A Grouping Genetic Algorithm Based on the GES Local Search for Pickup and Delivery Problem with Time Windows and LIFO Loading

This paper investigates the pickup and delivery problem with time windows and last-in-first-out (LIFO) loading (PDPTWL). In this problem, the compartment of a vehicle is modeled as a linear LIFO stack. The last picked up goods are placed on the top of the stack, and the goods can be delivered only when they are on the top of the stack. The LIFO constraint makes the feasible solution space more tightly constrained and the design of an effective algorithm more difficult. A grouping genetic algorithm combined with the guided ejection search is proposed to solve the PDPTWL problem of large-size, in which, an evaluation function is defined to guide the selection of genes for crossover and mutation, and a local search based on the guided ejection search is embedded into the genetic algorithm to improve the quality of the solutions. Then, a population-based metaheuristic is ready for the PDPTWL problem. It can solve instances with 50–300 requests in the Li and Lim’s benchmarks. Compared with the existing state-of-the-art algorithms, the experimental results confirm that the proposed algorithm works more efficiently. It improves 164 best-known solutions out of 236 instances and reduces 424 vehicles.