A Comparison of Particle Swarm Optimization and Mixed-Integer Nonlinear Programming Techniques for Maintenance Scheduling in Power Systems

Maintenance scheduling remains an essential part of power system operations, and finding mathematical solutions to this task is a priority for the scientific and engineering community. Its accurate and fast resolution would increase power system operation efficiency, as well as reduce the risk of energy not supplied to consumers. There are two main mathematical approaches to solving this problem: mixed-integer nonlinear programming methods and metaheuristic approaches. The well-known advantages of the latter are greater speed and the ability to avoid local optima to get closer to the global optimum. This article compares the effectiveness of these two approaches for the problem of maintenance scheduling one month ahead, considering unit commitment. The first algorithm is an example of a metaheuristic approach—the particle swarm optimization method. The second algorithm is a mixed-integer nonlinear programming algorithm based on the branch and bound method. The experiment demonstrated that, despite relatively similar optimization results, the PSO-based approach was significantly more computationally intensive. The MINLP approach completed the task in 6 hours, while the metaheuristic one took 7 days. The study shows that metaheuristics are not a one-size-fits-all solution, at least in the case of maintenance scheduling, and they may not be efficient for real-world power systems.