Fault Detection And Isolation In Smart Grid Devices Using Probabilistic Boolean Networks

The area of smart power systems needs continuous improvement of its efficiency and reliability, to produce power with optimal quality in a resilient, fault-tolerant grid. Components must be highly reliable, properly maintained, and the occurrence of faults and failures has to be studied. Guaranteeing correct system operation to performance specifications involving the aforementioned activities is an active research area that applies novel methodology to the detection, classification, and isolation of faults and failures, modeling and simulating processes using predictive algorithms, with innovative AI techniques. To maintain complex power grids, predictive analytics is necessary, as employing it to plan and perform activities lowers maintenance costs and minimizes downtime. Detecting multiple faults in dynamic systems is a difficult task. Biomimetic methodologies have been applied widely in engineering systems to solve many complex problems of this field. This contribution presents a complex-adaptive bioinformatic, self-organizing framework, probabilistic Boolean networks (PBN), as a means to understand the rules that govern dynamic power systems, and to model and analyze their behavior. They are used to describe Gene Regulatory Networks, but have been recently expanding to other fields. PBNs can model system behavior, and with model checking and formal logic, assure the process' mathematical correct-ness. They enable designers, reliability and electrical engineers, and other experts to make intelligent decisions, since PBNs self-organize into attractors that model the system's operating modes, permit design for reliability, create intelligent fault diagnosis systems, assist the reliability engineering design process, and use data to analyze a system's behavior, achieving predictive maintenance.