Artificial neural networks for optimal control of serial flexible structures

A radial basis function-artificial neural network is proposed for the optimal control of serial flexible structures based on substructure synthesis. The artificial neural network is trained using the global LQR controller assembled from the subcontrollers designed at substructure levels. Furthermore, the neural network training is carried out through only the sensors collocated with the actuators. The resulting artificial neural network controller is compared with the global LQR controller designed using the whole structural model. The numerical results of the two controllers closely match each other.