In this study, a method based on a convolutional neural network was proposed to estimate the participation factor of each vibration mode in forced vibration response, expressed by the mode superposition method, and the effectiveness of the method was validated by applying it to a cantilever beam. After the transient response decays out, the forced vibration behaviour of a structure due to a harmonic excitation force can be expressed as a linear combination of its vibration modes. Since the participation factor of each vibration mode varies depending on the frequency of the excitation force, its location, and distribution, only a few vibration modes with large participation factors can be used to approximate the forced vibration behaviour. To estimate this participation factor, a convolutional neural network model was trained using appropriate input and output data. The proposed and trained deep learning model showed high accuracy against test data. Furthermore, its areas of application were discussed.
