The invention discloses a task
planning method for an agile
earth satellite. The method comprises the following steps of 1, selecting a target function; 2, according to the target function in the step1, generating an initial solution x, and taking x as a starting point of local search; 3, performing parameter initialization, wherein parameters include a neighborhood number kmax, an
iterative search frequency lmin used for judging whether a process of falling into
local optimum exists or not, a maximum
iterative search frequency lmax used for judging whether an
algorithm is stopped or not, iterative counters m and n, and a neighborhood counter k, and k, m and n are set to be 1, 0 and 0 respectively; 4, randomly determining a feasible solution y in a kth neighborhood structure of the solution x, and enabling n to be equal to n+1; 5, if y is superior to x, enabling x to be equal to y and m to be equal to 0, or otherwise, enabling m to be equal to m+1; 6, if n==lmax, going to the step 9;7, if m==lmin, indicating that the process of falling into the
local optimum already exists and k=k+1, or otherwise, going to the step 4; 8, if k<=kmax, enabling m to be equal to 0, and going to the step 4; and 9, ending the
algorithm, and outputting x. According to the method, resources and time can be allocated for observation tasks or data return tasks; and on the premise of meeting task constraints and
resource constraints, the observation benefits are maximized.