An artificial intelligence technique: experimental analysis of population-based physarum-energy optimization algorithm

The traveling salesman problem (TSP) is an NP-hard problem being studied by many researchers. Metaheuristic algorithms generally depend on nature-inspired phenomena successfully applied to combinatorial optimization, such as routing, scheduling, assignment problems, engineering, optimization, genetics, robotics, nanotechnology, and various fields. In this paper, new versions of physarum-energy optimization algorithms (PEOs) which are population-based optimization algorithms were applied to the symmetric traveling salesman problems. PEOs use multiple solutions (population), multiple conductivities (dij) for the stochastic disturbance model and appropriate parameter strategies to update the solutions in each generation. To measure the effectiveness of the PEOs and the hybrids, the algorithms have been evaluated on several benchmark problems compared to the recent metaheuristics. The computational results show that the PEOs and their hybrids (k-NN, 2-opt, 3-opt, k-opt) can find high-quality results compared to the original physarum-energy optimization algorithm (PEO), ant colony optimization (ACO), black hole algorithm (BH), tabu search (TS), and other hybrid algorithms such as whale optimization algorithm + 4-opt heuristic (WOA + 4-opt), camel algorithm + 4-opt heuristic (CA + 4-opt), genetic algorithm + 2-opt heuristic (GA + 2-opt), and the k-NN, 2-opt, 3-opt, k-opt algorithms. Moreover, the population-based PEOs and their derived forms solve the optimization problem quite competitively in CPU time compared to other test algorithms.