A mutation operator self-adaptive differential evolution particle swarm optimization algorithm for USV navigation

Abstract

Read online

To keep the global search capability and robustness for unmanned surface vessel (USV) path planning, an improved differential evolution particle swarm optimization algorithm (DePSO) is proposed in this paper. In the optimization process, approach to optimal value in particle swarm optimization algorithm (PSO) and mutation, hybridization, selection operation in differential evolution algorithm (DE) are combined, and the mutation factor is self-adjusted. First, the particle population is initialized and the optimization objective is determined, the individual and global optimal values are updated. Then differential variation is conducted to produces new variables and cross over with the current individual, the scaling factor is adjusted adaptively with the number of iterations in the mutation process, particle population is updated according to the hybridization results. Finally, the convergence of the algorithm is determined according to the decision standard. Numerical simulation results show that, compared with conventional PSO and DE, the proposed algorithm can effectively reduce the path intersection points, and thus greatly shorten the overall path length.

Keywords

• differential evolution algorithm
• hybridization
• mutation
• particle swarm optimization
• unmanned surface vessel path planning
• scaling factor