Spacecraft inertial parameters estimation using time series clustering and reinforcement learning
This addresses a domain-specific challenge for spacecraft operations, offering an incremental improvement in parameter estimation under dynamic conditions.
The paper tackles the problem of estimating changing inertial parameters of a spacecraft during operations like payload deployments or propellant consumption, using a machine learning approach that combines time series clustering and reinforcement learning to generate optimized actuation sequences, with results showing the algorithm is resilient to common disturbances in such operations.
This paper presents a machine learning approach to estimate the inertial parameters of a spacecraft in cases when those change during operations, e.g. multiple deployments of payloads, unfolding of appendages and booms, propellant consumption as well as during in-orbit servicing and active debris removal operations. The machine learning approach uses time series clustering together with an optimised actuation sequence generated by reinforcement learning to facilitate distinguishing among different inertial parameter sets. The performance of the proposed strategy is assessed against the case of a multi-satellite deployment system showing that the algorithm is resilient towards common disturbances in such kinds of operations.