Current ATM researchers commonly rely on Base of Aircraft Data (BADA) as the aircraft performance model (APM). However, the license and openness of BADA sometimes presents challenges for researchers and other users.
This research focuses on applying different machine learning, data mining, and modeling methods to utilize the big data from ADS-B, together with other open data sources, to build an open aircraft performance model that can be used freely without restriction of license. It also aims to provide the core performance models for our open source ATM simulator BlueSky.
The work can be divided into the following main areas of research:
- Gathering, processing, and storing of large quantity of air traffic data
- Integrating different open data sources with ADS- B data (e.g. weather data, Mode-S EHS data)
- Deploying machine learning methods to explore, extract, and analyze ATM big data
- Developing advanced modeling methods on aggregated data
- Studying and evaluating our open source APM, versus the performance of current APM
- Developing estimation and inference methods to be able to predict certain key parameters such as mass and fuel flow of aircraft
The challenges of the research are:
- Limited knowledge on individual flights in our data
- Atmospheric conditions also affect performance, and these are difficult to be modeled for each trajectory
- Assumptions sometime need to be made, which can introduce more uncertainty in estimations
- The level of fidelity of the performance models
