Università di Pisa - Italy
Hall effect thrusters are a key technology for space propulsion due to their high efficiency, yet their operation may be affected by low-frequency oscillations of the discharge current, known as “breathing mode”. Understanding and controlling this instability remains a major challenge, due to the complex coupling between ionization, transport, and electric field dynamics.
Reduced one-dimensional fluid models offer a computationally efficient framework to reproduce these phenomena and to perform advanced analyses, such as stability and sensitivity studies. However, their reliability critically depends on the calibration of uncertain parameters, which is traditionally carried out in a heuristic and non-systematic way.
In this work, we present a Bayesian data assimilation approach for the calibration of reduced plasma models, enabling the consistent integration of heterogeneous data sources within a probabilistic framework. The method combines physical modelling with both time-resolved and steady measurements, allowing for uncertainty quantification and improved parameter identifiability. In particular, we introduce tailored likelihood formulations to incorporate information from oscillatory data sources.
The proposed strategy highlights the limitations of calibrating models using single diagnostics and demonstrates the benefits of data fusion in constraining model parameters and enhancing predictive capability.