AIrcraft - Increasing the efficiency of the Environmental Control System (ECS) through AI-assisted regulation

Project: Research project

Project Details


Commercial aviation causes around 2-3% of global CO2 emissions (2.4% in 2018 - that is 873 million tons of CO2 per year) and is expected to grow by around 50% until 2036 (based on data before the corona crisis). The environmental control system (ECS) - responsible for the regulation of hot bleed air for deicing systems and the desired pressure, temperature and humidity conditions in the cabin and the cargo compartment - is the most energy demanding subsystem of an aircraft (approx. 5% of total energy consumption). Currently used standard controllers (e.g. PI or PID controllers) can cause unstable states in transient phases which lead to overshooting or undershooting behavior, which have a negative impact on the energy efficiency. The use of artificial intelligence (AI) methods offers a wide range of opportunities for improving the control quality. The aim is to apply AI-based controllers in aviation industry to take advantages from existing knowledge in other branches. For this purpose, knowledge acquired about AI technologies (use of machine learning for predictive maintenance applications in rail vehicles and machine learning data-based prediction techniques for battery systems) and AI-based controls (of heating systems in buildings and ovens) will be transferred to the field of aviation. Due to the improved control ability by means of AI, energy savings in the range of 10-15% are expected for the ECS while maintaining or increasing cabin comfort. The high requirements with regard to safety issues will also be taken into account. If the AI-based control leaves defined ranges, the standard control will take over the control again. In addition, the AI-models should be able to detect process changes in order to be able to warn of degradation effects of components or possible errors in the system.
Effective start/end date1/08/2131/07/22


Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.