The traditional simulated annealing algorithm is difficult to determine the initial temperature and the cooling function as well as easy to receive the inferior solution at the same time losing the current optimal solution. In order to improve the global search capability and the computational efficiency of the simulated annealing,this paper considered to take the tabu table of tabu search algorithm into the simulated annealing to avoid the loss of the optimal solution and searching repeatedly for a solution. At the same time by determining the initial temperature according to the complexity of the function and defined the new cooling function to improve the efficiency and accuracy of the algorithm. Then it introduced predator search strategy to balance the search ability and development ability to avoid getting trapped into local optima. The results on five typical standard test functions show that the improved simulated annealing algorithm not only improves the global search ability,but also the search efficiency, search accuracy and convergence rate are better than the traditional simulated annealing algorithm.