@Article{ee.2026.080048,
AUTHOR = {Haiqing Cai, Liang Tu, Wei Chen, Wencong Wu, Qingyan Zhang, Jian Wang},
TITLE = {Reactive Power Optimization Strategy for Distribution Networks Based on Analytical Transformation of Probabilistic Power Flow Sensitivity},
JOURNAL = {Energy Engineering},
VOLUME = {},
YEAR = {},
NUMBER = {},
PAGES = {{pages}},
URL = {http://www.techscience.com/energy/online/detail/26292},
ISSN = {1546-0118},
ABSTRACT = {To address the difficulty of adapting reactive power optimization strategies in distribution networks to diverse scenarios due to source-load uncertainty, which increases the risk of over-voltage and overloads, a reactive power optimization strategy for distribution networks is proposed based on probabilistic power flow sensitivity. Firstly, considering the impact of source-load uncertainty on the dispatching strategy of distribution networks, a reactive power optimization model based on chance constraints is constructed, and the probabilistic models of random variables for voltage and branch power fluctuations in the chance constraints are respectively characterized by probabilistic power flow sensitivity. Then, the quadratic nonlinear branch safety constraints are linearized by the polygonal approximation method to decouple the state variables and random variables. Through the affine transformation of probabilistic power flow sensitivity, the probabilistic analytical expressions of voltage and branch safety chance constraints are constructed. Finally, based on the probability transformation theory, the nonlinear chance constraints in probability form are analytically transformed, and the reactive power optimization of distribution networks is converted into a solvable mixed-integer second-order cone programming problem. Through simulation verification, the proposed method in this paper can fully exploit the reactive power regulation capacity of photovoltaic power, and has obvious advantages in voltage optimization effect and computational efficiency compared with the conventional sample mean method.},
DOI = {10.32604/ee.2026.080048}
}