exclusion condition 1404 as much as possible. At any time, a highly reliable system can be realized by utilizing the processing capacity of the entire system.

Further, in the above embodiment, a management mechanism is provided for managing the input / output relationship. FIG.
, The input / output devices J to M are fixedly connected to the devices A to D. The input / output information of these devices is transmitted and received to and from a multi-drop communication network. The input / output relationship management mechanism 1313 manages in the database which task in the arithmetic unit uses the V and I information of which input device to transmit the control signal to which output device.

As an example, consider an operation on the system shown in FIG. 13 in which two input devices 1314 and J15 are prepared as input devices connected to the device A. The input / output relationship management mechanism has received designation 1503 that the input device J 1501 and the input device J ′ 1502 are in a mutually interchangeable relationship, based on the database contents schematically shown in FIG. Therefore, when the input / output relation management mechanism detects a failure of the input device J, each of the arithmetic units is configured to replace the input data transmitted by the input device J with the input data transmitted by the input device J ′. Make it known. Thereby, the influence of the failure of the input device J can be avoided.

In the above description, the contents of the database as shown in FIGS. 14 and 15 are implemented by designating them in a rule base at the stage of designing the processing system or at the time of maintenance. The method will be described with reference to columns.

FIG. 16 shows a data format given to the execution arithmetic unit management mechanism. The contents schematically represented in FIG. 14 can be given as text data according to a certain grammar.

In FIG. 16, the first line 1601 is a comment sentence. In the second line 1602 and thereafter, the protection function for the devices A and B is set.

First, a task group regarding protection of the device A is declared. The description in the third line 1603 declares a program module for processing the functions of the device A, the main, and the normal system under the name "module_A1" as a task for performing the normal process. The character string “device A / main / always” after the # mark is a comment sentence. Further, the description of the fourth line 1604 declares a program module for processing the functions of the device A, the main, and the standby system under the name "module_A2" as a task for performing the standby process. Similarly, a total of four tasks are declared for device A. It is assumed that the description regarding the real-time property of each task, for example, the task processing start time, end time, priority, and the like are set separately.

Next, the relationship between the always-on system and the standby system for protection of the device A is declared. In the seventh line 1605, the module
Defines that module_A2 performs backup of _A1. Hereinafter, the backup of the fail-safe function is similarly described.

At the end of the definition data for the device A, an exclusive relationship between the tasks is declared. In the ninth line 1606, the “device A / main / always” function,
The “device A / failsafe / always-on” function defines a constraint assigned to a different arithmetic unit with the highest priority. In the eleventh line 1607, it is determined that the “device A / main / always system” function and the “device A / fail safe / always system” function are desirably assigned to different arithmetic units as a condition for lowering the priority. Has defined. Hereinafter, similarly, five relationships existing between the four tasks are defined.

The text data given to the execution arithmetic unit management mechanism shown in FIG. 16 also describes the settings for the device B as well as the protection for the device A described above. In this way, it is possible to give the rule of the inter-task constraint condition in terms of reliability schematically shown in FIG.

Next, the data format given to the input / output relationship management mechanism will be described. The contents schematically represented in FIG. 15 can be given as text data as shown in FIG. 17 according to a certain grammar.

In FIG. 17, the first line 1701 is a comment sentence. In the second row 1702 and thereafter, settings related to the input device, installation related to the output device, and settings related to the input / output are performed.

First, a declaration is made about the input amount transmitted by the input device J. According to the description in the third line 1703, the name “a_bus_v1” is set to the first input amount of the input device J. Hereinafter, declaration and name setting are performed for all input amounts transmitted by the input device J.

Similarly, the output device receives a declaration of the output amount. In the 17th line and after 1704, the output device L
Is defined. In the same way as the description about the input amount,
At the eighteenth line 1705, the output amount is declared and its name is set.

Lastly, the input / output amount used by each task is set. From line 1706 on line 22,
The input / output amount of task module_A1 is being set. Arguments of the program module that implements the task are
Set in order. In the 23rd line 1707, it is defined that the input a_bus_v1 is used as the first argument. Similarly, the program module defines all necessary input / output amounts. On top of that, an alternative to the input / output amount used by the task is set.
In the twenty-eighth line 1708, the input a_ used for the first argument is
Defines that a2_bus_v1 is used instead if bus_v1 is abnormal.

The text data given to the input / output relation management mechanism shown in FIG. 17 makes it possible to give a connection between the input / output amounts schematically shown in FIG.

By introducing an expression method based on the grammar as shown in FIGS. 16 and 17, it becomes possible for a designer / operator to directly write down rules in text. In addition, by using a mechanism for outputting a text file of the above expression from a design / operation tool having a GUI, a designer / operator can obtain an environment in which design / operation can be performed more easily.

