US-20260126789-A1 - SYSTEMS, APPARATUSES, METHODS, AND COMPUTER PROGRAM PRODUCTS FOR INTELLIGENT FAULT DETECTION AND MITIGATION

Abstract

Embodiments of the present disclosure provide techniques for fault detection and mitigation. Plant data associated with one or more process assets of a process plant may be received. Fault data comprising at least one fault with at least one process asset may be generated based on the plant data and a set of conditions by applying the plant data to a fault monitoring machine learning model. Fault mitigating data for the at least one fault may be generated. The fault mitigating data may comprise one or more recommendations for minimizing potential impact of the at least one fault. Performance of one or more fault mitigation actions may be generated.

Inventors

• Junda Zhu
• Jan Zirnstein
• Andrew Harper
• Sabina Azizli
• Louis Lattanzio

Assignees

• HONEYWELL INTERNATIONAL INC.

Dates

Publication Date

20260507

Application Date

20241107

Claims (20)

1. 1 . A computer-implemented method comprising: receiving, by one or more processors, plant data associated with one or more process assets of a process plant; generating, by the one or more processors, fault data comprising at least one fault with at least one process asset of the one or more process assets based on the plant data and a set of conditions by applying the plant data to a fault monitoring machine learning model configured to: (i) identify from the plant data, one or more process measurements for one or more process parameters, (ii) apply the one or more process measurements to the set of conditions, wherein the set of conditions comprise the one or more process parameters and one or more thresholds for the one or more process parameters, and (iii) generate the fault data indicative of the at least one fault in response to determining that the one or more process measurements satisfy the set of conditions; generating, by the one or more processors, fault mitigating data for the at least one fault, wherein the fault mitigating data comprises one or more recommendations for minimizing potential impact of the at least one fault; and initiating, by the one or more processors, performance of one or more fault mitigation actions.
2. 2 . The computer-implemented method of claim 1 , wherein the one or more process measurements comprise sensor data from one or more sensors associated with the one or more process assets.
3. 3 . The computer-implemented method of claim 1 , wherein the one or more process measurements comprise laboratory data from one or more measuring devices.
4. 4 . The computer-implemented method of claim 1 , wherein the fault monitoring machine learning model is a first principles-based machine learning model.
5. 5 . The computer-implemented method of claim 4 , wherein the fault monitoring machine learning model comprise one of a decision tree algorithm, random forest algorithm, XGBoost algorithm, or CatBoost algorithm.
6. 6 . The computer-implemented method of claim 1 , further comprising: retrieving historical plant data from one or more databases; applying first principles to the historical plant data to generate training data, wherein the training data comprises a plurality of ground truth exception-based alerts; and training the fault monitoring machine learning model based on the training data and using a tree-based algorithm.
7. 7 . The computer-implemented method of claim 1 , wherein the fault monitoring machine learning model defines a fault tree, and wherein the set of one or more conditions comprise a portion of the fault tree.
8. 8 . The computer-implemented method of claim 1 , wherein initiating performance of one or more fault mitigation actions comprises transmitting instructions configured to cause modification of at least one process asset.
9. 9 . The computer-implemented method of claim 1 , wherein initiating performance of one or more fault mitigation actions comprises providing the fault data for display on a user interface of a client computing entity.
10. 10 . The computer-implemented method of claim 1 , wherein initiating performance of one or more fault mitigation actions comprises providing the fault mitigating data for display on a user interface of a client computing entity.
11. 11 . An apparatus comprising at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to: receive plant data associated with one or more process assets of a process plant; generate fault data comprising at least one fault with at least one process asset of the one or more process assets based on the plant data and a set of conditions by applying the plant data to a fault monitoring machine learning model configured to: (i) identify from the plant data, one or more process measurements for one or more process parameters, (ii) apply the one or more process measurements to the set of conditions, wherein the set of conditions comprise the one or more process parameters and one or more thresholds for the one or more process parameters, and (iii) generate the fault data indicative of the at least one fault in response to determining that the one or more process measurements satisfy the set of conditions; generate fault mitigating data for the at least one fault, wherein the fault mitigating data comprises one or more recommendations for minimizing potential impact of the at least one fault; and initiate performance of one or more fault mitigation actions.
12. 12 . The apparatus of claim 11 , wherein the one or more process measurements comprise sensor data from one or more sensors associated with the one or more process assets.
13. 13 . The apparatus of claim 11 , wherein the one or more process measurements comprise laboratory data from one or more measuring devices.
14. 14 . The apparatus of claim 11 , wherein the fault monitoring machine learning model is a first principles-based machine learning model.
15. 15 . The apparatus claim 14 , wherein the fault monitoring machine learning model comprise one of a decision tree algorithm, random forest algorithm, XGBoost algorithm, or CatBoost algorithm.
16. 16 . The apparatus of claim 11 , wherein the apparatus is further caused to: retrieve historical plant data from one or more databases; apply first principles to the historical plant data to generate training data, wherein the training data comprises a plurality of ground truth exception-based alerts; and train the fault monitoring machine learning model based on the training data and using a tree-based algorithm.
17. 17 . The apparatus of claim 11 , wherein the fault monitoring machine learning model defines a fault tree, and wherein the set of one or more conditions comprise a portion of the fault tree.
18. 18 . The apparatus of claim 11 , wherein initiating performance of one or more fault mitigation actions comprises transmitting instructions configured to cause modification of at least one process asset.
19. 19 . The apparatus of claim 11 , wherein initiating performance of one or more fault mitigation actions comprises providing the fault data for display on a user interface of a client computing entity.
20. 20 . At least one non-transitory computer-readable storage medium having computer coded instructions configured to, when executed by at least one processor: receive plant data associated with one or more process assets of a process plant; generate fault data comprising at least one fault with at least one process asset of the one or more process assets based on the plant data and a set of conditions by applying the plant data to a fault monitoring machine learning model configured to: (i) identify from the plant data, one or more process measurements for one or more process parameters, (ii) apply the one or more process measurements to the set of conditions, wherein the set of conditions comprise the one or more process parameters and one or more thresholds for the one or more process parameters, and (iii) generate the fault data indicative of the at least one fault in response to determining that the one or more process measurements satisfy the set of conditions; generate fault mitigating data for the at least one fault, wherein the fault mitigating data comprises one or more recommendations for minimizing potential impact of the at least one fault; and initiate performance of one or more fault mitigation actions.

