A Learning Probabilistic Boolean Network Model of a Smart Grid with Applications in System Maintenance
Name
energies-17-06399-v2.pdf
Size
4.52 MB
Format
Adobe PDF
Checksum (MD5)
959a6117e014efa6e9e8a35b57bbffb2
Author(s)
Rivera Torres, Pedro Juan
Chen, Chen
Macías-Aguayo, Jaime
Rodríguez González, Sara
Prieto Tejedor, Javier
Llanes Santiago, Orestes
García, Carlos Gershenson
Kanaan Izquierdo, Samir
Date Issued
December 19, 2024
Journal
Energies
Publisher
Multidisciplinary Digital Publishing Institute
Citation
Rivera Torres, P.J.; Chen, C.; Macías-Aguayo, J.; Rodríguez González, S.; Prieto Tejedor, J.; Llanes Santiago, O.; García, C.G.; Kanaan Izquierdo, S. A Learning Probabilistic Boolean Network Model of a Smart Grid with Applications in System Maintenance. Energies 2024, 17, 6399.
Version
Final published version
Abstract
Probabilistic Boolean Networks can capture the dynamics of complex biological systems as well as other non-biological systems, such as manufacturing systems and smart grids. In this proof-of-concept manuscript, we propose a Probabilistic Boolean Network architecture with a learning process that significantly improves the prediction of the occurrence of faults and failures in smart-grid systems. This idea was tested in a Probabilistic Boolean Network model of the WSCC nine-bus system that incorporates Intelligent Power Routers on every bus. The model learned the equality and negation functions in the different experiments performed. We take advantage of the complex properties of Probabilistic Boolean Networks to use them as a positive feedback adaptive learning tool and to illustrate that these networks could have a more general use than previously thought. This multi-layered PBN architecture provides a significant improvement in terms of performance for fault detection, within a positive-feedback network structure that is more tolerant of noise than other techniques.
MIT Department
Massachusetts Institute of Technology. Center for Transportation & Logistics
Terms of Use
Creative Commons Attribution
Persistent DSpace Link
DOI of Published Version
https://doi.org/10.3390/en17246399
