Cost function for 1st approach is complete. Calibration results converge while using the solver lsqnonlin, algorithm "trust-region-reflective", Tolx = 0.6deg, boundaries = +/-20deg. Converges in 7 iterations.
Initial bucket was filtered and down sampled from 3000 to 1000 samples. Optimisation was run on data subsets of 100 samples (randomly selected over the total of 1000 samples).
Cost function for 1st approach is complete. Calibration results converge while using the solver
lsqnonlin, algorithm "trust-region-reflective", Tolx = 0.6deg, boundaries = +/-20deg. Converges in 7 iterations. Initial bucket was filtered and down sampled from 3000 to 1000 samples. Optimisation was run on data subsets of 100 samples (randomly selected over the total of 1000 samples).