Description

TECHNICAL FIELD The present disclosure relates, generally, to fault detection. Example embodiments provide systems, apparatuses, methods, and computer program products for intelligent fault detection and mitigation. BACKGROUND In various contexts, processing plants include various assets, such as equipment, machines, and/or devices. Applicant has discovered problems with current implementations of fault detection in processing plant systems. Through applied effort, ingenuity, and innovation, Applicant has solved many of these identified problems by developing solutions embodied in the present disclosure, which are described in detail below. BRIEF SUMMARY In accordance with one aspect of the present disclosure, a computer-implemented method for early anti-ice valve fault is provided. The computer-implemented method is executable using any of a myriad of computing device(s) and/or combinations of hardware, software, and/or firmware. In some example embodiments, an example computer-implemented method includes receiving, by one or more processors, plant data associated with one or more process assets of a process plant; generating, by the one or more processors, fault data comprising at least one fault with at least one process asset of the one or more process assets based on the plant data and a set of conditions by applying the plant data to a fault monitoring machine learning model configured to: (i) identify from the plant data, one or more process measurements for one or more process parameters, (ii) apply the one or more process measurements to the set of conditions, wherein the set of conditions comprise the one or more process parameters and one or more thresholds for the one or more process parameters, and (iii) generate the fault data indicative of the at least one fault in response to determining that the one or more process measurements satisfy the set of conditions; generating, by the one or more processors, fault mitigating data for the at least one fault, wherein the fault mitigating data comprises one or more recommendations for minimizing potential impact of the at least one fault; and initiating, by the one or more processors, performance of one or more fault mitigation actions. In some embodiments, the one or more process measurements comprise sensor data from one or more sensors associated with the one or more process assets. In some embodiments, the one or more process measurements comprise laboratory data from one or more measuring devices. In some embodiments, the fault monitoring machine learning model is a first principles-based machine learning model. In some embodiments, the fault monitoring machine learning model comprise one of a decision tree algorithm, random forest algorithm, XGBoost algorithm, or CatBoost algorithm. In some embodiments, the example method further includes retrieving historical plant data from one or more databases; applying first principles to the historical plant data to generate training data, wherein the training data comprises a plurality of ground truth exception-based alerts; and training the fault monitoring machine learning model based on the training data and using a tree-based algorithm. In some embodiments, the fault monitoring machine learning model defines a fault tree, and wherein the set of one or more conditions comprise a portion of the fault tree. In some embodiments, initiating performance of one or more fault mitigation actions comprises transmitting instructions configured to cause modification of at least one process asset. In some embodiments, initiating performance of one or more fault mitigation actions comprises providing the fault data for display on a user interface of a client computing entity. In some embodiments, initiating performance of one or more fault mitigation actions comprises providing the fault mitigating data for display on a user interface of a client computing entity. In accordance with another aspect of the present disclosure, an apparatus is provided. In some example embodiments, the apparatus comprises at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to receive plant data associated with one or more process assets of a process plant; generate fault data comprising at least one fault with at least one process asset of the one or more process assets based on the plant data and a set of conditions by applying the plant data to a fault monitoring machine learning model configured to: (i) identify from the plant data, one or more process measurements for one or more process parameters, (ii) apply the one or more process measurements to the set of conditions, wherein the set of conditions comprise the one or more process parameters and one or more thresholds for the one or more process parameters, and (iii) generate the fault data indicative of the at least one fault in response to determining that the one or more process measurement