Possibilistic space object tracking under epistemic uncertainty

•The epistemic uncertainty in the process of space object tracking is properly modeled by OPMs.•A refined PAR+ algorithm is developed to handle different levels of prior knowledge in IOD.•A rigorous parameter estimation method of Gaussian possibility functions is proposed to initiate the proposed possibilistic filter.•Improved accuracy of orbital state estimation can be achieved by the fusion of TLEs with real radar measurements.