Improving Lagrangian relaxation: an application to the scheduling of pressurized water reactor outages

摘要

This paper presents an improvement of Lagrangian relaxation techniques that are now widely used for optimizing the operational planning of power systems. It is a precise algorithm for maximizing the concave and nondifferentiable dual function: a proximal point algorithm. This algorithm is applied to an important operational planning problem: the scheduling of pressurized water reactor nuclear power plant outages. Compared to the more classical Uzawa algorithm, numerical examples show that the proximal algorithm significantly improves both the solutions and the indicators of their quality.