Design and dynamic analysis of superconducting magnetic energy storage-based voltage source active power filter using deep Q-learning

The voltage source active power filter (VS-APF) is being significantly improved the dynamic performance in the power distribution networks (PDN). In this paper, the superconducting magnetic energy storage (SMES) is deployed with VS-APF to increase the range of the shunt compensation with reduced DC link voltage. The proposed SMES is characterized by the physical parameter, inductive coil, diodes and insulated gate bipolar transistors (IGBTs). The deep Q- learning (DQL) algorithm is suggested to operate SMES based VS-APF for the elimination of harmonics under different loading scenarios. Apart from this, the other benefits like improvement in power factor (PF), load balancing, potential regulation are attained. The simulation studies obtained from the proposed method demonstrates the correctness of the design and analysis compared to the VS-APF. To show the power quality (PQ) effectiveness, balanced and unbalanced loading are considered for the shunt compensation as per the guidelines imposed by IEEE-519-2017 and IEC-61000-1 grid code by using dSPACE-1104-based experimental study.