The agile Earth observation satellite scheduling problem (AEOSSP) is a complex combinatorial optimization problem encompassing resource allocation and task sequencing. The objective is to optimize the satellite observation task sequences to maximize the observation profits. However, efficiently optimizing the satellite observation task scheduling has become a current problem that needs to be solved in AEOSSP. To address this challenge, we propose a simulated annealing method combined with attention model (SAM-AM). SAM-AM utilizes the framework of Simulated Annealing (SA). Initially, Attention Model (AM) is employed to acquire the initial solution, followed by iterative and continuous refinement and optimization. The results show that SAM-AM exhibits significant advantages in the case of different task scales and that AM has a significant impact on the stability of the method. SAM-AM provides an efficient solution for AEOSSP, which has considerable potential in achieving resource optimization.