As described above, the highly reliable real-time distributed processing system having the execution operation unit management mechanism shown in the first embodiment manages the task group for realizing the digital protection relay of the power equipment, thereby achieving high reliability. Protection relay system can be realized.

There are a plurality of correlations between tasks to be managed. For example, as described in the third embodiment, as the correlation between the continuous task and the standby task for the same calculation purpose, there are constraints or conditions for executing the tasks by different arithmetic units. This realizes a digital protection relay that prevents the protection relay function from stopping when an arithmetic unit fails.

Further, a protection relay having a main detection task for realizing the main detection function of the protection relay and an accident detection task for realizing the accident detection function as a fail-safe element includes:
As a correlation between the main detection task and the accident detection task of the same protection relay, there are restrictions or conditions that can be executed by different arithmetic units. Thus, a digital protection relay in which malfunction is prevented is realized.

In addition, in a protection relay having a collective bus protection task for realizing the collective bus protection function of the bus protection relay and a divided bus protection task for realizing the divided bus protection function, the batch bus protection task for the same bus is As a correlation of the bus protection task, there are constraints or conditions that are executed by different arithmetic units. As a result, a digital protection relay that prevents malfunction of the bus protection relay is realized.

Further, a protection relay having a transmission line task for realizing a transmission line protection relay function for each of the transmission lines formed of parallel lines is provided with a transmission line task for protecting a plurality of transmission lines belonging to the same parallel line. And restrictions or conditions for execution by different arithmetic units.
As a result, a digital protection relay that minimizes the effect of a computing unit failure on protection of parallel transmission lines is realized.

A high-reliability protection relay system managed by a high-reliability real-time distributed processing system as described above will be described as the fifth embodiment with reference to the drawings.

First, an overall image of the system is shown in FIG. A plurality of arithmetic units, in this example, three arithmetic units 1802 to 1804 are network-connected via an optical communication path 1801 for realizing communication of an optical medium. The arithmetic unit here is
There are computers that perform digital operations. One arithmetic unit independently has basic components such as a power supply, a CPU, and a memory, so that a stop or failure of another arithmetic unit does not affect the arithmetic performance. Also, since each computing unit has the same program execution environment, independent program modules can be executed by any computer. Further, each arithmetic unit obtains a signal synchronized with the commercial frequency via the optical communication network 1801. Therefore, the protection relay calculation synchronized with the electrical angle of the commercial frequency can be performed.

The optical communication network 1801 is connected to not only a computing unit but also an input / output device. This embodiment is illustrated focusing on one phase of the power line. A power line 1805 through which commercial power can be supplied includes a circuit breaker 1806 and a transformer 1807.
Are installed, and each device has an input / output device 18
08 to 1811 are provided. These input / output devices are in a system form connected to the optical communication network 1801.

The input / output devices 1808 to 1811 have an analog / digital conversion processing function, a communication processing function, and an electric / optical conversion function, and are mounted on the circuit breaker 1806 and the transformer 1807. The input / output device 18
Reference numerals 08 and 1809 serve as dual communication terminals of the circuit breaker 1806, and realize communication functions for voltage / current information from the circuit breaker to the arithmetic unit and cutoff control information from the arithmetic unit to the circuit breaker. Similarly, the input / output devices 1810 and 1811 realize a communication function as a dual communication end of the transformer 1807. In addition, from the viewpoint of power supply,
From the standpoint of insulation, each input / output device and arithmetic unit are independent.

As in the above-described embodiments, the present system has an execution operation unit management mechanism 1812 and an input / output relation management mechanism 1813 mounted on any one or a plurality of operation units. I have.

In the high-reliability protection relay system managed by the high-reliability real-time distributed processing system having the above configuration, processing of a plurality of tasks having real-time properties is required. These tasks are represented in an abstract manner in FIG. Box-shaped symbols 1901 to 19 in FIG.
06 denotes a task indicated by a symbol 1901 expressing each of the tasks processed by the arithmetic units 1802 to 1804,
Hereinafter, it will be referred to as “Task1” or “TrPrMN”. Task 1 is a task that implements a main transformer differential relay element 87T operation and a sequence operation that defines the firing condition of the element.

The task indicated by reference numeral 1902 is hereinafter referred to as “Ta
sk2 ”or“ TrPrFD ”Task2
Is a task that implements a different algorithm accident detection function that fulfills the fail-safe function of Task 1 and a sequence operation that defines a firing condition for that element.

The task indicated by reference numeral 1903 is hereinafter referred to as “Ta
sk3 ”or“ TrPrMN-FT ”Task
Reference numeral 3 denotes a standby function that performs the fault-tolerant function of Task 1 described above. When the above-mentioned Task1 becomes impossible to execute, the function can be backed up immediately.
For Task 1 and the same algorithm, it waits continuously while performing the operation of the minimum part. An operation for taking in an input amount and performing a process such as a digital filter process for continuing a process that needs to retain a previous value is performed.

