I was initially generating some test measurements using SPICE and trying to fit them using Orekit. I found that it wouldn’t converge, even when given the correct answer as a guess. As such, I went back and rewrote the propagation and measurement generation code to use Orekit also. Unfortunately, Orekit still fails to converge, even with the parameters given in most examples — 20 m and 0.001 m/s sigma on two-way range and range-rate, respectively.
My measurements are here: https://gist.github.com/autumnsault/96968fdd690194a192a56f0845d797a2
The initial state is:
t0 = 2022-06-16T21:44:43.373
x0 = [-6.45306258e+06, -1.19390257e+06, -8.56858164e+04,
1.83609046e+03, -9.56878337e+03 -4.95077925e+03] in J2000
If I use that initial state, I get an exception: “org.orekit.errors.OrekitException: minimal step size (1.00E-03) reached, integration needs 7.43E-04”. Even lowering the min step, I get that exception. If I use the initial state for the W3B.aer example, it iterates forever without converging.
I’m only using a single newtonian attractor, with mu = 398600435436095.9.
dP = 10.0, and convergence threshold is 0.1.