OPTIMAL ALLOCATION OF DISTRIBUTED GENERATION IN DITRIBUTION NETWORK USING ELIGIBILITY TRACES ALGORITHM

Today, one of the most challenging issues for power distribution network is handling the growing level of load demand. This has brought different issues, including voltage reduction and losses in the distribution network. Over the past years, owing to potential advantages of DG technologies, their utilization has rapidly grown all through the world. Apart from increasing utilization of DGs, their place and size in a grid are crucial. Hence, optimal selection of DGs can result in decreased voltage profile volatility, system losses. This can particularly lead to the enhanced reliability indices, as well. To solve such problems, several intelligent approaches have been represented. In this paper, the Reinforcement Learning (RL) approach is proposed, in which the Distribution Company can be viewed as a learning agent. The Eligibility Traces algorithm is known as one of the solving methods in the Reinforcement Learning (RL) approach. The RL agent is able to make an intelligent interaction with the environment, find the optimal capacity and location of the distributed generation. In order to select the most candidate buses, Voltage Loss Sensitivity Factor (VLSF) is used. In the adopted intelligent method, the DG placement problem is studied for a network of ۶۹ buses. To reveal the effectiveness of the proposed method, results of the Eligibility Traces algorithm are discussed and compared with those of Dynamic Programming (DP), particle Swarm Optimization (PSO), Genetic Algorithms (GA) and parallel Population-Based Incremental Learning (PPBIL). Simulations have shown that the Eligibility Traces algorithm outperforms mentioned methods and opens new promising horizons for future applications in the considered field.