Critical infrastructure systems (CISs), such as power  grids, transportation systems, communication networks and water systems are the backbone of a country’s national security and industrial prosperity. These CISs execute large numbers of workflows with very high resource requirements that can span through different systems and last for a long time. The proper functioning and synchronization of these workflows is essential since humanity’s well-being is connected to it. Because of this, the challenge of ensuring availability and reliability of these services in the face of a broad range of operating conditions is very complicated.  This paper proposes an architecture which dynamically executes a scheduling algorithm using feedback about the current status of CIS nodes. Different artificial neural networks (ANNs) were created in order to solve the scheduling problem. Their performances were compared and as the main result of this paper, an optimal ANN architecture for workflow scheduling in CISs is proposed. A case study is shown for a meter data management system with measurements from a power distribution management system in Serbia. Performance tests show that significant improvement of the overall execution time can be achieved by ANNs.

In this paper, we propose a Hybrid Genetic Algorithm for data model partitioning of power distribution network. Analytical functions are the core of Distribution Management Systems (DMSs). Efficient calculation of the functions is of the utmost importance for the DMS users; the necessary preconditions for the efficient calculation are optimal load balancing of processors and data model partitioning among processors. The proposed algorithm is applied to different real models of power distribution systems. It obtains better results than classical evolutionary algorithms (Genetic Algorithm and Particle Swarm Optimization). The Hybrid Genetic Algorithm also achieves better results than multilevel algorithm (METIS) in cases of small graphs.

No abstract

No abstract

No abstract