Optimal power flow (OPF) remains one of the most critical optimization challenges in modern electrical grids. As power systems integrate increasing amounts of renewable energy sources alongside traditional thermal generation, the mathematical problem becomes far more complex and computationally demanding.
Researchers have introduced a new optimization method based on electromagnetic wave propagation principles to tackle this challenge. The approach, called EMWPA, reformulates the OPF problem to explicitly account for the uncertainty and variability inherent in wind and solar generation. Rather than treating renewable outputs as fixed values, the method uses probability density functions to represent their natural variability, incorporating reserve costs and penalty functions directly into the optimization objectives.
The team extended the basic algorithm to handle multiple competing objectives simultaneously—a critical capability since operators must balance generation cost minimization, emissions reduction, and system stability all at once. This multi-objective version, called MOEMWPA, uses sorting and crowding distance mechanisms to identify and preserve a diverse set of optimal trade-off solutions rather than a single answer.
Validation testing on standard IEEE test networks demonstrated strong performance. On smaller 30-bus test systems, MOEMWPA achieved the highest hypervolume indicator across all multi-objective cases, indicating superior solution quality and diversity. Crucially, the algorithm maintained its performance when scaled to a 57-bus system, whereas competing methods like multi-objective particle swarm optimization failed to converge reliably on the larger network.
Computational efficiency proved competitive, with the tri-objective test case showing the lowest execution time among all algorithms tested. Statistical analysis confirmed MOEMWPA demonstrated significant superiority over NSGA-II, a well-established benchmark algorithm, across multiple test scenarios.
These results suggest the method could provide grid operators with a practical, scalable tool for real-time and planning-level optimization in increasingly renewable-heavy power systems, where computational speed and solution quality both matter considerably.