The task indicated by reference numeral 1904 is hereinafter referred to as “Ta
sk4 ”or“ TrPrFD-FT ”Task
Reference numeral 4 denotes a standby function that performs a fault-tolerant function of Task 2 as in Task 3.

The task indicated by the symbol 1905 is hereinafter referred to as “Ta
sk5 "or" SS-P / P '".
Task5 is the input / output device P (18 in FIG. 18) described with reference to FIG.
08) and the health of the input / output device P '(1809 in FIG. 18). For example, in the input / output device, a hardware function of superimposing a twelfth harmonic component with respect to a commercial frequency as a test signal on an electric quantity (analog value) input from a power line is added. As a result, the health of the input signal can be monitored by monitoring the inspection harmonic component included in the information transmitted to and received from the input / output device in an arbitrary arithmetic unit.

The task indicated by symbol 1906 is hereinafter referred to as “Ta
sk6 "or" SS-Q / Q '".
Task6 is, like Task5, an input / output device Q (1810 in FIG. 18) and an input / output device Q ′ (FIG.
8 out of 1811).

For these tasks, real-time properties are required. In the example of FIG.
The six tasks of Task 6 are combined with the three arithmetic units 18 in FIG.
It is required to execute the processing within a predetermined execution time by using 02 to 1804.

In FIG. 19, predetermined execution time constraints for each of these six tasks, Task 1 to Task 6, are indicated by 1907 to 1912. As in the above embodiment, in implementation, data relating to these execution time constraints is provided in advance to a database managed by the execution arithmetic unit management mechanism. The contents of each execution time constraint will be described below.

The content shown in the execution time constraint 1907 is that Task1 has an execution priority Level1, and a request for the task occurs at a time frequency of 30 deg. 30 deg. Here, the smaller the Level is, the higher the priority is. Therefore, Level 1 means that the task has the highest priority. In addition, the unit deg of the processing time limit here
Represents the electrical angle with respect to one cycle of the commercial frequency as a unit, and specifies the time step when performing the protection relay calculation synchronized with the electrical angle of the commercial frequency as described with reference to FIG.

Similarly, the execution time constraint 1908 corresponds to Task 2
Indicates that the execution priority is Level 1, the occurrence frequency is 30 deg, and the processing deadline is 30 deg. In other words, it defines that Task1 and Task2 that realize the main protection function are the highest priority tasks.

Execution time constraints 1909 and 191
0 is a task that realizes the fault tolerance of the protection function
3 and Task4 are the tasks with the highest priority next to the highest priority,
Similarly, it specifies that the occurrence frequency is 30 deg and the processing deadline is 30 deg.

On the other hand, execution time constraints 1911 and 1911
12 has the lowest execution priority by design. The occurrence frequency is 10 seconds, and the processing is
It stipulates that the task should be completed within 2 seconds. In addition, for Task 6, there is also described an aperiodic task generation rule, such as no occurrence of this task for one minute after the system is started.

In this embodiment, in addition to the execution time constraints defined in the above-described tasks, the task correlation is defined similarly to the highly reliable real-time distributed processing system embodiment. Therefore, the inter-task constraint on the reliability is required as a high reliability protection relay system.

The exclusion constraints 1913 to 1916 are defined as Task1 to Task1.
A constraint condition between tasks that depends on the relationship of Task 4 in terms of reliability is illustrated. Similar to the above embodiment,
The data related to the inter-task constraints on the reliability is provided in advance to a database managed by the execution operation unit management mechanism. Hereinafter, the contents of the restrictions between tasks in terms of reliability will be described.

The contents shown in the exclusion constraint 1913 are as follows:
This indicates that Task1 and Task3 are at the exclusive priority Level1. Here, the smaller the Level is, the higher the priority is. Therefore, the level 1 means that the exclusive arrangement of the arithmetic units is considered with the highest priority. This is because Task 3 that realizes the fault-tolerant function of Task 1 is completely meaningless unless it is processed by another computing unit.

Similarly, in the exclusive constraint 1914, Task 2 and
Task 4 specifies that exclusive placement of arithmetic units should be considered with the highest priority.

On the other hand, in the exclusion constraint 1915, Task 1
This indicates that Task2 is at the exclusion priority Level2. Implement the fail-safe function of Task1
Task should be processed by another computing unit in principle. However, since Task2 is configured with a different algorithm as an accident detection method, Task1 of the main detection
It is not meaningless to process in the same arithmetic unit. Therefore,
As much as possible, constraints between tasks are defined in terms of reliability so as to take into account the exclusive arrangement of arithmetic units.

Similarly, in the exclusive constraint 1916, Task 3
Regarding Task 4, constraints between tasks are defined in terms of reliability so that exclusive arrangement of arithmetic units is considered as much as possible.

