Seyed Mohammad Ali Naseri Javareshk
Zahed Safari
Mahmoud Pourbagheri
Mahmoud Niaemanesh
Mohammad pakseresht

Distribution networks, as the most important and high-cost part of electric power systems, are responsible for supplying electricity to customers. Severe voltage drop, multiple blackouts, load imbalance and significant losses are among the obstacles of such networks. Network reconfiguration is the easiest and cost-effective method, which can relatively meet the above goals. However, it is not required to add any equipment to the network. Reconfiguration is possible by changing the state of switches in the network. This is considered as a complicated optimization problem due to various states of switches and different constraints. On the other side, the addition of Distributed Generation sources (DGs) and their impacts on the network have made this to a challenging issue. The Eligibility Traces algorithm is one of the methods to solve the problem of Reinforcement Learning. In this algorithm, the agent looks for both reconfiguration and allocation of Distributed Generation sources in the network through interactions with its environment, simultaneously. Voltage Deviation and Voltage Stability indexes are used to solve the problem faster and speed up the convergence of the proposed method. Finally, the proposed method is applied to an IEEE ۶۹-bus system under three different scenarios with the aim of minimization of network losses The supremacy of the eligibility traces algorithm is discussed and compared with Harmony Search Algorithm (HSA), Genetic Algorithms (GA) and Refined Genetic Algorithms (RGA).

Distribution networks, Voltage Deviation Index, Voltage Stability Index