Optimal deployment of reactive power in a renewable energy sources integrated system with EVs demand using local randomized neural networks

The rising popularity of Electric vehicles (EV) has resulted in a substantial increase in the amount of charging stations, which extensively affects the electrical grid, causing problems like power quality degradation, voltage fluctuations and higher losses. This paper proposes the novel application of Local Randomized Neural Networks (LRNN) for optimal deployment of reactive power in a renewable energy sources integrated system with EVs demand. The main aim of the proposed work is to reduce both active and reactive power loss and maximize reliability. The LRNN method predicts the optimal location for the fast charging station. The proposed methods performance is excluded in the MATLAB working platform and compared with several existing techniques, with Genetic Algorithm (GA), Sea Horse Optimization (SHO) and Particle Swarm Optimization (PSO).The proposed technique demonstrates superior performance by significantly reducing power losses across all buses in the system. Compared to conventional optimization techniques, the LRNN achieves the lowest computational complexity at 1.82%, and the fastest convergence speed in just 25 iterations. In terms of execution time, it completes in 0.34 s, faster than the Genetic Algorithm at 0.44 s, Sea Horse Optimization at 0.59 s, and Particle Swarm Optimization at 0.65 s. While its efficiency is 98% it offers an excellent balance between computational speed, accuracy, and loss minimization. These results highlight its potential as a highly effective solution for modern power systems integrating renewable sources and electric vehicles.