An Industry Maintenance Planning Optimization Problem Using CMA-VNS and Its Variations

In this article, we consider the statement of a problem described in the competition ROADEF/EURO challenge 2020 [4] dedicated to a maintenance planning optimization problem in collaboration with RTE [5]. First of all, the main task is to build an optimal maintenance schedule for a high voltage transmission network to ensure the delivery and supply of electricity. To solve this problem we use several optimization methods to find the best possible maintenance schedule that would be consistent with all work-related constraints and would take into account the risk assessment. There are many existing heuristic, metaheuristic, and exact algorithms that can solve this optimization problem; yet it is highly feasible that the state-of-the-art methods currently available may offer some improvement. For this reason, taking into account the task at hand, we decided to use the CMA-VNS [27] algorithm after examining the results of the Combinatorial Black Box Optimization Competition [2]. The algorithm is hybrid and consists of Bipop CMA-ES and VNS algorithms that make it possible to obtain solutions with a zero penalty function and close to a minimal objective function in an admissible time. Other popular algorithms were carried out in comparison with our solution to prove the efficiency of this approach. The effectiveness of this approach to solving the problem is demonstrated.