Sensitivity study of the open loop identification methods based on the whitening of the prediction error

The objective of this paper is to choose an optimal solution for open loop identification with the following methods based on the whitening of the prediction error: Recursive Least Square, Extended Least Square, Generalized Least Square, Recursive Maximum Likelihood and Output Error with Extended Prediction Model. In order to test and find an optimal solution for open loop identification three situations where considered: variable perturbation amplitude, variable parameter initialization, variable SPAB amplitude for a decreasing adaptation gain. For the simulations with the open loop identification methods, the real process was considered to be a flexible robot arm with a model complexity of 6 as described in section 2 of the paper.