The execution operation unit management mechanism determines the task arrangement and manages the execution command based on the contents of the execution time constraint and the reliability-to-task constraint described above. The processing steps will be described step by step. First, at the time of system startup, that is, an initial operation, the case of the present embodiment will be specifically described.

FIG. 20 is a flowchart showing the operation of the execution operation unit management mechanism at the time of system startup. First, in processing 2001, all the arithmetic units are initialized. Then, a process 2002 for grasping the state of the processing capacity of each computing device and grasping the computing power of each computing unit is performed. Next, in process 2003, the execution arithmetic unit management mechanism calls the specified contents for all tasks from the DB. In particular,
The execution time constraint and the inter-task constraint in terms of reliability described above. From among these, a process 2004 of selecting a task having a start request from the beginning of system startup is performed. In the example of FIG. 19, Tasks 1 to 5 are initial activation request tasks, but Task 6 is not.

By the above-described process 2004, the grasp of the execution time constraint and the inter-task constraint in terms of reliability on the initial activation request task is completed. Based on the data related to the constraints, in process 2005, the violation amount is minimized, that is, while the execution time constraint of each task is kept,
In order to maximize the CPU usage efficiency and prioritize important tasks, not to impair constraints between tasks in terms of reliability,
The task to be executed is selected, and the operation unit to which the task is assigned is selected. For this purpose, arithmetic processing for solving the combination optimization problem is performed, and the task arrangement is determined based on the result. When the arrangement position of each task is determined, the execution arithmetic unit management mechanism performs a process 2006 for distributing a program module (hereinafter, a task module) in an executable form of the task and a process 2007 for issuing a command to execute the distributed task module. Do.

Further, the execution arithmetic unit management mechanism performs not only the activation of the initial task but also the task arrangement in the next time section in advance, and distributes the task module. The next time section is a time section in which there is a next task generation request. The next time section in this embodiment is, as shown in FIG. 19, 30 minutes after the start in synchronization with the electrical angle since the minimum task occurrence frequency is 30 degrees.

In the process 2008, the task request in the next time section is grasped from the real-time property of the task defined by the execution time constraint described above, and the next task placement optimization process 2009 is performed in the same manner as described above. Do. Then, the arrangement of the distributed initial task module is compared with the arrangement of the next task module that has just been determined, and processing 2010 for distributing the next task module to the target arithmetic unit is performed as necessary. As a result, in the next time slice, there is no occurrence of a task due to an external factor, and if the occurrence of a task request is on schedule, the task can be immediately executed in an optimal arrangement.

FIG. 21 shows the state of the initial task allocation allocated by the above-described operation at the time of starting the execution arithmetic unit management mechanism. In the arithmetic unit X (2101 in the figure), Task1 (21 in the figure)
02) and Task5 (2103 in the figure) are arranged.
(2104 in the figure) contains Task2 (2105 in the figure) and Task3
(2106 in the figure) is arranged, and the arithmetic unit Z (2107 in the figure)
, Task4 (2108 in the figure) is arranged. With this arrangement, constraints between tasks (exclusive constraints of two tasks in this case) are shown in 1913 to 1916 in FIG.
Are all satisfied. At this stage, all requested tasks are being executed, and the entire system is in a healthy state. For example, even if a task 6 request occurs after a lapse of time, the execution operation unit management mechanism can automatically assign the task 6 to the allocation portion of the task 5 and plan the process of returning to the task 5 after completion, so that the overall system health can be improved. The state can be maintained.

When an unplanned situation occurs in the entire system, for example, when one of the computing units becomes unusable, the execution computing unit management mechanism needs to reallocate the tasks. FIG. 22 shows the processing flow.

First, the execution arithmetic unit management mechanism starts when an assumed time (30 deg in this embodiment) has elapsed since the previous execution of the execution arithmetic unit management mechanism itself. This is because the signal synchronized with the electrical angle is used as a trigger for an interrupt.
01.

After starting, the execution arithmetic unit management mechanism first executes a process 2202 for determining whether there is any change in the state of the arithmetic unit, that is, the state of the arithmetic unit, based on various monitoring results. If a device abnormality is found, the process starts to make the fault tolerant function. This will be described later.
In the following, a process performed when there is no device abnormality will be described.

First, at step 2203, the current task request is read. When a task on the present system needs to start another task as a result of its own operation, it enters a task request to the execution arithmetic unit management mechanism. For example, the main detection task of Task 1 is provided with a function for rejecting an abnormal input amount. In this case, the main detection task is configured to issue a request for precise abnormality monitoring. In other words, Task 1 is based on the condition
An sk6 request is sent to the execution operation unit management mechanism. Such a request is read by the process 2203.

It is determined whether there is a difference between the current task request and the scheduled task of the next section in the previous execution of the execution arithmetic unit management mechanism by determining whether there is an unscheduled task request 2204.
And the process branches.

