State Uncertainty Normality Detection

• DOI: 10.1007/s40295-019-00201-3
• Publication Date: 2019-12-06T19:02:15Z
• Authors: Sven K. Flegel; James C. Bennett
• Keywords: 0101 mathematics; 01 natural sciences

Abstract Two fundamentally different approaches of determining normality the probability density function state estimation error are compared by application to a range test cases. The first method is Henze-Zirkler test, which operates on random particle sample. variability its result quantified. Using this method, departure from has been found occur in three stages detailed. second compares offset whitened space predicted covariance mean obtained unscented transform. This much more efficient than based approach and can be applied using any perturbations model. comparison performed Cartesian two-body motion without process noise. efficient, transform shows excellent agreement with for constructed Application orbit determination results passive optical observations assessed Batch-Least-Squares however reveals some discrepancies have yet understood underline importance rigorous testing.