An improved adaptive large neighborhood search algorithm for multiple agile satellites scheduling

Abstract The multiple agile satellites scheduling problem is a time-dependent scheduling problem which is considerably more difficult than the single agile satellite scheduling problem, due to its much larger solution space. We extend the adaptive large neighborhood search (ALNS) developed for the single satellite scheduling problem to the multiple satellite case. An adaptive task assignment mechanism is introduced into the ALNS framework by defining five assignment operators. In the adaptive task assignment based ALNS (A-ALNS), the removal operators remove tasks from the current solution, the insertion operators insert tasks in the destroyed solution, and if the solution has not improved for a number of iterations, the assignment operators will reassign tasks to different satellites. These operators are selected adaptively to guide the algorithm to search the solution space efficiently. The effect of the parameters on the algorithm performance is studied in the simulation experiments, and the operators are also compared. Extensive computational results show that the proposed adaptive task assignment mechanism is more efficient than competing state-of-the-art multi-satellite processing methods. The A-ALNS metaheuristic performs effectively, handling the complexity brought by the large number of satellites and fulfilling more tasks with a good robustness.