2019, Vol. 6, No. 2. - go to content...

Permanent address of this page - https://t-s.today/en/10sats219.html

Метаданные этой статьи так же доступны на русском языке

DOI: 10.15862/10SATS219 (https://doi.org/10.15862/10SATS219)

Full article in PDF format (file size: 875.7 KB)

For citation:

Afanasev V.S. Elimination of the effect of temperature on the dynamic parameters of bridge beams. Russian Journal of Transport Engineering. 2019; 6(2). Available at: https://t-s.today/PDF/10SATS219.pdf (in Russian). DOI: 10.15862/10SATS219

Elimination of the effect of temperature on the dynamic parameters of bridge beams

Afanasev Vladimir Sergeevich
Russian university of transport, Moscow, Russia
E-mail: a89162825980@ya.ru

Abstract. An article presents an overview on influence of environmental factors on the dynamic response of bridge beams. Examples of the determination of the technical condition by taking into account the influence of the environment on real bridges are given. The process of damage identification based on operational modal analysis with filtering effects is presented.

Author considered an influence of temperature effect of the environment on eigenfrequencies of beam bridges. A numerical experiment was performed using a large road collapsible bridge as an example: bridge temperature was determined with additional heating of the girder from solar radiation and without it; the values of dynamic parameters of the girder from the change in temperature during the calendar year were calculated; the dependence of eigenfrequency of the girder on changes in temperature of the bridge is constructed. Further, an analysis of a time series formed from the eigenfrequency changes of the girder was made. Two different methods were used to describe the time series: a model based on a statistical pattern in the given data and a model of a recurrent neural network. Before applying the statistical autoregressive integrated moving average model, the time series was decomposed into additive components: trend, seasonal component, residuals and noise. After that, construction of a mathematical model describing the stationary residuals of the time series was performed. Second method – based on recurrent neural networks, can be rephrased as a question of constructing a regression model. on the basis of the training and test data sets, a network was formed containing a visible layer with 1 input signal, a hidden layer with 4 blocks and an output level that makes one prediction value.

Keywords: seasonality elimination; structural health monitoring; bridges; time series; neural networks; forecasting; dynamic parameters

Download article in PDF format