Using an Artificial Neural Network for Vibration Analysis of Multi-Layered Composite Beams Located on the Elastic Foundation

Different engineering systems are governed by differential equations that can be solved using classical numerical approximation methods. With these methods, relatively accurate results can be achieved by extending the number of meshing points and utilizing high-capacity computers. Moreover, an approach development that is accurate and easy to utilize for solving differential equations has always been of interest. In the present study, we employ artificial neural networks to analyze the vibrations of microbeams with a functional structure resting on an elastic bed. Using the obtained analytical solution results as inputs, a neural network was trained to predict the vibration characteristics of these beams. According to the results, the beam frequency can be approximated accurately using an artificial neural network without solving the governing equations. As can be seen, the multi-layer perceptron artificial neural network of backpropagation type with a structure of 4-12-15-1 and training based on the Lm algorithm provides a good result.