Figure 4.

Process carried out to obtain the dataset with which the model has been trained. Firstly, the acquired readouts are labeled according to their conditions of temperature and deflection at the free end of the beam. Then, each readout is sliced into several samples, where it is assumed that the strain and temperature are constant. After, for each pair of slices with the same spatial position, the temperature and strain increment are computed. Finally, this set of data is composed of two slices of readouts with two polarization states (four signals), and their corresponding temperature and strain increments are appended to the dataset.