Application of hybrid swarming algorithm on flexible job shop scheduling problems

Abstract

In this article, we present an improved hybrid algorithm based on ant colony optimization and the polycephalum algorithm. First, we use improved the probability selection mechanism in the ant colony algorithm in order to improve the efficiency of next point searching. Second, in each iteration we update the pheromone concentration of the optimal route by using the polycephalum algorithm. We regard the starting point of the optimal route as the water injection point and the end point as the water outlet point.The hybrid algorithm is compared on multiple TSPLIB problems and flexible job shop scheduling problems. And experiments show that the improved algorithm has good application results and resultful in accuracy and optimal solutions. K E Y W O R D Sant colony algorithm, FJSP, Physarum polycephalum algorithm INTRODUCTIONIn 1990, Bruker and Schlie 1 created a polynomial graph algorithm for solving flexible job shop scheduling problem (FJSP), but this algorithm cannot effectively solve multijob, multimachine, and multiconstraint problems. Later, some researchers used heuristic algorithms to handle this problem.For example, Brandimarte 2 suggested a tabu search algorithm based on a hierarchical strategy, assigned each operation of each job to an equivalent computer, and solved FJSP. Some scholars used the neighborhood structure to enhance the development capabilities of tabu search algorithms, and then perform machine reallocation within a predetermined time to solve FJSP. 3,4 Alzaqebah et al. 5 proposed a BSO algorithm that mimics the brainstorming process of humans, and added clustering ideas to polish up the local intensification of the algorithm, and disposed of FJSP. Ding and Gu 6 proposed an improved particle swarm optimization algorithm to dispose of the FJSP by improving the chain-based coding scheme, information exchange mechanism between particles, and the selection regulations of the next generation. Zhu and Zhou 7 proposed an efficient evolutionary multiobjective gray wolf optimizer. By improving the social hierarchy and leadership strategy, the goal of reducing machine workload in FJSP is minimized. Li et al. 8 created a hybrid artificial bee colony algorithm to solve FJSP, introduced a replanning strategy and clustering grouping roulette method, which improved the efficiency of the solution. Phanden and Ferreira 9 presented proposed a population-based biogeographic algorithm inspired by nature to solve FJSP. By citing neighborhood search methods, local search capabilities are enhanced. Xing et al. 10 initiated an ant colony optimization algorithm for knowledge heuristic search to solve FJSP. It can be seen that heuristic algorithms are very effective and popular in solving FJSP problems, and their performance is far superior to traditional algorithms. Cai et al. 11 proposed an intelligent algorithm of sine function to implement multicloud and scheduling problems in the Internet of Things (IoT). Cui et al. 12 proposed a multiobjective RBM model combined with the NSGA-II algorithm to improve the data cla...