The mechanical property changes of bridge structures using a Viscoelastic Model and Machine Learning Techniques

Abstract The article evaluates changes in the mechanical structure of a bridge using a model of a viscoelastic oscillator system. The study uses the vibration spectrum of the bridge structure through vibration signals as a research model. The vibration response spectrum of the bridge is considered in two independent states in practice, including the bending and torsional states to investigate the change in material mechanical properties. The first proposal of this study is to use the actual vibration model of the bridge to build a response spectrum that is as close to reality as possible with the support of machine learning. Using a CNN network model, the draught was optimised and enriched with information extracted from the vibration spectrum. The features of the spectrum serve as a basis for determining the existence of defects in the structure. From there, this research model allows the author to explore the influence of mechanical components on the structure through the proposed dissipation coefficient. The second proposal is to build a model of the value of the regression surface of the dissipation coefficient in the structural material. This quantity helps to evaluate the degree of data dispersion of the dissipation coefficient obtained from the response spectrum data. The study shows that the proposed quantity can evaluate the model for good practical results. In the future, this research model can be applied to many different types of structures with complex load states.