Double-assistant evolutionary multitasking algorithm for enhanced electric vehicle routing with backup batteries and battery swapping stations

This study addresses the challenges of the electric vehicle routing problem with time windows, backup batteries, and battery swapping stations (EVRPTW-BB-BSS), reflecting complex issues in modern logistics. We treat each problem as an optimization task. Our main goal is to optimize various types of tasks at the same time. To achieve this, we present the double-assistant evolutionary multitasking algorithm (DAEMTA). This algorithm skillfully combines internal and external knowledge exchange. It helps in the transfer between different and similar types of tasks. DAEMTA includes a two-stage system for collaboration and adaptive methods for managing knowledge transfer. A population enhancement strategy deepens the search for valuable solutions. It also maintains diversity to avoid early convergence. Our experiments show that DAEMTA performs better than advanced EMTAs and traditional algorithms. We test it on 30 multitasking EVRPTW-BB-BSSs and four real-world problems. These findings highlight the practicality and effectiveness of our approach.