If there is no unscheduled task, the task module to be executed in this time slice is the distribution agent in the previous execution of the execution arithmetic unit management mechanism. Therefore, processing
At 2205, the current task execution command is immediately transmitted.

If there is an unscheduled task, process 22
At 06, the placement is re-optimized in the executable task, that is, the task group in which the task module has been distributed to the arithmetic unit. If the execution time constraint of the unscheduled task has a high priority, the execution operation unit management mechanism may relocate the task. Whether it is actually relocated depends on inter-task constraints in terms of reliability with other distributed tasks.

In any case, the execution arithmetic unit management mechanism automatically selects the optimum arrangement in the executable task. After that, the process 2207 sends an execution command for the optimally arranged task in the executable tasks. Also, if there is more time than the task module distribution, a mechanism for distributing some tasks as supplementary tasks by processing 2208 and immediately transmitting an execution instruction to those supplementary tasks by processing 2209 is also provided. I have it.

Regardless of whether or not there is an unscheduled task, placement optimization processing is performed on the task request in the next time section by processing 2210 to 2212, and the next task module is subjected to the target operation as necessary. Execute the process of delivering to the device. This is exactly the same as the procedure of 2008 to 2010 described in FIG.

Next, a case will be described in which the apparatus abnormality is recognized in the determination processing 2202 and the fault-tolerant function is activated.

First, in the process 2213, it is searched for a task that has been executed in the apparatus in which an abnormality has occurred and whether there is a standby task (hereinafter, a fault-tolerant task) that performs a fault-tolerant function. FIG.
In the example of No. 9, [Task1] TrPrMN, [Task
3] It is specified that TrPrMN-FT is a fault-tolerant task. In this way, the fault tolerant task corresponding to the task being executed in the device at the time of occurrence of the abnormality is searched from the DB.

Here, if a fault-tolerant task exists, it is necessary to immediately switch the task from the standby processing mode to the main processing mode. Process 221
In step 4, by transmitting a master-slave mode switching command to the fault-tolerant task searched in process 2113,
Realize this. The function is maintained immediately by switching the master-slave mode.

After the switching of the master-slave mode, the tasks are rearranged. Basically, the concept is the same as the processing for the occurrence of the unscheduled task described above. This is a mechanism for selecting an optimal execution module in a combination of already arranged task modules. However, since an arithmetic unit abnormality has occurred, first, in step 2215, the arithmetic power table is recreated, and the state of the arithmetic unit known by the execution arithmetic unit management mechanism is updated. Then, by the processes 2206 to 2209 described above, optimal tasks that can be executed at present are selected, and an execution command is transmitted to these. Further, the task placement optimization in the next time section is also performed by the processes 2210 to 2212 described above.

According to the processing flow of FIG. 22 described above, when an arithmetic unit abnormality occurs, the fault tolerant function is immediately activated, and then the next time section, ie, 3
After the 0 deg step, the task arrangement is autonomously obtained and the execution is continued.

In the following, taking the case of the task arrangement shown in FIG.
This will be described with reference to FIG. The specified contents of the tasks (execution time constraints and constraints between tasks in terms of reliability) are the same as those in FIG. 19, and the task arrangement is the same as the task arrangement in FIG.

In FIG. 23, arithmetic units 2301 to 230
3 is the same as the arithmetic units 2101, 2104, and 2107 in FIG. 21, and the task arrangement is the same. here,
When the operation unit 2301 becomes unusable [Task1] TrPr
MN (2304 in the figure) and [Task5] input / output P / P 'monitoring (2305 in the figure) cannot be executed. The operation of the execution arithmetic unit management mechanism at this time will be described below with reference to the flowchart of FIG.

First, the execution arithmetic unit management mechanism that has started the interrupt in the processing 2201 judges an abnormality of the arithmetic unit X (2301 in FIG. 23) in the processing 2202. In this case, the process immediately shifts to the fault tolerant activation process. In process 2213, for [Task1] TrPrMN, the fault-tolerant task [Task3] TrPrMN-FT is executed by the arithmetic unit Y (2 in FIG. 23).
302). [Task5]
No fault-tolerant task exists for I / O P / P 'monitoring. Thereafter, immediately in process 2214,
[Task3] Instruct TrPrMN-FT to switch master-slave mode. As shown in FIG. 24, [Task3] Tr 2401 in the figure
PrMN-FT is equivalent to [Task1] TrPrMN,
The first function will be provided continuously.

Subsequently, in a process 2215, after reflecting the situation in which the computing unit X cannot be used, a process of planning whether or not there is a better task execution form in the distributed task module.
Implement 2216. Here, specifically, 2401 in the figure
[Task1] TrPrMN and 2402 [Task2] TrPrFD in task reliability violation between tasks (exclusion level 2) and execution of any operation unit [Task3] TrPrMN-FT execution time constraint violation (execution priority level 2) Is searched for a combination that can solve the problem, but in this example, the solution does not exist. Therefore, the form of the task arrangement shown in FIG. 24 is an urgent arrangement for the abnormality of the arithmetic unit X. In process 2207, task 24
An execution command is transmitted to 01 to 2403. Since there is no task that can be supplementarily executed, processes 2208 to 2208
At 209, task distribution and execution instructions are not performed.

