Three-stage multi-objective optimization approach based on sparsity knowledge for multi-airship earth-observation task scheduling

In recent years, due to the increasing demands for environmental monitoring, disaster warning, urban management, and national defense development, the construction of earth observation platforms has become a strategic requirement for countries in the aerospace sector. Autonomous airships in near space are highly competitive in terms of endurance, power consumption, and coverage area. An important aspect of earth observation is task scheduling, which generates allocation decisions under certain constraints. However, in the current literature, there is insufficient research in this field, which is crucial for human safety. Therefore, it is imperative to establish models to fill the research gap in this area. The novelties and our contributions to the paper are as follows. (i) Based on the consideration of distance, resolution, energy, and time constraints, the multi-airship earth-observation task scheduling (MAEOTS) model is developed to maximize the observation quality and minimize the energy cost. (ii) To solve this model efficiently, a three-stage multi-objective evolutionary algorithm (TSEA) based on sparsity knowledge combined with three indicators (SKTI) is proposed, also called TSEA/SKTI. (iii) Using three types of sparsity knowledge to guide the genetic operator and introducing the fitness evaluation strategy of three indicators for population selection, the competitiveness of populations in diversity and convergence is improved. (iv) A combination of decision variable score and number of tasks is used to initialize the populations. An adaptive strategy is designed to determine the range of operator crossover and mutation. The degree of crowding determined by the sharing function between individuals is used for environmental selection. The experiment was repeated in 4 data sets while comparing with 4 relevant popular evolutionary algorithms. Experimental results show that the proposed algorithm has advantages in terms of effectiveness and diversity of solutions, and demonstrates excellent characteristics in solving the MAEOTS model, exhibiting good time robustness and complexity. An actual case study in Zhuo Zhou City has been conducted to assess the effectiveness and applicability of the proposed model.