Similarity Estimation for Assessing the Accuracy of a Non-Invasive Identification Method

In the paper, an approach for parameter identification during regular movements of servo axes is discussed. In contrast to the usual case, where test signals are applied to the process, there is no deterministic excitation present in this manner. Consequently, the accuracy of the identified parameter depends on the available excitation and needs to be assessed after the identification is completed. Considering that, several methods of error calculation and similarity measurement are presented in the paper and discussed for the specific use case of non-invasive parameter identification.