An Analysis of the DG Placement for Primary Radial Distribution System Using Q-Leaning Algorithm

This paper presents an analysis of the distributed generation (DG) placement for the primary radial distribution system. The machine learning (ML) technique was applied by using the Q-leaning algorithm for optimal DG placement. The reward function was a key for investigating the exact optimal solution that consists of the total power loss and the voltage deviation index (VDI). The IEEE 33 bus testing system was used for power flow analysis of this problem. The simulation results showed that the Q-learning algorithm is capable of determining the optimal placement of DGs based on the variation in reward. The DG installation at bus number 6 (Case I) can be decreased the total energy loss. While installing the DG at bus number 9 (Case II) can be improved the VDI. Therefore, applying ML techniques for electrical power system analysis will enhance efficiency and reliability, which are important guidelines for planning improvements to the electrical power system in the future.