Abstract: The non-stationarity and randomness of airport passenger arrivals result in uncertain and highly time-varying requirements for security checkpoints, increasing the difficulty of resource allocation. To address this issue, this paper proposes a method embedding a non-stationary queuing model into a genetic algorithm. Initially, a non-stationary queuing model for a single-stage service system with multiple parallel service desks is established, where the input and service processes are fitted to general distributions. Furthermore, an extended stationary backlog-carryover (SBC) approximation is proposed to quickly solve system performance metrics. Then, the extended SBC approximation is embedded into the genetic algorithm for optimization. Finally, experimental and optimization examples are designed. The accuracy of the extended SBC approximation and the effectiveness of the optimization method are verified by comparing the analytical results with the simulation results. The impact of non-stationarity on system performance and optimization results is also analyzed. Results of applying this method to a practical case show that the optimized allocation scheme reduces the number of security checkpoints by 14.80% compared to the actual allocation scheme, proving its effectiveness.

Key words: optimization of airport security checkpoint allocation, non-stationary queuing modeling, service desk allocation, stationary backlog-carryover (SBC) approximation, genetic algorithm