A multi-objective invasive weed optimization for task assignment in prefabricated systems

A novel discrete invasive weed optimization (DIWO) algorithms is proposed to solve the task assignment problems in the prefabricated systems. Three objectives are considered simultaneously, i.e., minimization of the total execution time, the total transportation cost and the total balance of execution time. In the proposed algorithm, a novel reproduction heuristic based on the p-optimality criteria is embedded to decide the number of offspring for each weed during the generation, which can obviously increase the searching quality and diversity of the algorithm. Moreover, five types of neighborhood structures are proposed to realize the dispatching mechanism in the canonical IWO algorithm, which can only applicable for the continuous optimization problems. Furthermore, to decrease the computational complexity, during each generation, the newly-generated weeds are stored and applied the non-dominated sorting algorithm to update the Pareto archive set. Detailed comparisons on the randomly-generated instances show the efficiency and diversity of the proposed DIWO algorithm.