Computational Intelligence Methods for Optimising Airport Security Process
- Publication Type:
- Thesis
- Issue Date:
- 2022
Open Access
Copyright Clearance Process
- Recently Added
- In Progress
- Open Access
This item is open access.
Airport security screening processes are essential to ensure the safety of both passengers and the aviation industry. Security at airports has improved noticeably in recent years through the utilisation of state-of-the-art technologies and highly trained security officers. However, maintaining a high level of security can be costly to operate and implement. It also lead to delays for passengers and airlines. Nowadays, research is focused to build efficient and effective systems to reduce the congestion caused by the security screening process while maintaining a high level of safety for passengers and the aviation industry. Two open security challenges motivates this thesis: optimize and design the security process at airport, and build an effective intelligent system to detect anomalies in X-ray images.
This thesis proposes a series of novel using queuing theory and machine learning models to handle the aforementioned challenges. Particularly, this thesis addresses the issues related to passengers’ congestion at the waiting area and improve the performance of the security detection system to ensure the safety of both passengers and the aviation industry.
There are four contributions in this thesis. Contribution 1 proposes queueing theory method to optimise the security screening process with multi-servers operating in parallel to serve a different number of passengers during different seasons, such as Christmas, Easter and school holidays, and time of the day, as this strongly influences the number of passengers.
Contribution 2 proposes a novel method based on queueing theory augmented with particle swarm optimisation (QT-PSO) to predict passenger waiting time in a security screening context and to determine the required number of servers and security officers.
Contribution 3 propose a tensor-based learning approach to extract the informative latent features that will be used as an input to build a one-class model for anomaly detection.
Contribution 4 proposes a federated learning (FL) approach for anomaly detection in X-ray security imaging using OCSVM. The innovative machine learning approach can train a centralized model on data generated and located on multiple airports without compromising the privacy and security of the collected data.
The performance of all novel methods in this these is evaluated in the context of Sydney airport dataset, synthetic data, and public datasets for X-ray images. Further, all the results of the novel methods are compared to the state-of-the-art methods. The experimental results shows that our proposed methods in the contributions outperform the state-of-the-art and produce promising results.
Please use this identifier to cite or link to this item: