Model lifecycle management for closed-loop processes within industrial automation environments

Patent metadata

Field	Value
Publication number	US-12130611-B2
Application number	US-202117484691-A
Country	US
Kind code	B2
Filing date	Sep 24, 2021
Priority date	Sep 24, 2021
Publication date	Oct 29, 2024
Grant date	Oct 29, 2024

Abstract

Various embodiments of the present technology generally relate to solutions for integrating machine learning models into industrial automation environments. More specifically, embodiments include systems and methods for implementing machine learning models within industrial control code to improve performance, increase productivity, and add capability to existing programs. In an embodiment, a system comprises: a control component configured to run a closed-loop industrial process comprises a first machine learning model; a measurement component configured to measure a gap between outcome data predicted by the first machine learning model and actual outcome data; a determination component configured to determine, based on the gap, that the first machine learning model has degraded; and a management component configured to replace the first machine learning model with a second machine learning model, wherein the second machine learning model is trained based at least in part on the actual outcome data.

First claim

What is claimed is:

1. A system for managing model lifecycles in an industrial automation environment, the system comprising:
a memory that stores executable components; and
a processor, operatively coupled to the memory, that executes the executable components, the executable components comprising:
a control component configured to run a closed-loop industrial process, in the industrial automation environment, with a control program, wherein the control program comprises a first machine learning model;
a model lifecycle management (MLM) component comprising one or more lifecycle management machine learning models trained to:
compare outcome data predicted by the first machine learning model and actual outcome data of the closed-loop industrial process; and
upon detecting that a prediction error between the outcome data predicted by the first machine learning model and the actual outcome data exceeds a degradation threshold, identify an error reduction strategy, wherein the error reduction strategy includes replacing the first machine learning model in the control program with a second machine learning model; and

the MLM component configured to replace the first machine learning model in the control program with the second machine learning model, wherein the second machine learning model is trained based at least in part on the actual outcome data of the closed-loop industrial process.

2. The system of claim 1 , wherein the first machine learning model uses, in part, real-time data from the closed-loop industrial process as input and produces parameter values used in the control program as output.

3. The system of claim 1 , wherein the second machine learning model is a retrained version of the first machine learning model.

4. The system of claim 1 , wherein the executable components further comprise an interface component configured to display a graphical representation of the first machine learning model in the industrial automation environment, wherein the graphical representation includes a visual indicator representative of the prediction error.

5. The system of claim 1 , wherein the executable components further comprise an editing component configured to, prior to detecting that the prediction error exceeds the degradation threshold, adjust the first machine learning model based on real-time data from the closed-loop industrial process.

6. The system of claim 1 , wherein the first machine learning model comprises one of a predictive model and a prescriptive model.

7. The system of claim 1 , wherein the industrial automation environment is an autonomous manufacturing environment.

8. A non-transitory computer-readable medium having stored thereon instructions for managing model lifecycles in an industrial automation environment, wherein the instructions, in response to execution, cause a system comprising a processor to perform operations, the operations comprising:
running a closed-loop industrial process, in the industrial automation environment, with a control program, wherein the control program comprises a first machine learning model;
comparing, by one or more lifecycle management machine learning models, outcome data predicted by the first machine learning model and actual outcome data of the closed-loop industrial process;
upon detecting that a prediction error between the outcome data predicted by the first machine learning model and the actual outcome data exceeds a degradation threshold, identifying, by the one or more lifecycle management machine learning models, an error reduction strategy, wherein the error reduction strategy includes replacing the first machine learning model in the control program with a second machine learning model; and
replacing the first machine learning model in the control program with the second machine learning model, wherein the second machine learning model is trained based at least in part on the actual outcome data of the closed-loop industrial process.

9. The non-transitory computer-readable medium of claim 8 , wherein the first machine learning model uses, in part, real-time data from the closed-loop industrial process as input and produces parameter values used in the control program as output.

10. The non-transitory computer-readable medium of claim 8 , wherein the second machine learning model is a retrained version of the first machine learning model.

11. The non-transitory computer-readable medium of claim 8 , the operations further comprising displaying a graphical representation of the first machine learning model in the industrial automation environment, wherein the graphical representation includes a visual indicator representative of the prediction error.

12. The non-transitory computer-readable medium of claim 8 , the operations further comprising, prior to detecting that the prediction error exceeds the degradation threshold, adjusting the first machine learning model based on output data from the closed-loop industrial process.

13. The non-transitory computer-readable medium of claim 8 , wherein the first machine learning model comprises one of a predictive model and a prescriptive model.

14. The non-transitory computer-readable medium of claim 8 , wherein the industrial automation environment is an autonomous manufacturing environment.

15. A method for managing model lifecycles comprising:
running, by a system comprising a processor, a closed-loop industrial process, in an industrial automation environment, with a control program, wherein the control program comprises a first machine learning model;
comparing, by one or more lifecycle management machine learning models on the system, outcome data predicted by the first machine learning model and actual outcome data of the closed-loop industrial process;
upon detecting that a prediction error between the outcome data predicted by the first machine learning model and the actual outcome data exceeds a degradation threshold, identifying, by the one or more lifecycle management machine learning models on the system, an error reduction strategy, wherein the error reduction strategy includes replacing the first machine learning model in the control program with a second machine learning model; and
replacing, by the system, the first machine learning model in the control program with the second machine learning model, wherein the second machine learning model is trained based at least in part on the actual outcome data of the closed-loop industrial process.

16. The method of claim 15 , wherein the first machine learning model uses, in part, real-time data from the closed-loop industrial process as input and produces parameter values used in the control program as output.

17. The method of claim 15 , wherein the second machine learning model is a retrained version of the first machine learning model.

18. The method of claim 15 , further comprising displaying, by the system, a graphical representation of the first machine learning model in the industrial automation environment, wherein the graphical representation includes a visual indicator representative of the prediction error.

19. The method of claim 15 , further comprising, prior to detecting that the prediction error exceeds the degradation threshold, adjusting, by the system, the first machine learning model based on output data from the closed-loop industrial process.

20. The method of claim 15 , wherein the industrial automation environment is an autonomous manufacturing environment.

Assignees

• Rockwell Automation Tech Inc

Inventors

• Reynolds Jordan C
• Hagerbaumer John J
• Mahr Troy W
• Sperling Jeffrey S
• Jacobsen Thomas K
• Scaturchio Giancarlo

Classifications

• G05B19/4183

  characterised by data acquisition, e.g. workpiece identification · CPC title

• G06N20/20

  Ensemble learning · CPC title

• G06N3/08

  Learning methods · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title