Overview
The development of performance modelling methodologies able to translate new ATM concepts and technologies into their impact on high-level, system wide KPIs has been a long-time objective of the ATM research community. Bottom-up, microsimulation models are often the only feasible approach to address this problem in a reliable manner. However, the practical application of large-scale simulation models to strategic ATM performance assessment is often hindered by their computational complexity. The goal of SIMBAD is to develop and evaluate a set of machine learning approaches aimed at providing state of-the-art ATM microsimulation models with the level of reliability, tractability and interpretability required to effectively support performance evaluation at ECAC level. The specific objectives of the project are the following:
- Explore the use of machine learning techniques for the estimation of hidden variables from historical air traffic data, with particular focus on airspace users’ preferences and behaviour, in order to enable a more robust calibration of air traffic microsimulation models.
- Develop new machine learning algorithms for the classification of traffic patterns that enable the selection of a sufficiently representative set of simulation scenarios allowing a comprehensive assessment of new ATM concepts and solutions.
- Investigate the use of active learning metamodelling to facilitate a more efficient exploration of the input output space of complex simulation models through the development of more parsimonious performance metamodels, i.e., analytical input/output functions that approximate the results of a more complex function defined by the microsimulation models.
- Demonstrate and evaluate the newly developed methods and tools through a set of case studies in which the proposed techniques will be integrated with existing, state-of-the-art ATM simulation tools and used to analyse a variety of ATM performance problems.