A new differential evolution using a bilevel optimization model for solving generalized multi-point dynamic aggregation problems.

Abstract

The multi-point dynamic aggregation problem (MPDAP) comes mainly from real-world applications, which is characterized by dynamic task assignation and routing optimization with limited resources. Due to the dynamic allocation of tasks, more than one optimization objective, limited resources, and other factors involved, the computational complexity of both route programming and resource allocation optimization is a growing problem. In this manuscript, a task scheduling problem of fire-fighting robots is investigated and solved, and serves as a representative multi-point dynamic aggregation problem. First, in terms of two optimized objectives, the cost and completion time, a new bilevel programming model is presented, in which the task cost is taken as the leader's objective. In addition, in order to effectively solve the bilevel model, a differential evolution is developed based on a new matrix coding scheme. Moreover, some percentage of high-quality solutions are applied in mutation and selection operations, which helps to generate potentially better solutions and keep them into the next generation of population. Finally, the experimental results show that the proposed algorithm is feasible and effective in dealing with the multi-point dynamic aggregation problem.