A multi-layer neural network (MNN) based optimal system has been developed to control the vibrations in flexible structures subjected to deterministic and periodic excitation. Three cases of optimal vibration control are discussed: no external excitation, random excitation, and deterministic and periodic excitation. Numerical simulations indicate that steady state vibrations are eliminated almost completely by the neural network based control system for any excitation frequency contained in the frequency range used for training the MNN. In this sense, the neural network based controller can be described to be highly robust against variations in excitation frequencies.
All Science Journal Classification (ASJC) codes
- Condensed Matter Physics
- Mechanics of Materials
- Acoustics and Ultrasonics
- Mechanical Engineering