Memetic algorithm based on improved inver-over operator for TSP

The Inver-over operator is always stuck to local optima in solving the Traveling Salesman Problem(TSP). In this paper, two improved Inver-over operators are proposed which contain the noise method(NM) based local search with multiple different neighboring structures. An effective memetic algorithm(MA) based on two improved Inver-over operators is implemented, which conduct different operators in different stages, and then improve the convergence speed of the proposed algorithm while maintain the popular diversification. In addition, the adaptive Meta-Lamarckian learning strategy is applied in the local search, which decides the different neighboring structure in the evolution. Experimental results show that the proposed algorithm is efficient.