Elastic net with stochastic noise strategy and time-dependent parameters for TSP

In this paper, an improved elastic net algorithm is proposed to solve the traveling salesman problem by assembling a stochastic noise strategy and some time-dependent parameters. Based on the observation and analysis of solution status of TSP solving by elastic net, the stochastic noise strategy is introduced into elastic net to overcome the shortcoming of easily trapping in local minima. Being different with other stochastic algorithms, the stochastic noise strategy mainly modifies problem states by using city oscillation randomly and further affects elastic net performance. The time-dependent parameters controlling convergent process increase the ability of matching cities precisely and getting convergence quickly. It is verified by large numbers of simulations that the stochastic noise strategy and time-dependent parameters can enhance the performance of elastic net greatly. And especially the stochastic noise strategy to problem states probably reveals a novel way to optimize some deterministic algorithms.