In this paper, optimization problems such that the landscape of the objective function changes over time are treated. Conventional approaches for such time-varying functions by using Evolutionary Computations are designed to track moving optimal solutions. On the contrary, the proposed method in this paper tries to find out stable solutions, i.e., robust solutions, which may not be optimal at each time step but exhibit better performance for all time steps. Such stable solutions are useful if the acquired solutions are operated by human.