Tabu Search Based On Savings Algorithms And Direction Of Movement

In recent years, as optimization and heuristic were proposed, the study on intelligent optimization algorithms is achieving a climax at home and abroad now. Tabu Search (TS) is a novel intelligence optimization algorithm, which was formally proposed by Prof. Glover in 1986. TS has been being paid more attentions to prohibit from recurrence searching by many scholars and widely applied to combination optimization and function optimization. In this article, the author proposes an improved TS based on original TS, and applied it to practical problems.On the basis of preliminary work, the main work of this paper be summarized as follows: As the shortcomings that TS algorithm depends on the initial solution, we will use c-w algorithms to obtain initial solution. Initial solution is the sticking point of TS. The quality of initial solution restricts the convergence rate of search. Through c-w algorithms, we will get a high-quality solution, and construct a new search start point, and strengthened the focus of search efforts. In the simulation experiment we validate its effectiveness.In connection with the weak point that in the case of equally important of the intensification search and the diversification search, insufficient diversity, this paper support two kinds of strategy to improve algorithm, broad the search area, especial the unknown region. Improve the structure of TS list, define the parameters of the frequency of high-quality solutions and low-quality solution, by the two parameters, recording whether the improvements in these moving, consider such information in decision-making of the search, make the search to the better direction. Deterministic search process restricts its flexibility, determined on the basis of fixed TS, according to the information of moving, add a probability function to the evaluation function, increase the probability factors , broad the search area, makes a better solution a better chance of being selected.Traveling salesman problem (TSP) is a typical combinatorial optimization problem; people attach importance to effectively solve the TSP problem. In this paper, we support an effective form of a solution to TSP problem by the improved TS.