Spatiotemporal analysis of airport security lane throughput for lane-opening decision support via nonnegative matrix factorization

Passenger demand in airport security screening systems is characterized by pronounced temporal variability and uncertainty. An inappropriate configuration of open security lanes may lead to queue congestion or resource underutilization. To identify the spatiotemporal patterns of security screening operations and support the development of lane-opening plans, this study proposes a security flow pattern analysis method based on Nonnegative Matrix Factorization (NMF). Using 10-min aggregated throughput data collected over 92 days from 37 domestic security lanes at an airport, this study constructs a flow matrix in which each lane-day is treated as a sample and intra-day time intervals are treated as variables. NMF is then applied to extract typical intra-day operational patterns. The results show that security lane throughput can be decomposed into several operational patterns with clear temporal characteristics. Moreover, the activation levels of these patterns vary significantly across different lanes, reflecting spatial heterogeneity within the domestic security screening area. Furthermore, the extracted temporal patterns and lane activation features are used to estimate lane-opening requirements for different time periods, providing a basis for lane-opening planning. The findings demonstrate that the proposed method can extract operationally interpretable spatiotemporal patterns from historical operational data and provide data-driven support for operational state identification and resource allocation decisions in airport security screening systems.