TY - GEN
T1 - Scalability of Condition-based Maintenance Using Federated Learning
AU - Agarwal, Vivek
AU - Manjunatha, Koushik A.
AU - Palas, Harry
N1 - Funding Information:
This research was made possible through funding from the U.S. Department of Energy’s Light Water Reactor Sustainability program under contract no. DE-AC07-05ID14517.
Publisher Copyright:
© 2023 American Nuclear Society, Incorporated.
PY - 2023
Y1 - 2023
N2 - Condition-based monitoring (CBM) techniques are being widely used for maintenance activities at nuclear power plants (NPPs). As faults are rare events, it is highly unlikely that all potential fault modes are captured for a single component. In addition, fault signatures extracted from a single component cannot be robust enough to handle unseen fault patterns. On the other hand, privacy, security, legal, and commercial concerns restrict data-sharing across different plant systems, as well as across different systems within a given plant. This research presents a federated learning (FL) approach to scaling machine learning (ML) models for CBM across plant components and systems. FL enables a centralized server to develop an aggregated global CBM model, while the training data are safely and privately distributed across the plant system. FL was demonstrated using circulating water system (CWS) data from a plant site to diagnose the health condition of a circulating water pump (CWP). The FL framework was verified using a multi-kernel adaptive support vector machine (MK-A-SVM) and an artificial neural network (NN). The results show significantly improved prediction performance, reducing overfitting issues and data heterogeneity.
AB - Condition-based monitoring (CBM) techniques are being widely used for maintenance activities at nuclear power plants (NPPs). As faults are rare events, it is highly unlikely that all potential fault modes are captured for a single component. In addition, fault signatures extracted from a single component cannot be robust enough to handle unseen fault patterns. On the other hand, privacy, security, legal, and commercial concerns restrict data-sharing across different plant systems, as well as across different systems within a given plant. This research presents a federated learning (FL) approach to scaling machine learning (ML) models for CBM across plant components and systems. FL enables a centralized server to develop an aggregated global CBM model, while the training data are safely and privately distributed across the plant system. FL was demonstrated using circulating water system (CWS) data from a plant site to diagnose the health condition of a circulating water pump (CWP). The FL framework was verified using a multi-kernel adaptive support vector machine (MK-A-SVM) and an artificial neural network (NN). The results show significantly improved prediction performance, reducing overfitting issues and data heterogeneity.
KW - artificial neural network
KW - Condition-based maintenance
KW - federated learning
KW - support vector machine
UR - https://www.scopus.com/pages/publications/85183330626
U2 - 10.13182/NPICHMIT23-41271
DO - 10.13182/NPICHMIT23-41271
M3 - Conference contribution
AN - SCOPUS:85183330626
T3 - Proceedings of 13th Nuclear Plant Instrumentation, Control and Human-Machine Interface Technologies, NPIC and HMIT 2023
SP - 1043
EP - 1052
BT - Proceedings of 13th Nuclear Plant Instrumentation, Control and Human-Machine Interface Technologies, NPIC and HMIT 2023
PB - American Nuclear Society
T2 - 13th Nuclear Plant Instrumentation, Control and Human-Machine Interface Technologies, NPIC and HMIT 2023
Y2 - 15 July 2023 through 20 July 2023
ER -