Minimum energy driving a flexible link hammer using neural networks

This paper proposes a continuous hitting operation on objects with unknown coefficients of rebound by a flexible link hammer. This hammer system uses only the first mode of vibration for a desired hitting. The angular acceleration for a continuous hitting can be obtained from numerical calculations. However, this method is not useful for on-line processing Therefore, the multi-layered neural networks are applied to the calculation of the parameters of the angular acceleration. The outputs of the trained network agree well with the numerical solutions