An enhanced flower pollination algorithm with superiority of feasible solution for optimal power flow problem

Keyu Zhong,Fen Xiao,Xieping Gao

DOI: https://doi.org/10.1007/s00202-024-02822-w

IF: 1.63

2024-11-16

Electrical Engineering

Abstract:The optimal power flow (OPF) problem is a nonlinear, multi-constraint and multimodal optimization problem, which is widely used in the operation and planning of modern power systems. Due to these complex characteristics, solving the OPF problem is a challenging task in power system optimization. However, most of the existing OPF methods are unconstrained optimization methods, and there are drawbacks such as insufficient search and exploration–exploitation imbalance, which easily lead the optimal solution to deviate from the security constraints of the system. For this purpose, an enhanced version of flower pollination algorithm (FPA) called EFPA is suggested to cope with the OPF problem. In EFPA, a new step size is introduced to facilitate the algorithm to adequately explore in the search space, which can increase the probability of potential high-quality candidate solutions being discovered. In addition, the global optimal individual is introduced to guide the algorithm to search for more accurate solutions toward more promising local neighborhoods. In this way, the exploration and exploitation capabilities of the algorithm can be effectively traded off. Besides, the superiority of feasible solutions (SF) constraint handling technique is also integrated to guarantee the algorithm can adaptively deal with the constraints in the OPF. The performance of EFPA is validated on CEC2021 benchmark functions, and the statistical test results demonstrate the efficiency of EFPA. EFPA is also applied to the OPF problem incorporated into thermal, wind, solar and small hydro. The experimental results show that EFPA can provide the lowest total generation cost (892.6315 $/h) compared to the state-of-the-art algorithms. In terms of voltage deviation and voltage stability, EFPA can offer the minimum deviation (0.3889 p.u.) and the best L-index (0.1437), which is reduced by 3.90% and 2.51%, respectively, compared to FPA-SF.

engineering, electrical & electronic