An AI approach for optimizing multi-pallet loading operations

For pallet loading operations, it is found that space optimization does not necessarily lead to profit optimization, which is the ultimate goal of forwarders after numerous site evaluations and end-user feedbacks. To the best of the authors’ knowledge, there are only a few research studies related to profit optimization in this area. This paper presents a hybrid approach, using heuristic and genetic algorithms (GA), for solving the profit-based multi-pallet loading problem which was mathematically formulated as a nonlinear integer programming problem. The major novelties in this paper are the simultaneous consideration of priority for loading more profitable cargoes and cargo stability in heuristic and innovatively designed crossover and mutation operations in GA to suit the profit optimization. To validate the approach, simulations were carried out with 10 weakly and 10 strongly heterogeneous sets of cargoes. The simulation results obtained by our proposed GA were compared with those obtained by two other stochastic search methods, namely simulated annealing (SA) and Tabu search (TS), as well as a nonlinear integer programming-based method, branch-and-bound (BB). The results showed that GA can search more profitable solutions than SA, TS and BB in this multi-pallet loading problem.