In this paper an algorithm based on simulated annealing is developed to solve time optimal control problems (t-OCPs). In the beginning, a procedure that a t-OCP is converted into a nonlinear programming problem by parametrizing the control input and the time horizon is demonstrated. Then, all the unspecified parameters embedded in the converted problem, including the discretized control inputs and the associated time grids, are optimized by the proposed algorithm. For demonstrating the efficiency of our algorithm in the optimization, four typical examples are provided.