An Improved Evolutionary Multitasking Algorithm for Multitasking Berth Allocation Problem

The terminal’s resource management is a significant topic in maritime technology that directly affects the operational efficiency of ports. As an important resource in the terminal, the berth allocation problem (BAP) has been widely studied. In this article, to meet the requirement for unified management of multiple independent BAPs by terminal managers, we model such a complex BAP variant as a multitasking BAP (MBAP) that dealing with multiple berths simultaneously. To efficiently solve the MBAP, we propose an improved evolutionary multitask algorithm (IEMA) to simultaneously minimize the total waiting time with three novel designs. Firstly, the solution space is represented by a single population of multiple variables, which is suitable for implicit knowledge transfer and can avoid infeasible solutions. Secondly, a first come first served strategy is adopted to reinitialize a part of individuals to accelerate convergence. Finally, we develop a hybrid genetic mechanism based on the two-dimensional representation characteristics of the MBAP. Experimental results on a series of generated cases show the effectiveness and superiority of the proposed approach.