METHOD FOR THE DISTRIBUTED CALCULATION OF COMPUTATIONAL TASKS

Abstract

The invention relates to a method for the distributed calculation of calculation tasks by means of field devices of an industrial plant, wherein a plurality of field devices are coupled to a task distribution unit by means of a data link, the field devices effect a control of the industrial plant in an operating state in each case, at least one of the field devices receives a calculation task from the task distribution unit in an idle state and changes to a calculation state in which the calculation task is processed.

Claims

1. A method for the distributed calculation of calculation tasks by means of field devices of an industrial plant, wherein a plurality of field devices are coupled to a task distribution unit by means of a data link, the field devices effect a control of the industrial plant in an operating state in each case, at least one of the field devices, receives a calculation task from the task distribution unit and changes to a calculation state in which the calculation task is processed.

2. The method in accordance with claim 1, wherein the field devices report a current idle state to the task distribution unit and/or the task distribution unit polls the field devices for a current idle state, and the task distribution unit dynamically distributes calculation tasks to field devices currently in the idle state.

3. The method in accordance with claim 1, wherein the task distribution unit distributes the calculation tasks to the field devices in dependence on the bandwidth and/or processing capacity provided by the respective field devices and/or on the provisional duration of the idle state.

4. The method in accordance with claim 1, wherein the task distribution unit provides an interface for receiving new calculation tasks, and wherein the task distribution unit maintains a queue for new calculation tasks.

5. The method in accordance with claim 4, wherein the task distribution unit splits calculation tasks into a plurality of calculation tasks for different field devices and/or distributes calculation tasks as a container.

6. The method in accordance with claim 1, wherein the calculation tasks carried out in the calculation state are performed with a lower priority on the respective field device than the control of the industrial plant in the operating state.

7. The method in accordance with claim 1, wherein at least one of the field devices is real-time capable at least in the operating state and/or at least one of the field devices is coupled to the task distribution unit by means of a real-time capable field bus.

8. The method in accordance with claim 1, wherein the calculation task is a subtask of a larger task that was sent to the task distribution unit, and/or wherein the calculation task is part of a simulation, a prediction calculation, an application of artificial intelligence, a local analysis, or a self-optimization.

9. The method in accordance with claim 1, wherein at least one of the field devices changes from the calculation state to the operating state when an event or an interrupt occurs.

10. The method in accordance with claim 1, wherein at least one of the field devices and/or the task distribution unit makes/make a prediction as to when and/or how long the field device can be in the calculation state, whereupon the task distribution unit prepares a planning for calculation tasks to be distributed in the future based on the prediction.

11. A method in accordance with claim 1, wherein at least 100 field devices, are coupled to the task distribution unit.

12. A method in accordance with claim 1, wherein a plurality of the field devices cooperate in a machine of the industrial plant, wherein the machine comprises a secondary task distribution unit that distributes calculation tasks received from the task distribution unit to the field devices of the machine.

13. The method in accordance with claim 1, wherein the calculation task is independent of the control of the industrial plant.

14. An industrial plant, comprising a plurality of field devices and a task distribution unit, wherein the field devices are coupled to the task distribution unit by means of a data link, wherein the field devices are configured to effect a control of the industrial plant in an operating state in each case, at least one of the field devices is configured to receive a calculation task from the task distribution unit in an idle state and to change to a calculation state in which the calculation task is processed.

15. The method in accordance with claim 1, wherein at least one of the field devices, in an idle state, receives the calculation task.

16. The method in accordance with claim 5, wherein the task distribution unit splits calculation tasks, which are received via the interface.

17. The method in accordance with claim 5, wherein the task distribution unit distributes calculation tasks as a container for virtualization, as a script, or as an interpreter code to the field devices.

Description

[0044] The invention will be described purely by way of example with reference to the drawings in the following. There are shown:

[0045] FIG. 1 schematically an industrial plant with a plurality of field devices and a task distribution unit;

[0046] FIG. 2 schematically an industrial plant with a plurality of machines and a primary and secondary task distribution unit; and

[0047] FIG. 3 a state diagram of a field device with the different states which the field device can assume.

[0048] FIG. 1 schematically shows an industrial plant 10, for example, a manufacturing line for a product (not shown) that comprises a plurality of field devices 12a, 12b, 12c. The field devices are each coupled to a task distribution unit 16 by means of a field bus connection 14.

[0049] The task distribution unit 16 comprises an interface in the form of an API 18. Via the API 18, the task distribution unit 16 receives calculation tasks from an external source (not shown). The calculation tasks are converted into a form suitable for the field devices 12 in the task distribution unit 16 by means of a translation layer 20. A provisioning module 22 receives the calculation tasks from the translation layer 20 and “packages”/transforms the calculation tasks into a form that can be processed by the respective field device 12. In FIG. 1, a container 24 (for the virtualization), a script 26, and an interpreter code 28 are shown by way of example. The container 24 is transmitted to the field device 12a that has a container environment 30 for executing the container 24. The script 26 is transmitted to the field device 12b that has a script engine 32. Finally, the interpreter code 28 is transmitted to an interpreter 34 of the field device 12c.

[0050] After the processing of the calculation tasks, the field devices 12 transmit the processed calculation tasks via the field bus connections 14 to the task distribution unit 16 that outputs a result of the respective calculation task via the API 18.

[0051] FIG. 2 likewise shows an industrial plant 10 that, in contrast to FIG. 1, comprises a plurality of machines 36a, 36b, 36c. A plurality of field devices 12 are present in each machine 36 (not shown in FIG. 2). Each of the machines 36 has a secondary task distribution unit 38 that receives calculation tasks from the provisioning module 22 of the (primary) task distribution unit 16 via an API 18. The function within a machine 36 corresponds to the mode of operation that was already described with respect to FIG. 1.

[0052] FIG. 3 shows a state model of a field device 12. After a start 100, the field device 12 is initially in the idle state 110. To control the industrial plant 10, the field device 12 is set into an operating state 120. If the field device 12 is currently not necessary or not completely or not predominantly necessary for the control of the industrial plant, the field device 12 returns to the idle state 110. The fact that the field device 12 is in the idle state 110 is reported to the task distribution unit 16 that then assigns a calculation task to the field device 12. The field device 12 then changes to a calculation state 130 in which the calculation task is processed. After the processing of the calculation task, the field device 12 returns to the idle state 110 and can then, for example, be supplied with a further calculation task.

[0053] The operating state 120 has a higher priority than the calculation state 130. If the field device 12 now detects an event or an interrupt that indicate that the field device is required for the control of the industrial plant, the processing of the calculation task is immediately paused and the field device transitions from the calculation state 130 directly to the operation state 120 to effect the control of the industrial plant.

[0054] Once the control of the industrial plant has been completed again, a return can be made via the idle state 110 to the calculation state 130.

[0055] By providing for the calculation state 130, processing time that is typically not used can be sensibly used to increase the efficiency of the industrial plant with respect to the power consumption. The utilization of the field devices 12 can in this respect be dynamically reacted to by the task distribution unit 16.

REFERENCE NUMERAL LIST

[0056] 10 industrial plant [0057] 12 field device [0058] 14 field bus connection [0059] 16 task distribution unit [0060] 18 API [0061] 20 translation layer [0062] 22 provisioning module [0063] 24 container [0064] 26 script [0065] 28 interpreter code [0066] 30 container environment [0067] 32 script engine [0068] 34 interpreter [0069] 36 machine [0070] 38 secondary task distribution unit [0071] 100 start [0072] 110 idle state [0073] 120 operating state [0074] 130 calculation state