**6. Model Results**

This section presents the different model behaviors for the validation data. Figure 8 represents 30% of the data set described in Equation (22). All the results will be compared to those obtained by a NLARX NN with two hidden layers and 25 neurons each. To train the neural network, the 70% of the FSS data set described in Equation (22) was used.

As a form of comparison, the fitting value for the Normalized Root Mean Squared Error (NRMSE) will be taken into consideration. This tool finds the difference between the measured data and the model response as the sum of the squared individual errors throughout the entire signal. Using this method, the large errors will have a bigger quantitative penalization than small errors.

Finally, to validate the results, three independent different tests will be conducted that compare the methodologies used for static movement, dynamic movement, and normal operation mode.
