Enhancing Smart Grid Stability: Data-Driven Predictive Modeling in Distribution Systems

Abstract

The system's ability to retain the equilibrium state during regular and under disturbance decides the power system stability. The power system stability is highly affected by continuous load variation, voltage variation, frequency variation, power flow variation, topology and the work environment. Hence the stability analysis is made to ensure the acceptable equilibrium state throughout the operation of the power system while meeting the demand. As there has been numerous inclusion of renewable energy sources into the electric network, there occurs challenge to maintain the equilibrium level of this decentralized supply with temporary needs. So to establish this kind of scenario, a Decentralized smart grid control (DSGC) is developed. In DSGC, demand is evaluated with supply through price information and the customers are allowed to decide on usage based on Pricing. The optimal hyperparameter tuning through grid search optimization for DSGC stability prediction is presented in this paper. The local frequency provides the details on equilibrium/power balance, to match supply with demand. Using an ensemble grid search optimization approach, we examine the power grid performance on dynamic stability. Our findings imply that DSGC stability is best predicted by ensemble gradient boost machine grid search with best R2 index performance and accuracy of 93.92%.