Next, the next time section, that is, the electrical angle of 30 de
g Plan the optimal layout in the future cross section that has passed and distribute task modules. First, in a process 2210, a task request is checked. Here, task requests for Tasks 1 to 4 were grasped based on the task definition contents of FIG. Then, in process 2211, the next task arrangement optimization is performed. As a result, the solution (task arrangement) shown in FIG. 24 can be obtained.

Comparing the task arrangement of FIG. 25 with FIG. 24, the arrangement of [Task1] TrPrMN 2501 in the figure does not change, but the [Task4] TrPrFD-FT of 2502 in the figure and 2 in the figure.
503 [Task2] TrPrFD arrangement has been swapped.
As a result, the violation between tasks (exclusion level 2) due to the reliability of [Task1] TrPrMN and [Task2] TrPrFD can be eliminated. However, with respect to [Task 3] TrPrFD-FT execution time constraint violation (execution priority Level 2), no arrangement that resolves this is found. Therefore, the task arrangement is as shown in FIG.

Finally, the next task module is delivered by the process 2212. Specifically, [Task4] TrPr
FD-FT task module to computing unit Y, [Task2] TrP
The task module of rFD is distributed to the computing unit Z. As a result, in the next time section, a task module necessary for achieving the task arrangement of FIG. 25 is prepared.

If there is no change in the state of the arithmetic unit or the state of the task request in the next time slice, that is, the future slice after the electrical angle of 30 deg has elapsed, the execution arithmetic unit management mechanism sets the task 25
Optimal task processing can be continued only by transmitting an execution command to 01 to 2503.

[0150]

According to the distributed processing system of the present invention, task assignment that achieves both real-time performance and reliability is executed in accordance with the configuration and capabilities of the arithmetic unit. Further, if task scheduling is performed, the distributed processing system can always achieve systematically high reliability. Even for a process in which a task activation request is generated at any time, or even in the case of a failure or maintenance of a computing unit, it is possible to obtain a function maintenance that achieves a balance between CPU usage efficiency and reliability.

[Brief description of the drawings]

FIG. 1 is a configuration of a highly reliable real-time distributed processing system to which the present invention is applied.

FIG. 2 shows another embodiment 1 of a highly reliable real-time distributed processing system.

FIG. 3 is another embodiment 2 of the highly reliable real-time distributed processing system.

FIG. 4 shows an operation 1 when a device fails.

FIG. 5 is a second operation in the case of a device failure.

FIG. 6 is an example of task scheduling by a highly reliable real-time distributed processing system.

FIG. 7 is an example of a schematic diagram of an execution arithmetic unit management mechanism database.

FIG. 8 is a one-hour section 1 in task scheduling.

FIG. 9 is a one-hour section 2 in task scheduling.

FIG. 10 is a one-hour section 3 in task scheduling.

FIG. 11 is a design operation tool from an external communication network.

FIG. 12 shows the configuration of a distributed processing system having a communication path monitoring task.

FIG. 13 shows a configuration of a digital protection relay device to which the present invention is applied.

FIG. 14 is an example of a schematic diagram of a database of an execution arithmetic unit management mechanism in a digital protection relay.

FIG. 15 is an example of a schematic diagram of an input / output management mechanism database in the digital protection relay.

FIG. 16 shows an example of a text format of an execution arithmetic unit management mechanism database.

FIG. 17 is an example of a text format of an input / output management mechanism database.

FIG. 18 shows an example of a digital protection relay device to which the present invention is applied (system configuration).

FIG. 19 is a conceptual diagram of a task in an example of a digital protection relay device.

FIG. 20 is a startup operation flow of an execution operation unit management mechanism in an example of a digital protection relay device.

FIG. 21 shows an initial task arrangement at the time of startup in an example of a digital protection relay device.

FIG. 22 is a flowchart of a constant operation of an execution arithmetic unit management mechanism in an example of a digital protection relay device.

FIG. 23 is a diagram illustrating an assumption of an arithmetic unit abnormality in an example of a digital protection relay device.

FIG. 24 shows a fault-tolerant operation (1) in an example of a digital protection relay device.

FIG. 25 shows a fault-tolerant operation (2) in an example of the digital protection relay device.

[Explanation of symbols]

101 communication network, 102 arithmetic unit X, 103 arithmetic unit Y, 104 arithmetic unit Z, 105 task 1, 106 task 2, 107 task 3, 108 task 4, 109
... Task 5, 110: execution operation unit management mechanism, 116: execution operation unit evaluation index.

