Zero-Order Optimization in Security Constrained Power System Scheduling
Qi An,Jianxiao Wang,Shengyu Wu,Weikai Ren,Ruanming Huang,Gengyin Li
DOI: https://doi.org/10.1109/icecet61485.2024.10698329
2024-01-01
Abstract:The security and resilience of power systems are gradually challenged. As a typical index measuring the security and resilience of power systems, the frequency stability issue is becoming prominent. Due to the increase in the complexity of power systems, the frequency stability boundary is gradually becoming fuzzy, leading to traditional analytical-modeling algorithms hardly applicable for frequency stability constrained dispatch. Existing data-driven methods for characterizing frequency stability boundary typically requires massive simulation calculations, leading to unnecessary computing power consumption. To this end, zeroth-order (ZO) optimization, which aims to solve optimization problems embedding non-analytic functions, is adopted for solving the frequency stability constrained dispatch in this paper. In contrast to traditional data-driven methods, we directly embed frequency stability simulation check into power system dispatch. The gradient of frequency stability simulation check is estimated through ZO optimization, thereby realizing the interaction optimization between power system dispatch and real-time simulation. In this way, the solution for the frequency stability constrained dispatch is modified towards the optimal point along the gradient descent direction. Cases studies based on IEEE-9 bus system verifies that, compared with traditional data-driven methods, the solution approach we propose could reduce the system cost while significantly improving the calculation efficiency.