Wybrane zastosowania sztucznych sieci neuronowych w dynamice konstrukcji

Recently Artificial Neural Networks (ANN) have been applied in numerous fields of civil engineering, e.g. damage identification, structure parameters identification, indestructive testing, structure state estimation, and others. Some results of ANN application are discussed in short, related to the research done at Faculty of Civil and Environmental Engineering of Rzeszow University of Technology. Four examples of application of ANN in static and dynamic of structures are presented: 1) identification of load causing partial yielding in the cross-section of a simple supported beam, 2) identification of parameters of the model of semi-rigid beam-to-column connection, 3) detection of a damage in structural elements using wave propagation, 4) updating of a numerical model of a multi-storey frame. The non-destructive methods of detection of damage and the change in structural elements have been analysed. This methods allow to make state estimation of a structure as well as to predict period of safety usage. Neural networks are applied in both detection of the damage and estimation of location and scale of the damage. To this end finite element models are used. This approach allows analysis of the way of solution and plan the experiment. The method of updating of a discrete model of a multi-storey frame is presented. Two methods of compression of the input vector are presented: compression by a neural network and calculating some geometrical characteristics of the input signal. The method of compression by neural network is used in compression of the time signal and frequency response function. Neural networks can be efficiently applied in the field of dynamic of structures and they can deal with the numerical and experimental data. ANNs can be valuable tool for the analysis of structural damage problems.