The simulated annealing (SA) algorithm can be employed to solve the traveling salesman problem (TSP). When the scale of the problem increases, the convergence of SA will be extremely slow. To address this issue, we propose an improved SA (ISA) algorithm, which can adaptively search an optimal solution with multi-modal pathway variants, and introduces jump-cooling and quadratic variation mechanism during the search process to improve the convergence speed. A quadratic annealing is also introduced at the end of the algorithm to prevent from falling into a local optimum. A historical optimal solution memory is proposed to ensure that the optimal solution will not be lost due to probabilistic acceptance of poorer solutions. The results of comparative experiments on different data sets of TSP benchmarks, which are obtained from TSPLIB, justify that ISA outperforms SA as well as other heuristics in terms of faster convergence speed and higher search accuracy. We also evaluate and show the efficiency and optimization of ISA by real data, e.g., Hebei province tour data.