Automated Hyperparameter Tuned Deep Learning Enabled Reactive Power Optimization Model for Power Distribution System

With a great quantity of Electric Vehicles and Distributed Generator (DG) complied in the power distribution system, the complications of distribution systems' function are higher, which generates the superior need for online Reactive Power Optimization (RPO). The RPO is a distribution network that could enhance the quality of voltage and the economical function, and diminish the power losses of a dispersal network. RPO could understand rational dispersal of reactive power in the dispersal network and decrease the node voltage deviations and power losses. Currently, only a few heuristic intellectual methods are broadly employed for RPO. Therefore, this article introduces a new Jellyfish Search Optimization with Deep Stacked Autoencoder (JSO-DSAE) model for RRO in power distribution systems. The proposed JSO-DSAE model enables the DSAE model to receive previous data from DGs to identify the connection among power control and system characteristics. To bolster the performance of the JSO-DSAE algorithm, the JSO method is used. The experimental validation of the JSO-DSAE model is tested and the outcomes are examined over distinct aspects. The simulation outcome demonstrated the supremacy of the JSO-DSAE model over the recent approaches.