Improved slime mould algorithm based on hybrid strategy optimization of Cauchy mutation and simulated annealing

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Operations Research

Background:

• The standard Slime Mould Algorithm (SMA) faces limitations in global optimization capabilities.
• Existing metaheuristics may struggle with complex optimization tasks like the Capacitated Vehicle Routing Problem (CVRP).

Purpose of the Study:

• To propose an enhanced Slime Mould Algorithm with Cauchy mutation and simulated annealing (SMA-CSA).
• To improve the global optimization performance and efficiency for solving the CVRP.

Main Methods:

• Integration of Cauchy mutation strategy to escape local optima.
• Incorporation of simulated annealing with the Metropolis sampling criterion for expanded global search.
• Evaluation using CEC 2013 benchmark functions and the CVRP.

Main Results:

• SMA-CSA demonstrated superior performance compared to ten state-of-the-art metaheuristics.
• Statistical analysis using Friedman and Wilcoxon rank-sum tests confirmed SMA-CSA's competitiveness.
• The algorithm showed significant efficiency and discrete solving ability on CVRP instances.

Conclusions:

• The proposed SMA-CSA algorithm effectively addresses the limitations of the standard SMA.
• SMA-CSA offers a competitive and often superior solution for global optimization and CVRP.
• The enhanced exploration and exploitation phases contribute to achieving optimal solutions.