<p>With the increasing use of renewable energy, power grids require better control methods to maintain stability and efficiency. This study examines how Electric Spring (ES) performance parameters, such as its real, reactive and voltage (Pes, Qes, and Ves) change with variable load impedance of NCL and CL, using polynomial regression and an artificial neural networks for prediction and modeling. The results show that non-critical load characteristics strongly affect ES performance, revealing complex nonlinear interactions that traditional linear regression cannot capture. Polynomial regression provides consistent and reliable results, making it suitable for real-time applications due to its simple calculations and clear equations. In contrast, ANN can handle complex, nonlinear relationships but shows more variation in extreme cases and needs higher computational power. While ANN can adapt better to changing conditions, it requires careful tuning and is harder to use in real-time systems. To overcome these challenges, a hybrid approach is suggested, where PR is used for quick estimations and ANN for adaptive learning and fine-tuning. This method improves the efficiency, reliability, and adaptability of Electric Spring technology in smart grids, ensuring stable operation under different conditions. The study highlights that combining polynomial regression and an artificial neural network can be a practical solution for better power system control and voltage regulation.</p>