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Takanori Yokoyama 7-1-1, Omika-cho, Hitachi City, Ibaraki Prefecture Inside the Hitachi Research Laboratory, Hitachi, Ltd. (72) Inventor Tomoji Nakamura 1-1-1, Kokubuncho, Hitachi City, Ibaraki Prefecture No. 1 Hitachi, Ltd. Electric Systems Division (72) Inventor Sanyo Kido 1-1-1, Kokubuncho, Hitachi City, Ibaraki Prefecture F-term (Reference) 5B045 AA05 BB07 BB12 GG02 JJ13 JJ14 JJ22 JJ26 5B098 AA10 GA04 GC01 GD02 GD14

Claims (24)

[Claims] 1. An information processing apparatus for processing a plurality of tasks for realizing a function of a protection relay by a plurality of arithmetic units connected via a network, wherein a specific arithmetic unit is used among the plurality of tasks. Means for storing a combination of tasks that can be processed; and selecting any of a combination of tasks that can be processed using the specific arithmetic unit, and executing the selected combination of tasks in the specific operation. An information processing apparatus comprising means for causing a device to execute. 2. The method according to claim 1, wherein the plurality of tasks include a regular task that constantly performs processing included in the protection relay, and a standby task that performs processing included in the protection relay in an emergency. An information processing apparatus characterized in that the combination of tasks that can be processed using the arithmetic unit does not include the combination of the continuous task and the standby task. 3. The specific arithmetic unit according to claim 1, wherein the plurality of tasks include a main detection task for implementing a protection relay main detection function and an accident detection task for implementing a protection relay accident detection function. An information processing apparatus characterized in that a combination of tasks that can be processed by using the same does not include a main detection task and an accident detection task of the same protection relay. 4. The plurality of tasks according to claim 1, wherein the plurality of tasks include a collective bus protection task for realizing a collective bus protection function of the bus protection relay and a divided bus protection task for realizing a divided bus protection function of the bus protection relay. An information processing apparatus, wherein a combination of tasks that can be processed using the specific arithmetic unit does not include a collective bus protection task for the same bus and a divided bus protection task for the same bus. 5. The transmission apparatus according to claim 1, wherein the plurality of tasks include a transmission line task for realizing a transmission line protection relay function of each transmission line formed of parallel lines, and the plurality of tasks can be processed using the specific arithmetic unit. An information processing apparatus, wherein a combination of tasks does not include transmission line tasks for protecting a plurality of transmission lines belonging to the same parallel line. 6. An information processing apparatus for processing a plurality of tasks for realizing a function of a protection relay by a plurality of computing units connected via a network, wherein a specific computing unit is used among the plurality of tasks. An information processing apparatus comprising: means for storing a combination of tasks that are not to be processed; and means for causing the specific arithmetic unit to execute a combined task excluding a combination of tasks that are not to be processed using the specific arithmetic unit. 7. The specific task according to claim 6, wherein the plurality of tasks include a regular task that constantly performs processing included in the protection relay, and a standby task that performs processing included in the protection relay in an emergency. The information processing apparatus according to claim 1, wherein the combination of tasks that are not processed using the arithmetic unit includes a combination of the continuous task and the standby task. 8. The specific arithmetic unit according to claim 6, wherein the plurality of tasks include a main detection task for realizing a protection relay main detection function and an accident detection task for realizing a protection relay accident detection function. An information processing apparatus characterized in that a combination of tasks that are not processed by using the same includes a main detection task and an accident detection task of the same protection relay. 9. The plurality of tasks according to claim 6, wherein the plurality of tasks include a collective bus protection task for realizing a collective bus protection function of the bus protection relay and a divided bus protection task for realizing a divided bus protection function of the bus protection relay. An information processing apparatus, wherein the combination of tasks that are not processed using the specific arithmetic unit includes a collective bus protection task for the same bus and a divided bus protection task for the same bus. 10. The plurality of tasks include a transmission line task for realizing a transmission line protection relay function of each transmission line formed of a parallel line, and a combination of tasks that are not processed using the specific arithmetic unit is: An information processing apparatus including transmission line tasks for protecting a plurality of transmission lines belonging to the same parallel line. 11. A distributed processing system for executing a plurality of tasks on a plurality of arithmetic units connected by one or a plurality of networks, wherein a constraint or condition for completing the task execution within a predetermined time for each of a plurality of tasks. In executing the calculation processing to which is added, for each of a plurality or all of constraints or correlations between arbitrary plural tasks,
