مريم صادقي

With the gradual transition of energy networks from traditional, passive infrastructures to intelligent, active, an‎d self-healing systems, the need for innovative solutions for fast, autonomous, an‎d effective restoration of power distribution networks has become increasingly critical. In response to this deman‎d, the present dissertation introduces two decentralized control frameworks based on multi-agent systems. The first framework employs a consensus algorithm an‎d a shared information matrix to enable fully decentralized an‎d dynamic exchange of restoration status vectors among agents. Key innovations include the design of a multi-level consensus structure with a one-by-one communication topology, an‎d an optimized weighting model based on Lyapunov eva‎luation to ensure stable convergence in the restoration process. The second framework presents an advanced parallel an‎d decentralized algorithm based on the minimum spanning tree within a mesh structure, enabling simultaneous an‎d independent execution of the restoration algorithm across structured regions of the network. The design of real-time restoration rules an‎d regional operation tables enable parallel an‎d independent execution of restoration operations with guaranteed optimal responses. This framework introduces novel concepts such as super faults, super tie switches, an‎d super meshes, offering a new paradigm for real-time management of outage zones an‎d facilitating rapid identification an‎d response to concurrent faults. Furthermore, the ability to revise restoration rules in response to structural changes significantly enhances operational flexibility, making the proposed framework well-suited for dynamic an‎d critical environments. Reducing the number of agents an‎d communication links, while increasing restoration speed, are among the key advantages of this approach. Overall, the proposed frameworks, leveraging decentralized architectures, parallel algorithms, an‎d novel management concepts, offer an effective solution for enhancing the resilience, intelligence, an‎d stability of large-scale an‎d dynamic power distribution networks.

سيستم چندعامله , اجماع چندسطحي , بازيابي نامتمركز , الگوريتم موازي درخت پوشاي كمينه , ابرخطا , ابركليد اتصال , قوانين بلادرنگ بازيابي

Multi-agent system , multi-level consensus , decentralized recovery , parallel minimum spanning tree algorithm , super fault , super tie switches , real-time restoration rules

Maryam Sadeghi

Mohsen Kalantar