A Self-Learning Evolutionary Multi-Agent System for Distribution Network Reconfiguration

The paper proposes a self-learning evolutionary multi-agent system for distribution network reconfiguration. The network reconfiguration is modeled as a multi-objective combinational optimization. An autonomous agent-entity cognizes the physical aspects as operational states of the local substation, the agent-entities establish relationship network based on the interactions to provide service. Multiple objectives are considered for load balancing among the feeders, minimum deviation of the nodes voltage, minimize the power loss and branch current constraint violation. These objectives are modeled with fuzzy sets to evaluate their imprecise nature and one can provide the anticipated value of each objective. The method completes the network reconfiguration based on the negotiation of autonomous agent-entities. Simulation results demonstrated that the proposed method is effective in improving performance.