By performing processing to determine the degree of satisfaction of constraints or conditions that are executed by different arithmetic units, maximum or sufficient reliability is ensured within the range practicable by all arithmetic units. Reliable real-time distributed processing system. 12. The highly reliable real-time distributed processing system according to claim 11, wherein an execution time evaluation index for completing execution within a predetermined time is provided for each of the plurality of tasks, and a correlation between the plurality of tasks is provided. For each of them, an execution operation unit evaluation index for execution by a different operation unit is provided, and a process of selecting a task to be executed is performed so that the evaluation index group is maximized within a range executable by all the operation units. And a high-reliability real-time distributed processing system including an execution operation unit management mechanism for executing a selection process of an operation unit that calculates a selected task. 13. The highly reliable real-time distributed processing system according to claim 12, wherein, for each task, an execution time evaluation index for the task and an execution computing element evaluation index for each correlation between tasks other than the task are set in advance. A high-reliability real-time distributed processing system comprising an execution computing unit management mechanism that executes the task selection process and the computing unit selection process with reference to the database stored in the database. 14. A highly reliable real-time distributed processing system according to claim 12, wherein each task comprises an execution time evaluation index for the task and an execution operation unit evaluation index for each correlation between tasks other than the task. A program module that implements the task is provided in advance, and includes an execution operation unit management mechanism that executes the task selection process and the operation unit selection process with reference to an index transmitted from the task. A highly reliable real-time distributed processing system characterized by the following. 15. The high-reliability real-time distributed processing system according to claim 12, wherein, for each task, when a name for managing the task is provided, an execution time evaluation index for the task is included in a symbol indicating the task name. By including a symbol to describe and a symbol to describe the execution computing element evaluation index for each of the correlations between tasks other than the task, the execution time evaluation index and the execution computing element evaluation index are calculated from the task name given in advance. A highly reliable real-time distributed processing system characterized by having an execution arithmetic unit management mechanism for extracting. 16. A highly reliable real-time distributed processing system according to claim 12, wherein a plurality of computing devices on said network execute processing of an execution operation unit management mechanism. Real-time distributed processing system. 17. The high-reliability real-time distributed processing system according to claim 12, wherein the processing of the execution operation unit management mechanism is performed by one or a plurality of computing devices that are not always connected to the network. A highly reliable real-time distributed processing system characterized in that: 18. The highly reliable real-time distributed processing system according to claim 12, wherein the operating state of said computing unit is constantly monitored to determine that a state of any computing unit on the network has changed. Upon detection, the execution arithmetic unit management mechanism executes the task selection processing and
A highly reliable real-time distributed processing system characterized by re-executing the arithmetic unit selection processing. 19. The high-reliability real-time distributed processing system according to claim 18, wherein the execution arithmetic unit management mechanism manages the presence / absence of a task to be generated in the future from the current calculation section and the evaluation index group related to the task. A highly reliable real-time distributed processing system characterized in that the task selection processing and the computing element selection processing determine the optimal solution through a plurality of time slices and execute task scheduling. 20. The high-reliability real-time distributed processing system according to claim 18, wherein a start request of a new task not included in the contents managed by the execution operation unit management mechanism is generated, or a start instruction of the new task is issued from outside. When the execution is performed, the execution arithmetic unit management mechanism executes the task selection processing, and
A highly reliable real-time distributed processing system characterized by re-executing the arithmetic unit selection processing. 21. The high-reliability real-time distributed processing system according to claim 18, wherein, when a command is issued via an external computing device regarding a dead state of the computing unit, the execution computing unit management mechanism operates in accordance with the content of the command. The task selection processing described above, and
A highly reliable real-time distributed processing system characterized by re-executing the arithmetic unit selection processing. 22. The highly reliable real-time distributed processing system according to claim 11, wherein an execution time evaluation index for completing execution within a predetermined time is provided for each of the plurality of tasks, and a correlation between the plurality of tasks is provided. For each of them, an execution operation unit evaluation index for execution by a different operation unit is provided, and a process of selecting a task to be executed is performed so that the evaluation index group is maximized within a range executable by all the operation units. Also, a highly reliable real-time distributed processing system characterized by including a development tool capable of performing an initial design capable of executing a selection process of an arithmetic unit for calculating a selected task. 23. The highly reliable real-time distributed processing system according to claim 11, wherein said task is to verify the normality of communication contents for each of a plurality of communication paths connecting a plurality of arithmetic units. It has a set of communication path monitoring tasks, and performs processing to determine the degree of satisfaction of constraints or conditions such that individual tasks constituting the set of communication path monitoring tasks that verify the same communication path are executed by different arithmetic units. A highly reliable real-time distributed processing system characterized by grasping the normality of each communication path. 24. The high-reliability real-time distributed processing system according to claim 11, wherein, as said task, a test for verifying a predetermined arithmetic expression is performed to communicate with another plurality of arithmetic units connected by a network. A plurality of computing unit monitoring tasks that verify the results via the, and each of the computing unit monitoring tasks performs processing to determine the degree of satisfaction of constraints or conditions that are executed by different computing units. A highly reliable real-time distributed processing system characterized in that a device can be specified or a network failure can be detected.