Improvement and Application of Discrete State Transition Algorithm

Abstract

Discrete state transition algorithm relies on the initial solution and can easily fall into the local optimum. This paper proposes an improved discrete state transition algorithm (CDSTA) for the above problem. Firstly, the genetic algorithm is used to initialize to obtain the initial solution with high quality and quickly approximate the optimal value. Secondly, the optimal recovery strategy reduces the number of iterations to accelerate the algorithm's convergence rate. Finally, the chaotic perturbation strategy is introduced. When the algorithm falls into the stagnation state, a chaotic sequence is generated by Tent mapping to get rid of the local extremum. Two single-peaked and three multi-peaked functions are used to experiment with the improved algorithm, and the performance is compared and analyzed with other algorithms. The results show that the improved algorithm's solution accuracy and convergence rate are better than other comparative algorithms. The application of the improved algorithm to the traveling salesman problem demon-strates that CDSTA has good practical engineering application potential.