UTFacultiesETDepartmentsMS3Research ChairsApplied Mechanics and Data AnalysisEducationMaster AssignmentsCan machine learning approaches for assessing damage of oscillating beams be explainable?

Can machine learning approaches for assessing damage of oscillating beams be explainable?

Master Assignment

In this assignment we will look at a cantilever beam which is oscillated for many hundreds of cycles at always a constant displacement amplitude and excitation frequency. Hence, the dynamic system is excited with a force which makes the tip of the beam oscillating always at the same frequency and amplitude. The relative phase between the input force and the output response shows a linear decrement which suddenly changes slope. That sudden change in slope is caused by a damage appearing inside our oscillating beam. This damage becomes bigger over the number of cycles. In this assignment the data driven techniques will be investigated in order to analyze:

1.      Why the energy of vibrations breaks through high order harmonics when a damage occurs?

2.      How is the damage initiation and early growth affecting the super harmonic amplitudes over number of cycles?

3.      Can data mining find a physical meaning?

                        

Contact Person(s)