Acceleration in optimization using bayesian optimization for broad permutation space

The challenges in air traffic control includes various management of aircraft, vehicles, and staffs, e.g. aircraft sequencing management, grand handling management. To enhance the efficiency of airports, it is important to propose comprehensive measures for these issues. As a first step, we focused on the take-off or landing aircraft management and formulated it as a permutation-space optimization problem. The accuracy and calculation speed are critical. To solve the optimization problem reasonably, we employ the simulated annealing and pave the way to implement its variant, the quantum annealing. However, the instantaneous state is trapped in local minima in the standard simulated annealing. In addition, since the solution space is extremely broad, the efficient search in the permutation space can not be performed during the limitation of the computational time. We then use a technique of Bayesian optimization to search for the next candidate in the solution space. We will compare the pure metropolis method, a standard way to implement the simulated annealing, and the method driven by Bayesian optimization regarding solution accuracy and computation time.